Infantile-onset spinocerebellar ataxia (IOSCA) is a severe neurodegenerative disorder caused by the recessive mutation in PEO1, leading to an Y508C change in the mitochondrial helicase Twinkle, in its helicase domain. However, no mitochondrial dysfunction has been found in this disease. We studied here the consequences of IOSCA for the central nervous system, as well as the in vitro performance of the IOSCA mutant protein. The results of the mtDNA analyses were compared to findings in a similar juvenile or adult-onset ataxia syndrome, mitochondrial recessive ataxia syndrome (MIRAS), caused by the W748S mutation in the mitochondrial DNA polymerase (POLG). We show here that IOSCA brain does not harbor mtDNA deletions or increased amount of mtDNA point mutations, whereas MIRAS brain shows multiple deletions of mtDNA. However, IOSCA, and to a lesser extent also MIRAS, show mtDNA depletion in the brain and the liver. In both diseases, especially large neurons show respiratory chain complex I (CI) deficiency, but also CIV is decreased in IOSCA. Helicase activity, hexamerization and nucleoid structure of the IOSCA mutant were, however, unaffected. The lack of in vitro helicase defect or cell culture phenotype suggest that Twinkle-Y508C dysfunction affects mtDNA maintenance in a highly context and cell-type specific manner. Our results indicate that IOSCA is a new member of the mitochondrial DNA depletion syndromes.
Defects in the mitochondrial DNA (mtDNA) maintenance proteins, such as the Twinkle helicase and the mitochondrial DNA polymerase (POLG), lead to severe neurodegenerative diseases. Infantile onset spinocerebellar ataxia (IOSCA; OMIM no. 271245), a disease resulting from a recessive Twinkle mutation (c. 1523 A>G leading to amino acid change Y508C; from now on IOSCA-Y508C), manifests after 9–18 months of age with muscle hypotonia, athetosis, ataxia, ophthalmoplegia, sensorineural hearing deficit, sensory axonal neuropathy, epileptic encephalopathy and female hypogonadism (1,2). Neuropathologically, the disease is characterized by progressive atrophy of the cerebellum, brain stem and the spinal cord, as well as sensory axonal neuropathy (3). IOSCA is the second most common inherited ataxia in Finland, with 22 patients and a population carrier frequency of more than 1:230 (2).
We have reported the IOSCA-Y508C as compound heterozygous with a mutation leading to an A318T change in patients with severe early onset encephalopathy, signs of liver involvement and mtDNA depletion in the liver (4). A hepatocerebral form of mitochondrial depletion syndrome (MDS) is also caused by a recessive Twinkle mutation leading to a T457I change (5). In contrast, dominant Twinkle mutations manifest with progressive external ophthalmoplegia (PEO), mitochondrial myopathy and sometimes additional features (OMIM no. 609286) (6–8). As a hallmark of PEO, mtDNA deletions accumulate in the muscle, brain and the heart of the patients (6,9,10). IOSCA muscle, however, has not shown any signs of mtDNA defects or respiratory chain deficiency (1,2).
A key functional partner of Twinkle, the replicative mtDNA polymerase (polymerase γ, POLG, POLG1), is associated with a similar clinical spectrum: dominant POLG1 mutations cause PEO, and recessive mutations underlie diverse phenotypes including Alpers-Huttenlocher syndrome with mtDNA depletion (OMIM no. 203700) (11–14), and mitochondrial recessive ataxia syndrome (MIRAS) (15–18). MIRAS closely resembles IOSCA, except for the later onset of 5–41 years in MIRAS compared to 1–2 years in IOSCA.
Twinkle helicase, with strong homology to the bacteriophage T7 primase/helicase, is considered to be the replicative helicase of mtDNA, but is not thought to function as the primase in replication (7,19). Most of the PEO mutations cluster in the linker region that is considered essential for subunit–subunit interactions in the hexameric Twinkle, whereas the Y508C mutation resides in the helicase domain of the protein, just N-terminal to the conserved Walker B motif. The importance of Twinkle for mtDNA replication has been unequivocally shown (20), but the CNS-specific manifestations of its diseases cannot be directly explained by an mtDNA replication defect.
In order to study the pathogenesis of IOSCA and MIRAS, we performed a quantitative and qualitative analyses of mtDNA as well as the respiratory chain proteins from the tissues of the patients. Further, we performed a comprehensive in vitro analysis of the mutant Twinkle protein. We provide here evidence of tissue-specific depletion of mtDNA in IOSCA and neuronal complex I (CI)-defect in both of these mitochondrial ataxias.
IOSCA shows severe mtDNA depletion and MIRAS mtDNA deletions and subtle depletion in the brain.
Both Twinkle and POLG are essential mtDNA maintenance proteins. To study whether mtDNA integrity was compromised in the tissues of IOSCA and MIRAS patients, we studied the mtDNA amount, point mutation load and presence of mtDNA deletions in the brain and other available tissues.
Figure 1 illustrates the results of mtDNA analysis by real-time quantitative PCR (q-PCR) analysis. Also Southern blot analysis was done on most samples, replicating the results of the q-PCR. IOSCA cerebrum and the cerebellum showed mtDNA depletion with residual mtDNA amounts numbering 5–20% (q-PCR and Southern), and also in the IOSCA liver, the mtDNA amounts were decreased to 10–30% as determined by q-PCR or to 10–70% as determined by Southern analysis. In the skeletal muscle, our IOSCA patient showed mtDNA amounts comparable to those seen in the controls. The two MIRAS patients showed somewhat decreased levels of mtDNA in the cerebellum and the cerebrum by q-PCR, but the finding of mtDNA depletion was confirmed by Southern analysis only in the cerebellum of one MIRAS patient (M1), whereas the mtDNA content of this patient was within normal range in the cerebrum. In the Southern analysis, the mtDNA content of the MIRAS liver was barely within low normal range, and the mtDNA amount in the MIRAS muscle within normal range (patient M1). Therefore, Twinkle-IOSCA was associated with clear brain-specific mtDNA depletion, whereas POLG-MIRAS showed reduced levels of mtDNA.
No mtDNA deletions were detected by long-range PCR from the cerebral cortex, the cerebellum, the liver or the muscle of our IOSCA patients, although mtDNA deletions were readily detected in the positive controls (Fig. 2). Our two MIRAS patients showed multiple mtDNA deletions in the frontal cortex, and the cerebellum, as well as a minimal amount of mtDNA deletions in the muscle, but no deletions in the liver.
Figure 3 illustrates the results of mtDNA point mutation analysis; the mutation load was comparable in the cerebellum of the control (aged 50 years), and the patients with IOSCA (P3, 21 years). The 2-fold increase of mtDNA point mutations detected in control region of the MIRAS patient (52 years) sample was not considered significant, as no increase could be seen in the cytochrome b gene. When the same mutation occurred in multiple clones, the mutation was taken into account only once, because this was considered to best reflect de novo mutagenesis, and not reflect clonal expansion of a single mutation in vivo. This approach, however, ignored same-site mutations. Since point mutations are considered to accumulate with age, the somewhat lower amount of point mutations in the IOSCA brain may be explained by the younger age of the IOSCA patient compared to the MIRAS patient and the control. Our results are in agreement with previous studies on mtDNA point mutation accumulation in human tissue (21): healthy control individuals aged 32–72 years accumulated ∼0.2–2 mutations/10 kb in the control region and ∼0.2–1 mutations/10 kb in the cytochrome b gene region in the skeletal muscle. Twinkle-IOSCA and POLG-MIRAS did not increase mtDNA point mutation load in the affected tissues.
IOSCA and MIRAS show reduced CI immunoreactivity in the large neurons of cerebellum and frontal cortex
We stained formalin-fixed paraffin-embedded sections from the frontal cortex and the cerebellum of two IOSCA patients, two controls and one MIRAS patient, with antibodies against the mitochondrial respiratory chain CI, CII and CIV, including three antibodies against different CI subunits (Fig. 4). The CII and CIV stainings appeared similar in the frontal cortex and the cerebellum of the IOSCA patients, the MIRAS patient and the controls. However, CI levels appeared drastically reduced or were below the level of detection both in the frontal cortex and the cerebellum of the two IOSCA patients [antibodies against the CI NDUFB4, ND1 subunits (Fig. 4); antibody against NDUFS3 subunit showed similar results, data not shown], although signals were readily detected in the controls (Fig. 4). The evaluation of Purkinje cells in the MIRAS patient was challenging, as her cerebellum showed almost complete loss of this neuron type. However, all Purkinje cells that were found showed CI deficiency with all the used antibodies, and were positive for CII and CIV (Fig. 4).
Low CI and CIV in IOSCA frontal cortex on western analysis
We analyzed the respiratory chain complex subunits in the IOSCA and MIRAS brain by SDS–polyacrylamide electrophoresis (SDS–PAGE) and western blotting (Fig. 5). In the frontal cortex of the IOSCA patients (n = 4), the levels of CI and CIV were clearly reduced when normalized to the amount of mitochondrial outer membrane protein, porin, levels and compared to the controls (n = 5). The MIRAS frontal cortex did not show changes in CI or CIV levels. In the cerebellum, the levels of CI and CIV were slightly reduced in the IOSCA patients (n = 2), but unchanged in the MIRAS patient, when compared with the controls (n = 7). In the analysis of the cerebral samples by blue native gel electrophoresis (BNGE), the four IOSCA patients showed CI/CII—ratios ranging from 22 to 64% of the average of three controls with an antibody against NDUFA9, or ranging from 60 to 80% with an antibody against NDUFB4 (IOSCA n = 2, controls n = 3). The CIV/CII ratios in the IOSCA frontal cortexes were 30–80% of the average ratios in three controls. We did not observe any changes in the amounts of the complexes in the frontal cortex of the MIRAS patient. These findings supported the SDS–PAGE findings of CI and CIV deficiency in IOSCA brain.
Y508C Twinkle helicase shows normal nucleoid localization
In 293 FlpIn TREx cells overexpressing myc-His-tagged Twinkle wild-type or Y508C, co-localization of both variants with mtDNA was confirmed using anti-myc and anti-DNA antibodies. No obvious difference in nucleoid size or distribution was observed comparing wt and Y508C protein expression (Fig. 6). These results suggest that the overexpressed Y508C Twinkle localizes normally in mtDNA nucleoids.
The Y508C Twinkle helicase has high unwinding activity and unchanged nucleotide affinity, multimerization capacity and ssDNA binding activity in vitro
The double-strand DNA unwinding activity of wild-type Twinkle and the variant bearing the Y508C mutation was compared in an in vitro assay using a commonly used ssM13-oligonucleotide substrate and a nucleotide or deoxynucleotide as energy source. The highest unwinding activity of wild-type protein was observed with UTP, while ATP, GTP and dGTP supported ca. 22, 10 and 8% of maximal enzyme activity, respectively. The Y508C variant showed a significant increase of ca. 45% in unwinding activity with UTP. Also with the other nucleotides tested, the helicase activity of the mutant form was increased (Fig. 7A). We measured the helicase activity over a broad UTP concentration range, but did not observe altered kinetics for the mutant protein (Fig. 7B).
Based on a recent molecular model of the Twinkle protein, it was suggested that the Y508C mutation might affect multimerization of the protein (22) and thus ssDNA binding, as the two processes are intimately linked in hexameric helicases. Thus, we compared the subunit–subunit interaction and formation of oligomers of the Y508C variant to those of the wild-type Twinkle using size exclusion chromatography (Fig. 7C) and mild glutaraldehyde cross-linking (data not shown). Both variants formed stable hexamers under conditions in which Twinkle exhibits helicase activity (low ionic strength), under high ionic strength conditions, and in the absence of magnesium, suggesting that the Y508C does not affect multimerization. The purified Twinkle Y508C did show a tendency to reduced affinity for single-stranded DNA compared to wt Twinkle (Fig. 7D), but this was not statistically significant.
Y508C-Twinkle overexpression in TREx cells did not result in specific changes in mtDNA replication intermediates or in 7S DNA content
To see any direct effect of overexpression on mtDNA replication, we performed native two-dimensional DNA agarose gel electrophoresis (2DNAGE) (Fig. 8). Previously, we showed that overexpression of wild-type Twinkle does not affect mtDNA replication itself, but reduces forms such as termination intermediates (23). A similar effect as in overexpressed control was now observed with the Y508C variant; in fragments containing the D-loop region, the levels of termination intermediates were strongly reduced, but no Y508C-specific changes were observed. Other regions of mtDNA did not give any indication of altered replication speed or mode. On Southern analysis, the 7S DNA content in both wild-type and Y508C overexpressing TREx cells was reduced, when compared to the non-induced cells (Fig. 9), but no Y508C specific changes were seen.
MtDNA levels and mitochondrial transcripts were unchanged upon Y508C Twinkle overexpression in FlpIn TREx cells
When wild-type and Y508C Twinkle were overexpressed in 293 FlpIn TREx cells, only minor changes in mtDNA copy number were observed. The overexpression of wild-type Twinkle increased the content of mtDNA by 20% after 2 or 3 days of induction (P = 0.02 for 3 days induction). Cells expressing the Y508C variant at a similar level did not show any significant changes in mtDNA content. Overexpression of either wild-type or the Y508C protein did not influence steady-state levels of mitochondrial 12S, COXI, ND2 or ND6 transcripts, when studied by northern blot and compared to the nuclear 18S rRNA transcript (data not shown).
IOSCA mutation creates a heme-binding motif in Twinkle, but does not bind heme covalently
The IOSCA mutation (Y508C), changing a tyrosine to a cysteine, creates a conserved heme-binding motif, the CXXCH-motif. This motif is found in c-type cytochromes and binds heme covalently, the two cysteines forming the covalent disulfide bonds and the histidine functioning as a ligand for the heme iron. B-type cytochromes bind heme non-covalently, but can be altered into covalently heme-binding cytochromes by site-directed mutagenesis in the heme-binding pocket (24,25). Also an artificial heme-binding tag, with the heme bound covalently to the CXXCH-motif, has been created (26). Our database search for proteins with the CXXCH-motif resulted in 330 proteins, of which the amount of hemoproteins numbered 266. We therefore investigated the possibility that the Y508C Twinkle protein could bind heme covalently. Based on the protein structure of the helicase domain of the bacteriophage T7 primase/helicase (with a helicase domain homologous to the human Twinkle) and our sequence alignment of the T7 primase/helicase and Twinkle, the Y508C change resided in a domain not involved in subunit interactions of Twinkle (Fig. 10A). We could therefore compare the structure of one Twinkle subunit with the c-type cytochromes. The Y508C change occurred on an alpha helix facing a beta-sheet in a similar conformation as seen in the heme-binding site of cytochrome c.
Based on the similarities between the heme-binding sites of Twinkle-Y508C and cytochrome c (Fig. 10B), it seemed plausible that the Y508C mutant could bind heme covalently. However, direct detection of heme by chemiluminescence in overexpressed wild-type or Y508C Twinkle did not show a heme signal, although the positive control cytochrome c was readily detected (Fig. 11). The presence of the Twinkle proteins on the membrane was verified by immunostaining with an antibody for the c-Myc-tag (Fig. 11). This suggested that the Twinkle variants did not bind a substantial amount of heme in a covalent manner.
Twinkle and POLG defects have recently been shown to be important causes of neurological disease. However, it remains unresolved why some mutations cause a CNS-specific phenotype, whereas other mutations in the same proteins cause muscle diseases. We studied in detail IOSCA, which is a recessive childhood disorder characterized by major cerebellar dysfunction and sensory neuropathy, but lack of muscle symptoms, and caused by the Y508C mutation in the Twinkle helicase. IOSCA was compared to the findings in a similar adult or juvenile-onset ataxia, MIRAS, caused by POLG defects. We show here that IOSCA is associated with brain-specific depletion of mtDNA, and that MIRAS is associated with multiple mtDNA deletions and subtle mtDNA depletion in the brain. However, when the Y508C mutant protein was analyzed in detail in vitro, no defects of the basic replicative helicase functions could be identified. These results suggest that IOSCA is caused by a CNS-specific defect of Twinkle—either a context-specific dysfunction in replication or a defect not directly associated with the catalytic helicase function.
As PEO patients with dominant Twinkle mutations accumulate mtDNA deletions in their muscle and brain, it has remained a possibility that brain-specific mtDNA deletions would account for the neurological phenotype in patients with recessive Twinkle mutations. However, the high amount of mtDNA deletions in the brain of PEO patients—up to 60% of total mtDNA amount—and the absence of cerebellar dysfunction in these patients (10) rendered it unlikely that IOSCA would be caused by mtDNA deletions. We show here that mtDNA deletions are indeed absent in IOSCA brain, and that the cerebellum does not show abnormal mtDNA point mutation accumulation. Instead, we detected depletion of mtDNA in the cerebellum and the cerebrum of our four IOSCA patients. We and collaborators previously found mtDNA depletion in the liver of patients compound heterozygous for the A318T and Y508C Twinkle mutations (4) or in patients homozygous for the T457I Twinkle mutation (5), but no previous data on the brain has been available. In patients with a myopathic form of mtDNA depletion syndrome, mtDNA levels varied between individual muscle fibers and the findings supported a certain threshold level required for normal cytochrome c oxidase activity (27). The level of mtDNA reduction in the brain of our IOSCA patients was considerable, but it remains to be established whether specific cell types are more severely affected than others. The severity of mtDNA depletion does not necessarily directly correlate with the amount of mitochondrial respiratory chain complexes, as compensatory mechanisms may help in maintaining relatively high steady-state levels of mitochondrial transcripts, as shown in the muscle of patients with severe mtDNA depletion (28).
In line with the mtDNA depletion in the brain, we found a reduction of the protein amounts of respiratory chain CI and CIV in IOSCA brain by western analysis. CI includes seven mtDNA-encoded subunits from a total of at least 45, CIV has three mtDNA-encoded out of 13 subunits, whereas CII is completely nuclear encoded. The western findings were relatively modest, but in immunohistochemical analysis we found with different antibodies drastically reduced immunoreactivity for CI in Purkinje cells and in pyramidal neurons of IOSCA and MIRAS patients. The cells showed, however, reactivity to CIV and CII. We suggest that IOSCA and MIRAS show cell-specific manifestations of the complex defects: CI seems to be severely affected at least in the somas of the two large neuron types. A typical phenotype caused by recessive CI defects in children is Leigh syndrome, a subacute necrotic encephalopathy severely affecting the brain stem, basal ganglia and the thalami. Ataxias have also been associated with CI-deficiency before: mice with CI deficiency, due to knockout of the Ndufs4 subunit, manifest with ataxia and encephalomyopathy progressing to death at ∼7 weeks of age (29), and Harlequin mice with reduced expression of the apoptosis inducing factor show CI deficiency (30). As the ATP levels in the resting muscle of the Ndufs4 knockout mice appeared normal, the encephalomyopathy was suggested to result either from increased local demands of energy production in certain tissues, such as the maturing brain, or from indirect effects of CI. As the mtDNA depletion and CI deficiency in IOSCA are likely to be progressive, partial and form gradually, the disease represents a later form of CI deficiency compared to Leigh disease, which may explain the differential brain region manifestations. Our results suggest that the presence of CI in the cerebellum is crucial for its normal function, and indicate that CI deficiency can cause severe ataxia syndromes in humans.
To search for the functional defect in the Y508C Twinkle, leading to brain-specific mtDNA depletion, we produced and purified the mutant protein and analyzed its in vitro helicase functions. We found the helicase activity, nucleotide kinetics hexamerization and affinity to ssDNA to be unaffected, indicating that the enzymatic function of Twinkle as the replicative mtDNA helicase, as can be measured in vitro, was unaffected. Previous nucleotide titration studies on the helicase activity of an other recessive helicase domain Twinkle mutant (recessive T457I mutation) showed a ∼50–60% decreased relative maximal helicase activity (5), which may explain the neonatal onset and early-fatal phenotype in these patients when compared with IOSCA patients with the disease-onset at 1–2 years and survival to adulthood. In spite of no apparent defect in nucleotide affinity of the Y508C mutant, it should be noted that helicase assays are performed in optimal conditions with high concentrations of substrates and co-factors, unlikely to reflect the physiological conditions in cells. Further, in vitro assays do not recapitulate the in vivo protein interactions of Twinkle within the replisome/nucleoid (7,19,31,32). The finding of functional defects in dominant PEO Twinkle mutants in vitro, including compromised helicase activity (22, Goffart et al., manuscript in preparation), supports the involvement of different mechanisms of pathogenesis in patients homozygous for PEO mutations and in IOSCA. Interestingly, ataxia-associated recessive POLG1 mutants may also lack catalytic polymerase defects in vitro, although the dominant defects are readily demonstrated (33,34). Twinkle-Y508C may therefore share the pathogenic pathway with POLG-associated MIRAS, either involving a replicative defect highly specific for the cell type and context, or be involved in e.g. DNA repair, shown to be crucial for the cerebellum (35,36).
We then decided to study the effects of induced expression of the Y508C protein for mtDNA replication and transcription in cultured cells. However, the mutant performed exactly as the wild-type protein in these cells, maintaining normal mtDNA copy number and transcript levels. These results agree with the in vitro results, and corroborate the cell-type specificity of the phenotype. An alternative explanation is that, despite the overexpression of the Y508C mutant, the minimal amount of the endogenous Twinkle rescued a possible loss-of-function phenotype of the mutant protein. This option is, however, unlikely since a similar decrease in 7S DNA levels was achieved by overexpression of wild-type and Y508C Twinkle, indicating that the activity responsible for these changes, probably the helicase activity itself, is comparable in both. Thus, in proliferating cells, Twinkle-Y508C seems to be able to maintain mtDNA as well as the wild-type.
The Y508C mutation creates a heme-binding CXXCH-motif that is a highly conserved site for covalent heme-binding in c-type cytochromes. Furthermore, previous findings in IOSCA-patients have suggested aberrant levels of heme-biosynthetic enzymes, and the patients sometimes experience crisis reminiscent of porphyrias (37). These observations led us to hypothesize that the Y508C mutation would confer a gain-of-function to the IOSCA mutant protein, resulting in covalent heme-binding in this protein. Protein modeling supported the hypothesis, indicating a similar structural context for Y508C as for cytochrome c heme binding site. We therefore tested heme-binding on isolated Twinkle proteins, but did not detect covalent heme-binding in the IOSCA, nor the wild-type Twinkle. This indicates that the IOSCA mutant, or the wild-type Twinkle, do not bind substantial amounts of heme, at least in cells overexpressing Twinkle.
The same complexity of dominant and recessive mutations as in Twinkle exists for POLG1 mutations: dominant defects cause a myopathy—often accompanied by CNS symptoms such as parkinsonism—and recessive defects cause ataxias such as MIRAS caused by the W748S mutation. The presence of mtDNA deletions in the brain of the MIRAS patients most likely does not explain the ataxia phenotype as also adPEO patients with a different phenotype harbor high amounts of mtDNA deletions in the brain (10). However, it remains unclear what cell types contain the highest amounts of mtDNA deletions in the brain of the MIRAS or adPEO patients. In this study, the levels of mtDNA in the MIRAS brain appeared slightly decreased. However, as compound heterozygous with other POLG1 mutations, the W748S mutation causes Alpers syndrome that has been associated with mtDNA depletion in the liver, with residual mtDNA amounts of 6–40% (12,13,38), as well as in the muscle, with residual mtDNA levels numbering 25% (38). Evidence of mtDNA depletion in the frontal cortex exists for Alpers patients harboring other POLG mutations (13). It is therefore tempting to speculate that a cell type specific mtDNA depletion in large neurons, such as Purkinje cells, might account for the phenotype seen in the MIRAS patients, but this remains to be studied further. Although the levels of respiratory chain complexes appeared unchanged on western analysis, lowered immunoreactivity with antibodies against CI in the MIRAS brain suggests dysfunction of the respiratory chain. The similarities in recessive POLG and Twinkle diseases suggest a common CNS-specific need for mtDNA maintenance proteins.
We show here evidence for mtDNA depletion in the brain of IOSCA patients, indicating that IOSCA is a new member of the mtDNA depletion syndromes. We also show that CI-deficiency in the brain can underlie progressive ataxias. However, the helicase properties and nucleoid structure of IOSCA-Twinkle appeared normal, suggesting that mitotic cells, and perhaps also the non-mitotic neurons, are able to maintain sufficient levels of basal mtDNA replication. The precise molecular mechanism, accounting for the brain-specific mtDNA depletion in IOSCA, cannot be explained by current knowledge on Twinkle functions and remains to be elucidated.
MATERIALS AND METHODS
Patients and controls
Autopsy tissue samples were studied with authorization from the National Authority for Medicolegal Affairs and by permission from HUSLAB Division of Pathology (Head of Section, Meilahti Laboratories of Pathology).
Tissue samples of four IOSCA patients, comprising three females and one male, were taken at autopsies performed latest 24–48 h post mortem. The clinical course of the disease was typical for IOSCA in all these patients and further clinical details as well as the neuropathology of Patient 1 (P1) and Patient 2 (P2) have been previously described (3). In brief, all four patients died at 21–26 years of age, due to treatment-resistant epileptic attacks as the direct cause of death in all except one of the patients.
P1 developed her first generalized tonic-clonic seizure at the age of 26 years and died a couple of months later due to a prolonged status epilepticus.
P2 developed epilepsy at the age of 20 years and died at the age of 24 years, within 2 months after the onset of a treatment-resistant epileptic episode.
The disease manifestation of P3 began at 13 months of age with increasing clumsiness, head tilt, severe ataxia and loss of deep tendon reflexes. She soon lost her ability to walk independently, and developed ophthalmoplegia and severe hearing deficit by the age of 5 years, this necessitating communication by means of sign language. At the age of 21 years, she was living in a nursing home and developed her first prolonged generalized epileptic seizure postoperatively after appendicectomy. Recurrent epileptic seizures followed and she died 3 months later.
P4 developed head tilt, muscle hypotonia, ataxia and athetoid movements (most prominent in the face) at the age of 11 months. Oxcarbazepine therapy was started at 13 years of age, after one short tonic seizure (during night). She developed her first status epilepticus at the age of 15 years after scoliosis surgery, her EEG showing profuse discharges with occipital predisposition. One month after the onset of seizures, she was unable to communicate with sign language or recognize her family members. After 3 months of hospitalization and treatment for intractable epilepsy, she was relocated to an institution for mentally retarded and severely ill patients. Four years after the onset of epilepsy, her MRI showed not only spinocerebellar degeneration, but also severe central and cortical brain atrophy. She continued having episodes of intractable seizures and severe infections, and died of pneumonia at the age of 21 years.
Similarities between the acute crises in IOSCA, in response to epileptic drugs metabolized through the CYP450 pathway, and the neurological crises seen in porphyrias (disorders of the heme biosynthetic pathway), had raised an early suspicion of abnormal heme metabolism in IOSCA. This was further supported by the finding of decreased or low levels of ferrochelatase (heme synthase, the last enzyme in the heme biosynthesis pathway) when analyzed in 17 IOSCA patients (37).
The typical neuropathological findings in the four IOSCA patients included atrophy of the brain stem and the spinal cord, degeneration of the posterior spinocerebellar tracts, atrophy of the cerebellar cortex with loss of Purkinje neurons, as well as atrophy and gliosis of the dentate nuclei. In addition, the neuropathological examination of P2 and P4 revealed laminar necrosis of the cerebral cortex, especially in the visual cortex of P4. The thalami and subthalamic nuclei of P2 showed severe atrophy.
A MIRAS patient (M1), homozygous for the W748S+E1143G amino acid changes in the catalytic subunit of POLG, has previously been described as patient D2 (16). The patient died at 52 years of age, and the autopsy samples were taken 80 h post mortem. The other MIRAS patient (M2, who died at 36 years of age) has been described as patient III-7 by Van Goethem et al. (15); the samples of this patient were sufficient only for the deletion PCR analysis and quantitative PCR (frontal cortex, cerebellum and muscle).
As control samples, we used post-mortem samples from individuals of 19–65 years of age (19, 24, 33, 36, 50, 54, 57, 60 and 65 years of age), who died of non-neurological causes (cause of death: cardiac n = 3, sudden death/accident n = 2, other n = 4). The samples from the cerebral cortex (frontal cortex), cerebellum (cerebellar cortex), liver and the vastus lateralis were taken 4–7 days post mortem. In addition, we used two cerebral cortex (temporal cortex) samples taken during brain surgery for benign cerebral tumor (15 and 17 years of age). The controls comprised three females and eight males.
Tissue samples of the patients (cerebral cortex, cerebellum, liver and muscle) and controls were snap-frozen in liquid nitrogen or frozen in isopentane/liquid nitrogen and stored at −80°C. Total DNA was extracted using 0.75 mg/ml proteinase K treatment overnight at 37°C in the lysis buffer (0.5% SDS, 0.1 M NaCl, 20 mm EDTA, 50 mm Tris pH = 8.1) followed by a standard phenol-chloroform DNA extraction protocol. The DNA was precipitated with ice-cold ethanol and 0.3 M sodium acetate, and dissolved in Tris-EDTA-buffer (pH 8.0).
Analysis of mtDNA deletions
Analysis of mtDNA deletions was performed by long-range PCR as previously described (4) using 10 ng of total DNA extracted from the frontal cortex, the cerebellum, the liver and the muscle of IOSCA patients, two MIRAS patients (except one MIRAS patient for the liver) and one control sample for each tissue. Different template amounts (5, 10 and 20 ng) were used to search for the potential mtDNA deletions in the IOSCA frontal cortex and cerebellum. As positive controls, we used muscle DNA samples from a patient with multiple mtDNA deletions and a Kearns–Sayre patient with a single mtDNA deletion.
Analysis of mtDNA quantity
MtDNA quantity was determined by Southern blotting and by quantitative PCR. For Southern, total DNA (3 µg) from the tissue samples was subjected to restriction digestion by PvuII restriction endonuclease (Fermentas), which linearizes mtDNA. The digested DNA was subsequently run on 0.8% agarose gel and blotted on Hybond-N+membrane (Amersham Biosciences). 32P-labeled hybridization probes were generated by random primer labeling (Amersham Rediprime II Random Prime Labelling System) utilizing the following templates: a DNA fragment corresponding to the region 1–740 of human mtDNA, cloned in pTZ19 and a 5.8 kb fragment of human 18S ribosomal RNA gene, cloned in pBR322. Following hybridization, the radioactive signals were detected with Typhoon 9400 Imager (Amersham Biosciences), and quantified with ImageQuantTL software (Amersham Biosciences): the mtDNA signal was compared to the nuclear genomic 18S signal. MtDNA was also quantified from the cerebral cortex, cerebellum, liver and the muscle DNA using real-time q-PCR with TaqMan probes, exactly as previously described in Hakonen et al. (4). A portion of cytochrome B gene (CytB) served as the mitochondrial gene target, and a portion of a single copy gene amyloid precursor protein as the nuclear gene target. The mean values of duplicate or triplicate measurements were used for the analyses.
Analysis of mtDNA point mutations by PCR amplification, cloning and sequencing
To identify potential mtDNA point mutations in the IOSCA and MIRAS cerebellum, we PCR amplified parts of mtDNA control region (577 nt) and cytochrome b gene (cyt b, 835 nt) with the primers previously described and used for mtDNA point mutation analysis (21). PCR reactions were performed with 50 ng of template DNA, 1U of Phusion High Fidelity DNA polymerase in HF buffer (Finnzymes), 200 µM dNTPs and 0.5 µm of each primer. As DNA template, we used DNA extracted from the cerebellum of one IOSCA patient (P3), the MIRAS patient, as well as one control (50 years of age). The PCR conditions for the control region fragment were the following: initial denaturation at 98°C for 30 s, followed by 30 cycles (98°C for 10 s, 65°C for 30 s and 72°C for 30 s) and a final extension at 72°C for 10 min. The PCR conditions were similar for the cyt b fragment, except for an annealing temperature of 68°C. Subsequently, the PCR fragments were cloned into Zero Blunt TOPO PCR Cloning vector (Invitrogen) and the resulting plasmid DNAs were extracted using QIAprep Spin Miniprep Kit (QIAGEN). Altogether, ∼28 000 nucleotides of the control region and more than 35 000 nucleotides of the cytochrome b gene region were sequenced from all three samples.
Immunohistochemistry of mitochondrial respiratory chain CI, CII and CIV from brain sections
Immunohistochemistry was performed on formalin-fixed paraffin-embedded sections from the cerebellum and the cerebral cortex of two IOSCA patients, and two control cases. The sections were first deparaffinized and then processed in a microwave oven (5 min 800 W, 2 × 5 min 500 W) in 10 mm citric acid buffer (pH = 6.0). After cooling down, the sections were treated with 0.9% H2O2 to deactivate endogenous peroxidase. Immunohistochemical stainings were performed using Vectastain ABC/Elite Mouse IgG Kit (Vectastain, Vector Laboratories), according to the manufacturer's instructions. The samples were first blocked for 45 min. The primary CI antibody (NDUFB4, MS 107, Mitosciences) was diluted to 10 µg/ml in Dako REAL™ antibody diluent (Dako) and the samples incubated in this solution for 1 h in RT. The slides were rinsed in three changes of phosphate buffered saline (PBS) after every staining step. The incubations with the secondary antibody and the pre-formed avidin/biotinylated horseradish peroxidase complex were performed according to the protocol of the Vectastain ABC/Elite Mouse IgG Kit (Vectastain, Vector Laboratories). The peroxidase staining was visualized using DAB (SIGMA FAST 3,3′-Diaminobenzidine tablets, Sigma). The slides were also stained with hematoxylin (Tissue-Tek Staining Solution, Sakura Finetek). Normal mouse IgG served as the primary antibody in all stainings of negative controls. The protocol for the remaining CI, CII and CIV immunostainings was essentially the same; the CI NDUFS3 antibody (MS110, MitoSciences), the CII primary antibody (70 kDa Fp subunit, MS204, Mitosciences) and the CIV primary antibody (COXII, MS405, Mitosciences) were diluted to 5 µg/ml. The antibody against ND1 subunit of CI was a kind gift from Anne Lombes: immunohistochemistry was performed with a dilution of 1:500 of the primary antibody, and serum was utilized for the negative controls.
Western blotting of subunits of the mitochondrial respiratory chain complexes
Isolation of mitochondrial proteins from patient and control samples was performed as previously described (39) without digitonin treatment. The isolated proteins (10 µg) were resolved on 10–20% Tris–HCl Ready Gels (Bio-Rad) and blotted on PVDF membranes (Immobilon-FL, Millipore). Monoclonal antibodies against CI-NDUFA9 (39 kDa, MS111, MitoSciences), CI-NDUFB4 (15 kDa, MS107 MitoSciences), CIV subunit II (MS405, MitoSciences) and Porin 3IHL (Calbiochem) were used for the detection.
Mitochondrial isolation and blue native gel electrophoresis
Isolation of mitochondrial proteins from patient and control samples was performed as previously described (39). Native gradient gels (5–12%) were casted and run according to a previously described protocol (40). Equal amounts of protein (10 µg) from control and patient tissues were analyzed with Coomassie staining and western blot using monoclonal antibodies against CI NDUFA9 subunit (39 kDa, MS111, MitoSciences), CI NDUFB4 subunit (15 kDa, MS107 MitoSciences), CIV-cox2 subunit and CII subunit 70 kDa Fp (Mitosciences).
Creation and maintenance of stable transfected inducible expression cell lines
The full-length cDNA of Twinkle variants were cloned in the pcDNA3.1(−)/Myc-His A (Invitrogen), as previously described (7). The constructs were re-cloned in the pcDNA5/FrT/TO vector (Invitrogen) taking advantage of two PmeI restriction sites flanking the multiple cloning sites of the original pcDNA3 vectors and target vector. The resulting fusion proteins contained the sequence of the respective proteins followed by a (C-terminal) Myc-His tag. The plasmid constructs were confirmed by DNA sequencing.
Using these constructs, stable cell lines expressing wild-type or IOSCA mutant Twinkle upon induction were created as described (23) in the FlpIn™ TREx™ 293 host cell line (Invitrogen). The resulting cells were grown in DMEM medium (Sigma) supplemented with 10% FCS (Sigma), 1 mml-Glutamine, 50 µg/ml uridine (Sigma), 150 µg/ml Hygromycin and 15 µg/ml Blasticidin in a 37°C incubator at 8.5% CO2.
To induce expression, the indicated amount of doxycycline (Sigma) was added to the growth medium, and after the indicated time, the cells were processed for further analyses.
Immunofluorescent detection in 293 FlpIn TREx cells was done as described previously (32) using the monoclonal c-myc 9E10 antibody (Roche Molecular Biochemicals) and the monoclonal anti-DNA antibody AC-30-10 (PROGEN, Shingle Springs, CA, USA) as primary antibodies. As secondary antibodies anti-mouse IgG-Alexa-Fluor®488 (Invitrogen, myc) and anti-mouse IgM-Alexa Fluor®568 (DNA) were used. Image acquisition using confocal microscopy was done as described.
Cell lysates were prepared and analysed for protein expression by immunoblotting after SDS–PAGE (41). A primary monoclonal c-myc (Roche Molecular Biochemicals) antibody was used for the detection of recombinant proteins. Peroxidase-coupled secondary antibody horse-anti-mouse was obtained from Vector Laboratories. Enhanced Chemiluminescence detection was done essentially as described (41).
Twinkle protein extraction
In vitro assays for the determination of helicase activities were performed with highly enriched Twinkle preparations derived from 293 FlpIn™ TREx™ cells. Cells induced for 36 h were disrupted after short cytochalasin treatment (42); mitochondria were isolated using differential centrifugation and sucrose gradient purification and then lyzed and sonicated in high salt buffer (50 mm KH2PO4 pH 7.4, 1 M NaCl, 0.05% Triton X-100, 10 mm Imidazol and 7 mm β-mercaptoethanol). The lysate was incubated with TALON Co2+ affinity resin for 1 h at 4°C, the resin was washed twice each with high and low salt buffer (25 mm Tris pH 7.6, 200 mm NaCl, 100 mm L-Arginine pH 7.6, 40 mm Imidazol, 10% glycerol, 7 mm β-mercaptoethanol) and His-tagged proteins binding to the resin were isolated with elution buffer containing 25 mm Tris pH 7.6, 200 mm NaCl, 100 mml-Arginine pH 7.6, 200 mm Imidazol, 50% glycerol. Protein extracts were aliquoted, shock frozen in liquid nitrogen and stored at −80°C. The concentration of Twinkle in the eluate was judged by SDS–PAGE and Coomassie Brilliant blue staining using BSA as a reference.
In vitro assays
Helicase assays were performed as described (23) with a radioactively end-labeled 60 nt oligonucleotide hybridized to M13 single-stranded (ss) DNA, forming a 20 nt double-stranded stretch with a 40 nt 5′ overhang. The reaction conditions were essentially as described in Wanrooij et al. (23), but the reaction mix did not contain other DNA besides the substrate. UTP, ATP, GTP or dGTP were added at 3 mm, if not indicated otherwise. Kinetics were determined by using the indicated nucleotide concentrations in a range from 0 to 3 mm.
Oligomerization of isolated Twinkle variants in vitro was tested by cross-linking the protein with low concentrations of glutaraldehyde followed by separation on a SDS–PAGE and western blotting. About 20 ng of Twinkle protein was diluted in 20 µl of oligomerization buffer (25 mm Tris–HCl pH 7.6, 100 mml-Arginine pH 7.6, 1 mm DTT, 10% glycerol) containing either 50 mm NaCl (helicase conditions) or 400 mm NaCl (high salt conditions). 0.02% Glutaraldehyde was added and the reaction mix incubated for 5 min at room temperature. The cross-linking was stopped by adding 10 µl of SDS-containing sample buffer and directly heat-denatured for 10 min at 95°C. The denatured sample was separated on a 3–8% Tris-Acetate gel (NuPAGE, Invitrogen) and Twinkle proteins were detected by regular western blotting.
The stability of formed hexamers was also analyzed by gel filtration using a Superdex 200 10/300 GL size exclusion column (GE Healthcare). Approximately 300 ng purified Twinkle protein was separated in high salt buffer (25 mm Tris pH 7.6, 400 mm NaCl, 100 mml-Arginine, 5 mm EDTA and 10% glycerol), the protein of 500 µl fractions was concentrated using TCA/Deoxycholate precipitation and analyzed by SDS–PAGE and western blotting. The size markers used were BioRAD Gel Filtration Chromatography Standard.
Affinity to ssDNA was measured using an electromobility shift assay described by Farge et al. (43). Binding reactions were carried out in 20 µl volumes containing 20 mm Tris–HCl (pH 7.5), 10 mm MgCl2, mM dithiothreitol, 0.1 mg/ml bovine serum albumin, 10% glycerol and Twinkle proteins (60, 120 or 180 ng). Ten femtomol of a 30 bp ssDNA template 5′ labeled with 32P γ−ATP and 2 mm ATP were added, the reactions were incubated at RT for 10 min and run immediately on a 5% native PAGE in 1x TBE for 15 min at 150 V. Gels were dried and quantified by phosphoimager (Storm 840, Molecular Dynamics) using ImageQuant TL software.
Brewer–Fangman 2D neutral/neutral agarose electrophoresis (2DNAGE)
Mitochondrial nucleic acids were extracted using cytochalasine (Sigma-Aldrich) as described (42). Purified mtDNA was digested with the indicated restriction enzymes and separated by 2DNAGE as described (44,45). The gels were blotted by capillary transfer, hybridized with 32P-labeled DNA probes for human mtDNA and exposed to film. Equal loading was ensured by quantifying the 1n spots by phosphoimager using ImageQuant TL software.
Northern blots and southern blots from 293 FlpIn TREx cells
RNA for northern blot analysis was extracted from 293 FlpIn TREx cells with Trizol (Sigma) using the manufacturer's recommendations. Four microgram of total RNA per sample was run on a 1% agarose MOPS/formaldehyde gel, blotted and hybridized using standard techniques. Blots were probed with ca. 500 bp long double-stranded DNA probes radioactively labeled by random-primed labeling. Exposures were performed using phosphoimager (Storm 840, Molecular Dynamics).
The relative levels of 7S DNA in mtDNA were analyzed by one-dimensional agarose gels and Southern blotting. Total DNA was extracted from cells by proteinase K digest and phenol-chloroform extraction. Two micrograms of total DNA were digested with PvuII, heat-denatured for 5 min at 75°C before loading and separated on a 0.8% agarose gel in TBE containing 0.1 µg/µl ethidium bromide. The gels were blotted and hybridized with 32P-labeled DNA probes containing the D-loop region of human mtDNA. The signal of 7S and mtDNA was quantified by phosphoimager.
Probes used for northern and Southern blots referred to the following sequences of human mtDNA: 12S 652–1156 bp, ND2 4481–4989 bp, COXI 5970–6478 bp, ND6 14261–14623 bp, CytB 14846–15358 bp and D-loop 16343–151 bp.
The structure of the helicase domain of bacteriophage T7 helicase/primase (PDB-ID: 1q57), homologous to the helicase domain of human Twinkle, served as a tool to estimate the structural location of the Y508C mutation in the Twinkle protein. According to the sequence alignment (7), the Y508C position roughly corresponded to the region near K416 in the T7 helicase. We then compared the local structure of the T7 helicase around K416 to a heme-binding motif of cytochrome c to evaluate the compatibility of the structural features for covalent heme-binding (Cytochrome c2 from Paracoccus denitrificans was used as a sample structure, PDB-ID: 1COT).
Detection of covalently bound heme using enhanced chemiluminescence
In order to obtain highly enriched Twinkle preparations from 293 FlpIn TREx cells, the cell cultures were induced with 20 ng/ml doxycyclin for 1 day before harvesting. Cells were pre-swollen for 10 min in 4 mm Tris–HCl pH 7.8, 2.5 mm NaCl 0.5 mm MgCl2 on ice and cells were homogenized using a glass-Teflon potter; the swelling was immediately stopped by adjusting the buffer to 40 mm Tris–HCl, 25 mm NaCl and 5 mm MgCl2. After several low-speed centrifugations discarding other cellular organelles, the mitochondria were pelleted by centrifugation at 12 000 g, and lyzed in buffer containing 50 mm NaPO4, 1 M NaCl, 20% glycerol, Complete™ protease inhibitors (Roche) and 1.5% laurylmaltoside. After addition of 10 mm Imidazole, the lysate was incubated with TALON Co2+ affinity resin for 1 h at 4°C. The resin was subsequently washed with buffer containing 50 mm NaPO4 (pH = 7), 600 mm NaCl, 20% glycerol and 20 mm Imidazol, and another buffer containing 50 mm NaPO4 (pH = 7), 200 mm NaCl, 20% glycerol. His-tagged proteins binding to the resin were isolated with elution buffer containing 25 mm NaPO4 (pH = 7), 200 mm NaCl, 100 mM l-Arginine-HCl (pH 7), 20% glycerol, 100 mm EDTA.
Covalently bound heme was detected with a previously described method (46): the purified Y508C and wild-type proteins were incubated for 15 min at 40°C in a loading buffer containing no β-mercaptoethanol (124 mm Tris, ph 7.0, 20% glycerol, 4.6% SDS). 1.5 µg of cytochrome c was used as a positive control. Following incubation, the samples were separated on 10–20% Tris–HCl gradient gels (Bio-Rad) and transferred to membranes by western blotting. Covalently, bound heme was then detected with ECL Plus Western Blotting Detection Reagents (Amersham Biosciences). The same membrane was subsequently used for the detection of the Twinkle proteins by immunostaining with the c-Myc antibody (Santa Cruz Biotechnology).
We thank the following organizations for financial support: the University of Helsinki, the Academy of Finland, Sigrid Juselius Foundation (to A.S. and J.N.S.); Finnish Foundation for Pediatric Research (to T.L.); the EU sixth Framework Programme for Research (to S.G and J.N.S.); Tampere University Hospital Medical Research Fund [9G072, 9H079] (to J.N.S.) and Biomedicum Helsinki foundationc (to A.H.H.).
We thank the patients’ families for their participation in the study. Anu Harju and Markus Innilä are thanked for skillful technical assistance, Dr Anne Lombes for ND1 antibody, and Dr Roland Lill for methodological advice.
Conflict of Interest statement. None declared.