MT-ND5 Mutation Exhibits Highly Variable Neurological Manifestations at Low Mutant Load

Mutations in the m.13094T>C MT-ND5 gene have been previously described in three cases of Leigh Syndrome (LS). In this retrospective, international cohort study we identified 20 clinically affected individuals (13 families) and four asymptomatic carriers. Ten patients were deceased at the time of analysis (median age of death was 10 years (range: 5·4 months−37 years, IQR = 17·9 years). Nine patients manifested with LS, one with mitochondrial encephalomyopathy, lactic acidosis and stroke-like episodes (MELAS), and one with Leber hereditary optic neuropathy. The remaining nine patients presented with either overlapping syndromes or isolated neurological symptoms. Mitochondrial respiratory chain activity analysis was normal in five out of ten muscle biopsies. We confirmed maternal inheritance in six families, and demonstrated marked variability in tissue segregation, and phenotypic expression at relatively low blood mutant loads. Neuropathological studies of two patients manifesting with LS/MELAS showed prominent capillary proliferation, microvacuolation and severe neuronal cell loss in the brainstem and cerebellum, with conspicuous absence of basal ganglia involvement. These findings suggest that whole mtDNA genome sequencing should be considered in patients with suspected mitochondrial disease presenting with complex neurological manifestations, which would identify over 300 known pathogenic variants including the m.13094T>C.

The m.13094TNC mutation, in the MT-ND5 gene, is considered a rare pathogenic variant that has been previously reported in association with LS but the full phenotypic spectrum has remained poorly understood (Valente et al., 2009;Ching et al., 2013). We present the clinical, radiological, and histopathological data of 24 subjects who harbor the m.13094TNC mutation. We have also studied the neuropathological changes in two patients with LS/MELAS overlap syndrome, to fully elucidate the spectrum of m.13094TNC-related mitochondrial disease and to offer guidance on management and genetic counseling.

Study Design and Patients
This retrospective, international cohort study was done at the NHS Highly Specialised Service-funded Mitochondrial Diagnostic Centers in Newcastle upon Tyne and London, UK. These included patients referred from four other countries: China, South Africa, Greece and Japan. Eligible participants were genetically confirmed to harbor the m.13094TNC mutation (p.Val253Ala) in MT-ND5 over a 17 year period (January 2000-October 2017); their maternal family members were traced and examined whenever possible.
This study was approved and performed under the ethical guidelines issued by our institution for clinical studies, and complied with the declaration of Helsinki.

Clinical Presentation and Phenotypic Evaluation
Each patient was assessed by a specialist (pediatrician, adult neurologist and/or ophthalmologist) at each center, and their medical records were comprehensively reviewed. The clinical presentation of individual patients and their family pedigree are provided in Supplemental data and Supplemental Fig. 1. Where possible, common clinical syndromes described in mitochondrial disease were assigned (and agreed by all authors).

Histopathological and Biochemical Studies
Standard histological (hematoxylin and eosin (H&E) and modified Gomori Trichome stains), histochemical (cytochrome c oxidase (COX), succinate dehydrogenase (SDH), and sequential COX-SDH and biochemical assessments of muscle tissue were performed as described elsewhere (Supplemental Table 1). Fresh tissue was fixed in 2.5% glutaraldehyde buffered to pH 7.4 with phosphate buffer and post-fixed in OsO4 and embedded in Epon LX-112. Ultra-thin sections were examined in an electron microscope (JEOL•1500, Japan) for Patient 5.

Neuropathological Studies
Neuropathological investigations were performed on postmortem brain tissues from two patients (patients 1.2 and 8). Cresyl fast violet (CFV), H&E and Luxol fast blue with H&E counterstain were used on formalin-fixed paraffin-embedded tissues to determine neuronal population density and degree of myelination. Immunohistochemistry to determine the expression levels of mitochondrial respiratory chain subunits, including complex I subunit NADH: ubiquinone oxidoreductase subunit B8 (NDUFB8) and complex II subunit succinate dehydrogenase subunit A (SDHA), were performed on 5 μm thick sections as previously described (Lax et al., 2012).

Statistical Analysis
Non-parametric, continuous data were presented as median (range and inter-quartile range (IQR)). Correlation of mtDNA heteroplasmy level of different tissues and age of disease onset was examined using Spearman rank correlation test. Statistical significance was determined at p b 0.05. Data were managed and analyzed with IBM SPSS for Windows version 22.

Patient Cohorts
Clinical features of 24 individuals from 13 families are summarised in Table 1. The median age of disease onset was 5.5 years (n = 20, range: 6 weeks to 34 years, IQR = 13 years; the age of onset not known for two patients). Four individuals were clinically unaffected; they were ascertained pre-symptomatically due to diagnosis in other family members. Ten patients (42%) were deceased, and the median age of death was 10 years (range: 5.4 months to 37 years, IQR = 17.9 years).
Eleven patients presented with distinctive clinical syndromes previously described in mitochondrial disease. However, the remaining patients (n = 9) manifested with an isolated neurological symptom or overlapping syndromes, of which central nervous system was most commonly affected (Table 1). Centrally mediated respiratory failure or apnoea was documented in nine patients (38%) with LS, including an adult patient who presented at the age of 22 years (Patient 10). Refractory focal onset seizures with or without evolution to bilateral convulsions occurred in eight patients (35%).

Radiological Imaging
Cranial magnetic resonance imaging (MRI) was available for analysis in 17 patients (Table 2, Fig. 1 and Supplemental Fig. 2). The most common T2/FLuid Attenuation Inversion Recovery (FLAIR)-signal abnormalities were present in the midbrain (n = 12) and thalamus (n = 9: bilateral, symmetrical changes; n = 3: unilateral) followed by pons (n = 10) and medulla (n = 9). Brainstem changes in two patients (Patients 6 and 8), initially was initially misdiagnosed as a low-grade glioma, although gadolinium-enhancement was absent. Subacute cortical and subcortical signal changes were identified in six patients that were consistent with stroke-like lesions. Two patients had multiple cortical lesions including lesions within the cerebellar hemispheres (Patients 1.2 and 5), suggestive of cross cerebellar diaschisis. Cervical cord lesions were also identified in six patients.
Signal abnormalities within the basal ganglia, classically seen in LS, were only identified in two patients (Patients 11 and 13).

Histopathological and Biochemical Analyses
Histopathological and histochemical evaluation was normal in most patients except minor changes detected in the skeletal muscle of three patients: minor ragged red fibers (n = 1), lipid droplets and abnormal mitochondrial ultrastructure (n = 1) and occasional angular atrophic fibers (n = 1). Mitochondrial respiratory chain activity was measured in skeletal muscle tissue from 10 patients. One patient demonstrated isolated complex I deficiency, one patient with combined complex I and III deficiencies, two patients with combined complex I and IV deficiencies and one patient had low complex I and II activity in postmortem tissue, which was caused by significant delay in tissue handling. The remaining six cases demonstrated normal respiratory chain activity (Supplemental Table 1).

Molecular Genetics
Thirteen probands were identified to harbor the m.13094TNC mutation through whole mtDNA sequencing after screening negative for common mtDNA point mutations including m.3243ANG, m.8344ANG, MTATP6, and MTATP8 genes. The mutation was identified in nine individuals by direct sequencing of the mtDNA point mutation through pedigree and segregation analysis. Maternal inheritance of the m.13094TNC mutation was confirmed in six family pedigrees. The m.13094TNC mutation was not detected in the blood of Patient 1 (measured at age 1·2 (F) Offspring 10/14* MELAS/LS  37 years) with a severe, adult-onset MELAS phenotype although was detected at 51% in urine. Patient 2.2 was presumed an obligate carrier even though the m.13094TNC mutation was not detectable in blood (measured at age 74 years). There was no tissue sample available for testing in Patient 2.3. The asymptomatic mother of patient 3 was demonstrated to harbor the m.13094TNC mutation at mtDNA heteroplasmy   levels of 19%, 27% and 45% in blood, muscle, and urine respectively (measured at age 24 years). The quantification of mutant mtDNA heteroplasmy level was not performed in two patients due to no access to tissue samples (Patient 2.3) and only Sanger sequencing was performed in Patient 11 (the mutant load in muscle appeared higher than blood; sensitivity N 25%). There was a statistically significant negative correlation between blood mtDNA heteroplasmy level and age (Spearman rho = −0.883, p b 0.001; Fig. 2A). Among the patients with mutant mtDNA load quantified in more than one tissue (n = 12), the mutant heteroplasmy levels in muscle (65% ± 20%; n = 10) and/or urine (40% ± 27%; n = 8) were consistently higher than blood (24% ± 22%; n = 16) (Fig. 2B). The discrepancy between muscle and blood mtDNA heteroplasmy ranged from 3 to 75% in seven patients; the discrepancy between urine and blood mtDNA heteroplasmy levels ranged from 5 to 51% in seven patients. Whole exome sequencing (methods described elsewhere (Neveling et al., 2013)) was performed to investigate the cause of optic neuropathy in Patient 7. This excluded other known causative genes (including OPA1).
The m.13094TNC mutation load of a number of post-mortem tissues was analyzed from Patients 1.2 and 13 exemplifying the varied tissue segregation that is characteristic of this mtDNA mutation (Table 3). The m.13094TNC mutation load was specifically determined in brain homogenate samples from patients 1.2 and 8 demonstrating lower mutant load within neuronal tissues (median = 61%, range: 47-69%) compared to other tissues despite both manifesting with extensive neurological sequelae (Table 3).
Muscle (Spearman rho = 0.192, p = 0.62) and urine (Spearman rho = −0.80, p = 0.20) mutant mtDNA heteroplasmy levels did not correlate with the age of disease onset. There was no significant difference in mean mutant heteroplasmy level in different syndromic categories (p = 0.122).

Neuropathological Findings
The major neuropathological findings for patients 1.2 and 8 are summarised in Supplemental Table 2). In patient 1.2, posterior cerebellar cortex demonstrated multiple areas of necrotic lesions ranging from atrophy of the molecular layer, Purkinje cell dropout and granule cell loss (Fig. 3A), to total necrosis of the cerebellar cortex that also affected the underlying white matter (Fig. 3B). Concerning topographical distribution, lesions were most severe in the brainstem nuclei ( Fig. 3C and D), the thalamic and subthalamic nuclei ( Fig. 3E and F) and primary visual cortex (Brodmann area 17) (Fig. 3G). At a microscopic level, changes such as prominent capillary proliferation, microvacuolation and severe neuronal loss were frequently observed in fixed cerebral and cerebellar hemispheres, and right brainstem compatible with LS. Downregulation of complex I subunit (NDUFB8) was evident in the in the pons and cerebellar cortex of patient 1.2, and lesioned thalamus and occipital cortex of patient 8 (Supplemental Fig. 3).

Discussion
In this study, we have identified marked clinical heterogeneity with a continuous spectrum of overlapping symptoms associated with the m.13094TNC mutation. However within that clinical heterogeneity, at certain points in the course of disease, several distinct clinical syndromes associated with mutations in the MT-ND5 gene were clearly discernible: early onset LS (38%), late childhood/early adulthood-onset LS/ MELAS overlap syndrome (13%) and LHON (4%) (not previously recognized in association with the m.13094TNC mutation). LS frequently manifested with a typical illness trajectory including hypotonia, cranial nerve palsies, cerebellar ataxia, and developmental regression; with relapses in clinical status often triggered by inter-current illness. Brainstem dysfunction (manifesting as centrally-mediated respiratory failure and lability in blood pressure), commonly occurred as a preterminal event. Patients with the LS/MELAS overlap syndrome typically presented with refractory focal seizures and stroke-like lesions (as classically seen in MELAS). Additional features of brainstem dysfunction, including acute-onset ptosis, a complex eye movement disorder and worsening ataxia (as classically seen in LS) were not infrequent. Neuropsychiatric symptoms such as excessive anxiety, low mood, and hypersomnolence, in addition to cognitive impairment were features accompanying stroke-like episodes. A progressive gait disorder with truncal ataxia was a prominent clinical feature in three adult patients (13%), and additional neurological features such as strabismus (Patient  Table 1. 58% -2.4), axonal neuropathy and stroke-like episode (Patient 2) were identified. Interestingly, optic neuropathy was the first clinical manifestation in four patients (Patients 1, 2.4, 7 and 7.1); however, additional severe central nervous system (CNS) involvement evolved in two patients (Patients 1 and 2.4). These findings highlight that as the disease progressed, discrete syndromes were no longer discernible. Indeed, the heterogeneity of the neurological manifestations in the m.13094TNC mutation appears similar to that observed in other mtDNA encoded complex I gene mutations. The most common T2/FLAIR signal abnormalities on cranial MRI were localized to the brainstem (71%), thalamus (71%), cerebral cortex (35%) and cervical cord (31%) and medial thalamic changes (with restricted diffusion) were identified in seven out of eight patients presenting with refractory epilepsy. Such findings suggest prolonged seizure activity (focal status epilepticus) as the underlying pathophysiological mechanism of the radiological (and clinical) changes observed (Katramados et al., 2009). MRI abnormalities in cortical and/or subcortical areas, cerebellar hemispheres (crossed cerebellar diaschisis, which refers to cerebellar hypometabolism is ascribed to functional disconnection of the contralateral hemisphere from the cerebral cortex) (Tien and Ashdown, 1992) corpus callosum, thalami and basal ganglia have been increasingly reported in cases of isolated seizures, seizure clusters, and status epilepticus (Katramados et al., 2009). Moreover, the development of cognitive dysfunction in these patients may be explained by the observed involvement of the cerebello-thalamo-cortical pathway (white matter tracts connecting the cerebellar cortex to the contralateral various cerebral cortices, passing through the superior cerebellar peduncle, red nuclear and the thalamus) (Palesi et al., 2015). T2/ FLAIR signal abnormality involving the brainstem extending into the upper cervical cord was identified in four patients. Interestingly, Patient 5 had quite extensive T2/FLAIR signal changes spanning from C2 to C6, mimicking a demyelinating disorder but CSF restricted oligoclonal bands and anti-aquaporin 4 antibodies both tested negative. Whilst MR imaging findings are consistent with previously documented changes in other complex I deficient patients, it is striking that only two patients with the m.13094TNC mutation had evidence of striatal lesions (Valente et al., 2009) compared to 90% of complex I cases reported by Lebre et al. (2011)) These findings are corroborated by relative preservation of the striatum and the presence of devastating destruction of the thalami and brainstem with marked capillary proliferation on autopsy examination. Whilst symmetrical signal abnormalities in basal ganglia are the most common radiological finding in patients affected by LS, lesions may merely be confined to the brainstem (Leigh, 1951;Rahman et al., 1996), as was observed in all our cases. This again serves as a reminder that the diagnosis of LS (including due to MT-ND5 mutations) should be considered even in the absence of symmetrical basal ganglia lesions. The radiological appearances of patients with LS harboring the m.13094TNC mutation bear similarities to mitochondrial diseases caused by autosomal recessive tRNA synthetase mutations, such as leukoencephalopathy with thalamus and brainstem involvement and high lactate (LTBL) caused by EARS2 mutations and leukoencephalopathy with brainstem and spinal cord involvement and lactate elevation (LBSL) caused by DARS2 mutations (Diodato et al., 2014). In two of our cases, the radiological diagnosis was thought to be a space-occupying lesion at presentation. However, the 'relapsingremitting' course of neurological deficits associated with m.13094TNCrelated mitochondrial disease, coupled with the evolution of signal abnormalities on interval brain imaging and demonstration of abnormal signals in different regions of structurally normal brain on T1 images (but abnormal cerebral metabolism on MR spectroscopy), should aid clinicians to discern between the two conditions.
The clinical features of these patients may also overlap at various stages of the disease with other more common forms of mitochondrial disease, such as LHON, OPA1, and POLG-related disease, and may mimic other forms of hereditary ataxia. In addition, m.13094TNC-related mitochondrial disease may mimic inflammatory CNS disorders such as acute disseminated encephalomyelitis (ADEM), multiple sclerosis (MS) and neuromyelitis optica (Patients 5, 6 and 7). Indeed, the link of an MS-like illness and common LHON mutations is well recognized in Harding disease, where imaging appearances are frequently indistinguishable from MS (Matthews et al., 2014). More recently, the coincidental occurrence of an MS-like disorder and OPA1 mutations in three unrelated patients has also been reported (Yu-Wai-Man et al., 2016).
We observed that the histochemical changes such as ragged red and COX deficient fibers, and biochemical analysis of mitochondrial respiratory chain activity in muscle tissue, was normal in 60% of the cases presented here, similar to other point mutations in MT-ND5 (Kirby et al., 2003;Liolitsa et al., 2003;Crimi et al., 2003). In addition, the quadruple immunohistochemistry technique could not detect the reduction of complex I subunit in muscle tissue in one of our patients (Ahmed et al., 2017), in accordance with the findings of preferential CNS manifestations in association with this mutation. Interestingly, the threshold level of mutant mitochondrial load necessary to cause symptoms in these cases appears to be much lower than in mitochondrial tRNA gene-related disorders associated with severe clinical phenotypes. However, our findings corroborate previous findings in transmitochondrial cybrid models of the m.13094TNC mutation (Valente et al., 2009). Furthermore, severe clinical phenotypes associated with a relatively low mutant load have also been observed in other common point mutations in the MT-ND5 gene, including the m.13513G N A mutation, where a mutant load of b50% in muscle has been reported in association with LS (Kirby et al., 2003) and MELAS (Shanske et al., 2008). Preliminary pedigree analysis did not initially prove informative; however, extensive segregation testing confirmed this to be a maternally inherited genetic disorder. In addition, our findings highlight the importance of recognizing that mtDNA disease can only be definitively excluded by performing whole mitochondrial genome sequencing in post-mitotic tissues, such as muscle. In the case of predictive testing, we would advocate screening for the m.13094TNC mutation in muscle and/or urine to minimize the chance of falsenegative results. The variable tissue segregation was further evident and exemplified by the variable mutation load in multiple tissues at autopsy (Patients 1.2 and 13). Intriguingly, the variable phenotypic threshold effect associated with the m.13094TNC mutation would support the assertion that MT-ND5 synthesis is a rate limiting step for complex I activity (Sazanov, 2015), given the relatively low mutant load resulting in severe phenotypic expression (Blok et al., 2007). Although we do not have serial measurements of the m.13094TNC mutant load in blood for individual patients, our data suggest that mutant mtDNA heteroplasmy level may decline with age, and we would hypothesize that there is a negative selection of this mutant mtDNA in bone marrow over time, as demonstrated in other commoner mtDNA mutations such as m.3243ANG and m.13513GNA. However, we concede there is a possibility of ascertainment bias in the older individuals with lower mutant heteroplasmy levels.
Patients 1.2 and 8 had MELAS/LS overlapping syndrome whilst Patient 13 had LS only. Patient 1.2 predominantly manifested with a childhood onset CNS disorder, with no evidence of myopathy (on clinical examination), cardiomyopathy, renal impairment or hepatic disease. However, in Patient 1.2, the mutant heteroplasmy levels were much higher in skeletal muscle (88%), heart (75%), kidney (83%) and liver (83%) compared to CNS tissues (~67%). Only patient 8 manifested with stroke-like episodes and subsequently a brainstem crisis in her 30s, yet the mutant mtDNA heteroplasmy levels in CNS tissues (~52%) were lower than in Patient 1.2. The tissue specificity of phenotypic expression in this mutation is intriguing, and the possible underlying reasons require further elucidation yet support the observation that mutant heteroplasmy level alone does not fully explain clinical manifestation (Gorman et al., 2016).
In summary, the m.13094TNC mutation exhibits highly variable neurological manifestations and is frequently associated with high disease burden and early mortality. Cortico-thalamic-cerebellar involvement appears to be a frequent finding in these patients compared to other rare mtDNA mutations, and may serve as a radiological biomarker of m.13094TNCrelated mitochondrial disease. Moreover, our findings would suggest m.13094TNC-related mitochondrial disease is perhaps not as rare as originally thought. We would suggest that consideration should be given to screening the whole mtDNA genome in clinically relevant tissues, prior to proceeding to whole exome sequencing, in those who have tested negative for the commonly recognized disease causing genes and mtDNA point mutations. Once again, the observed clinical heterogeneity, often apparent lack of maternal inheritance, normal histological and biochemical muscle biopsy findings, and variability in tissue segregation in these cases, highlights the diagnostic challenges of mitochondrial disease caused by rare mtDNA variants. We suggest, these findings support a better understanding of m.13094TNC -related syndromes and their inherent clinical trajectory and as such will serve to aid a more timely diagnosis, inform accurate genetic counseling, and facilitate tailored therapeutic interventions.