Complex hereditary peripheral neuropathies caused by novel variants in mitochondrial-related nuclear genes

Mitochondrial disorders are a group of clinically and genetically heterogeneous multisystem disorders and peripheral neuropathy is frequently described in the context of mutations in mitochondrial-related nuclear genes. This study aimed to identify the causative mutations in mitochondrial-related nuclear genes in suspected hereditary peripheral neuropathy patients. We enrolled a large Japanese cohort of clinically suspected hereditary peripheral neuropathy patients who were mutation negative in the prescreening of the known Charcot–Marie–Tooth disease-causing genes. We performed whole-exome sequencing on 247 patients with autosomal recessive or sporadic inheritance for further analysis of 167 mitochondrial-related nuclear genes. We detected novel bi-allelic likely pathogenic/pathogenic variants in four patients, from four mitochondrial-related nuclear genes: pyruvate dehydrogenase beta-polypeptide (PDHB), mitochondrial poly(A) polymerase (MTPAP), hydroxyacyl-CoA dehydrogenase/3-ketoacyl-CoA thiolase/enoyl-CoA hydratase, beta subunit (HADHB), and succinate-CoA ligase ADP-forming beta subunit (SUCLA2). All these patients showed sensory and motor axonal polyneuropathy, combined with central nervous system or multisystem involvements. The pathological analysis of skeletal muscles revealed mild neurogenic changes without significant mitochondrial abnormalities. Targeted screening of mitochondria-related nuclear genes should be considered for patients with complex hereditary axonal polyneuropathy, accompanied by central nervous system dysfunctions, or with unexplainable multisystem disorders. Supplementary Information The online version contains supplementary material available at 10.1007/s00415-022-11026-w.


Introduction
Peripheral neuropathy has various causes, one of which is mitochondrial abnormalities. Mitochondrial-related nuclear genes, such as MFN2 and GDAP1 that are involved in mitochondrial dynamics, are major causes of Charcot-Marie-Tooth disease (CMT) [1,2], the most common subtype of hereditary peripheral neuropathy. In MFN2, known as CMT2A and HMSN6A, phenotypes may cause optic atrophy, and in GDAP1, they may cause vocal cord paresis [2] so the spectrum of CMT has considerably broadened and multisystem involvement is frequently observed similar to other disorders caused by mutations in mitochondrial DNA (mtDNA) or mitochondria-related nuclear genes. Moreover, mitochondrial disorders associated with defects in mitochondrial DNA (mtDNA) maintenance and replication or defects in the respiratory chain complex are often associated with peripheral neuropathy [2]. Although the severity of these disorders is usually mild or subclinical, peripheral neuropathy can be severe and might be the main feature of a mitochondrial disorder [3]. Given that > 1100 mitochondrialrelated nuclear genes have been identified and > 240 nuclear genes cause mitochondrial disorders [4,5]; we speculate that more mutations in mitochondrial-related nuclear genes cause patients to manifest the CMT-like phenotype. In this study, through whole-exome sequencing (WES) data, we examined a large cohort of Japanese patients with clinically suspected hereditary peripheral neuropathy patients to determine the presence of variants in a panel of mitochondrial-related nuclear genes.

Patient selection and extraction of genomic DNA
After preliminary exclusion of the PMP22 duplication or deletion mutation, using fluorescence in situ hybridization or multiplex ligation probe amplification, we enrolled a nationwide cohort of 854 Japanese patients clinically suspected with pure or complex hereditary peripheral neuropathy, between April 2007 and July 2014. All of their clinical information, electrophysiological and radiological records, and pathological findings, were provided by local neurologists or pediatricians. The protocol was reviewed and approved by the Institutional Review Board of Kagoshima University (Kagoshima, Japan). All patients and family members provided written informed consent to participate in this study. The study conforms with the World Medical Association Declaration of Helsinki published on the website of the Journal of American Medical Association.
Genomic DNA was isolated from peripheral blood leukocytes, using the Qiagen Puregene Core Kit C (Qiagen, Valencia, CA, USA), or from the saliva, using the Oragene DNA self-collection kit (DNA Genotek, Ottawa, ON, Canada), according to the manufacturer's protocol.

Gene panel screening and WES
All 854 DNA samples were processed on 1 of the 2 types of CMT-related gene panel screening platforms. Between April 2007 and April 2012, 417 cases were screened using a customized MyGeneChip® CustomSeq® Custom Resequencing Array (Affymetrix, Inc., Santa Clara, CA, USA), targeting 28 genes known to cause CMT or related diseases following a protocol described previously [6]. Between May 2012 and July 2014, we used the Illumina MiSeq next-generation sequencing platform to screen 437 patients for 40 known CMT disease-causing and 20 candidate genes [7]. After target resequencing, we used WES to further analyze 399 mutation-negative patients, including 247 patients with autosomal recessive (AR) or sporadic inheritance.
We used a SureSelect v4 + UTRs or v5 + UTRs kit, then sequenced on Illumina HiSeq 2000® (Illumina, San Diego, CA, USA). WES data were aligned to the human reference genome (NCBI37/hg19) with Burrows-Wheeler Aligner [8], and variant call was performed using SAM tool [9], The CLC Genomic Workbench software program (Qiagen, Hilden, Germany) and an in-house R script were applied for variant annotation and filtering.

RNA extraction and reverse-transcription polymerase chain reaction (RT-PCR)
Total RNA was extracted from whole blood of Patient 4 using the PAXgene Blood RNA Kit (Qiagen). Subsequently, complementary DNA (cDNA) was produced with a highcapacity cDNA reverse-transcription kit (Applied Biosystems, Carlsbad, CA, USA) according to the manufacturer's instructions. To analyze the effect of the splice site variant in the succinate-CoA ligase ADP-forming beta subunit (SUCLA2) gene, we amplified SUCLA2 cDNA using the following primer pairs:

Clinical studies
Clinical findings and laboratory data, nerve conduction studies (NCS), and image examinations were based on the currently available information for all patients. The primary physician performed histological investigations of the sural nerve biopsy in Patient 1 as well as skeletal muscle biopsies in Patients 3 and 4, after obtaining informed consent. In Patient 4, respiratory chain enzyme activities in the skeletal muscle homogenate were also assayed as described earlier [11].

Results
From WES data of 247 CMT patients with AR or sporadic inheritance, we concentrated all uncommon variants (allele frequency < 0.05) in the mitochondria-related 167 nuclear gene panel. Therein, bi-allelic variants were identified in four patients, from four distinct genes. These genes were pyruvate dehydrogenase, beta-polypeptide (PDHB), mitochondrial poly(A) polymerase (MTPAP), hydroxyacyl-CoA dehydrogenase/3-ketoacyl-CoA thiolase/enoyl-CoA hydratase, beta subunit (HADHB), and SUCLA2. These variants comprise c.880G > A (p.G294R) homozygous variants in PDHB (Fig. 1a), c.833G > T (p.R278I) and c.1531C > T (p.Q511*) compound heterozygous variants in MTPAP (Fig. 1b), c.1192 T > C (p.F398L) homozygous variants in HADHB (Fig. 1c), and c.664-1G > A and c.1,300delG (p.D434fs) compound heterozygous variants in SUCLA2 (Fig. 1d). All the missense variants were novel and located at a highly conserved amino acid residues ( Fig. 1e-h), and most variants were located in each protein domain region ( Fig. 1i-l). All variants were absent in the control database and predicted to be damaging or deleterious during in silico analysis; the in silico analysis results as well as the classification by ACMG standards and guidelines are shown in Supplementary Table 2. The genetic and clinical presentations of the patients are summarized in Table 1. Segregation analyses were performed for the parents of all four patients, except Patient 3, whose parents were deceased. The same genotype was detected in an affected elder brother of Patient 2 (Fig. 2).

Patient 1 [PDHB c.880G > A (p.G294R)]
This patient was a woman in her 30 s who was born weighing 2400 g. By 2 years of age, she had a ventricular septal defect and seizures four times. She began walking at 1 year and 6 months but had gait disturbance. The surgery of the strabismus was performed at age 4 and of the ileus at age 15. Gait disturbance progressed, and sural nerve biopsy was performed at age 25. A decrease in the density of large myelinated fibers with thin small myelinated clusters was revealed by nerve analysis (Fig. 3a). Simultaneously, adjustment disorder associated with mild-to-moderate intellectual disability was observed. Cerebral atrophy with no white matter lesions was shown by brain magnetic resonance imaging (MRI) (Fig. 3b). Clinically, the patient had weakness and atrophy of the distal limb muscles, pes cavus, and hallux valgus. Disturbance of the leg vibration sense and decreased tendon reflexes were also noted. Axonal polyneuropathy was revealed by NCS (Table 1).

Patient 2 [MTPAP c.833G > T(p.R278I) c.1531C > T (p.Q511*)]
This male patient in his 30 s who had gait disturbance at the age of 1 year and 6 months. At 28 years, scoliosis, equinus foot, and paresthesia were present, and he could not walk. Sensory-dominant axonal polyneuropathy was revealed by NCS (Table 1), and he had skin acne through the lumbar region from the head. Low levels of vitamin B 12 (70 ng/ mL; normal: 180-914 ng/mL) and folate (1.3 ng/mL; normal: > 3 ng/mL) were revealed in laboratory investigations, but the intrinsic factor antibody and gastric parietal cell antibody were negative. An upper gastrointestinal endoscopy showed only atrophic gastritis. Patchy T2 hyperintensities in the white matter of the temporal and occipital lobes (Fig. 3c) were revealed by brain MRI.
During elementary school, his elder brother, who had the same MTPAP variants, also had gait disturbance, amblyopia, and scoliosis, and he began using a wheelchair at 38 years of age. We observed hypoesthesia and optic atrophy, and axonal polyneuropathy was revealed by NCS. His vitamin B 12 levels were low, and despite supplementation, the symptoms worsened.

Patient 3 [HADHB c.1192 T > C (p.F398L)]
Details about this case were previously established [12]. A man in his 50 s had languor of the lower extremity and brown urine after exercise at the age of 10. For a while, symptoms improved but recurred at age 45, and gait disturbance progressed. At age 55, deep tendon reflexes were and HADHB (g) genes and canonical GT-AG nucleotides (c.664-1 g) of the splice acceptor junctions in the SUCLA2 gene (h). These mutation sites were highly conserved. i-l Predicted positions of amino acids affected and domain structure of PDHB protein (i), MTPAP protein (j), HADHB protein (k), and SUCLA2 protein (l) based on an NCBI conserved domain search (http:// www. ncbi. nlm. nih. gov/ Struc ture/ cdd/ wrpsb. cgi). Arrows indicate mutation sites. aa amino acids, Mut mutant, WT wild type absent, and muscle weakness and atrophy in his distal lower limbs were present. His senses of vibration and position were deeply disturbed, and sensory ataxia was present. Axonal sensorimotor polyneuropathy was shown by NCS (Table 1) and normal creatine kinase levels by laboratory investigations. After gene analysis, he suffered from recurrent rhabdomyolysis. Neurogenic changes, such as marked fiber-type grouping were revealed by open muscle biopsy (Fig. 3d), but there was no mitochondrial abnormality. His 3-hydroxy-tetradecanoyl carnitine (C 14 -OH) level was slightly increased (0.06 nmol/mL; normal: < 0.05 nmol/mL).

Patient 4 [SUCLA2 c.1300delG (p.D434Mfs*8) c.664-1G > A (p.L222_Q224del)]
Details about this case were previously established [13]. This infant girl was born weighing 2,450 g. At 3 months, ptosis, deafness, scaphocephaly, and axial hypotonia were observed. At 7 months, profound intellectual disability was noted. At 1 year, she developed dysphagia, and tube feeding was started. At 2 years, gastroesophageal reflux and scoliosis were observed accompanied by cyclic vomiting and hypercapnia during sleep. NCS showed that axonal polyneuropathy predominantly affected the motor nerves (Table 1). Laboratory tests showed increased levels of mild methylmalonic aciduria (34.1 μg/mg creatinine; normal: < 10 μg/mg creatinine), blood lactate (58.5 mg/ dL; normal: 4-17 mg/dL), and pyruvate (2.17 mg/dL; normal: 0.3-0.9 mg/dL), and cerebrospinal fluid lactate and pyruvate levels were 27.9 mg/dL and 1.81 mg/dL, respectively. Brain MRI revealed cerebral atrophy and T2 hyperintensities in the bilateral putamen and caudate nuclei (Fig. 3e) and a lactate peak in magnetic resonance spectroscopy (Fig. 3f). Therefore, Leigh syndrome was suspected; however, there was no known mitochondrial DNA mutation in peripheral blood lymphocytes. Open muscle biopsy showed a neurogenic change in the fibertype grouping (Fig. 3g), but no insufficient cytochrome c oxidase activity, ragged red fibers, or increased staining for succinate dehydrogenase were observed. Hence, she was tentatively diagnosed with CMT2 or other complex hereditary peripheral neuropathy. After the gene analysis, the muscle respiratory chain enzyme activities showed deficiencies in complexes I (45%) and IV (43%), whereas complex II (the only complex that does not contain mtDNA-encoded proteins) was normal (105%; each activity was expressed as the citrate synthase ratio).
Regarding heterozygous splice acceptor site variant (c.664-1G > A) in intron 5, we detected two RT-PCR products comprising a wild-type band and a smaller band at 370 bp (Fig. 4a). Sanger sequencing of the RT-PCR product revealed 9 bp heterozygous deletion (Fig. 4b), leading to the in-frame deletion of three amino acids in exon 6 of SUCLA2 (p.L222_Q224del; Fig. 4c).

Discussion
Using WES, we performed further genetic analysis targeting 167 mitochondrial-related nuclear genes among a group of patients with clinically suspected with pure or complex hereditary peripheral neuropathy. In four patients, we identified four disease-causing genes, of which variants in PDHB, MTPAP, and SUCLA2 have been reported not as a cause of CMT but of other diseases associated with symptomatic or subclinical peripheral neuropathy [14][15][16]. Although recessive mutations in HADHB have been assumed to be associated with axonal CMT, no pathophysiological mechanism was elucidated [17,18]. We indicated the pathological significance of the PDHB and HADHB genes using Drosophila neuron-specific knockdown models in a previous study [19,20]. Although all four patients presented with various phenotypes, motor and sensory axonal neuropathy was the major clinical feature.
The pyruvate dehydrogenase alpha (PDHA) and PDHB encode pyruvate dehydrogenase, a component enzyme of the pyruvate dehydrogenase multienzyme complex in mitochondria [21]. Recessive PDHB mutation causes pyruvate dehydrogenase complex (PDC) deficiencies with severe clinical consequences that primarily affect the nervous system, such as developmental delay, seizures, central hypotonia, ataxia, peripheral neuropathy, microcephaly, congenital brain malformations, and degenerative changes, such as Leigh syndrome [14,21,22]. Given that cases of thiamine-sensitive PDC deficiency have been reported and a ketogenic diet may alleviate the symptoms [21], it is important to diagnose this disease. In 3-21% of patients, peripheral neuropathy has been found to be linked to PDC deficiencies [21], whereas progressive peripheral neuropathy in adult patients has been linked only to mutations in PDHA1 [22]. In Patient 1, in addition to peripheral neuropathy, central nervous system Genotypes of the variants are indicated at the bottom of the pedigree for each examined individual. Arrows and asterisks indicate the proband of each family and individual whose genome was applied for Sanger sequencing. del deletion, WT wild type involvement was suggested by multiple manifestations, comprising seizures, intellectual disability, and cerebral atrophy in brain MRI. The G294R variant is located at the transketolase coding region of the C-terminal domain of the PDHB protein (Fig. 1i) and is close to the previously reported mutations of PDC deficiencies [14].
MTPAP gene encodes the mitochondrial polyadenylation polymerase, involved in DNA maintenance and repair [23]. Recessive mutation in MTPAP would result in delayed DNA repair, elevation of the level of DNA double-strand breaks, reactive oxygen species, and cell death after irradiation, leading to spastic ataxia and optic atrophy or encephalopathy. A previous case report described progressive lower motor neuron involvement due to MTPAP mutation [15] but progressive peripheral neuropathy, especially sensory nerves, without pyramidal signs, as observed in family 2, has not previously been described.
Importantly, low levels of vitamin B 12 and folate, as found in Patient 2 and his brother, might also cause peripheral neuropathy. Folate and vitamin B 12 , especially methylcobalamin, are coenzymes for cytosolic methionine synthase involved in vital cellular processes, such as methylation and DNA synthesis [24,25]. Bi-allelic mutations of MTPAP would elevate both DNA double-strand breaks and cell death, whereas over-consumption of vitamin B 12 and folate for DNA synthesis could cause deficiencies of  (Fig. 1j); no mutation is reported in this domain. Because the p.Q511* variant is located in the last exon, and loss-of-function has not been a recognized mechanism in MTPAP-related disease, this variant was classified as "strong," as per the ACMG standards and guidelines. Further investigations are needed to clarify the role of vitamin B 12 and folate in the pathogenesis of MTPAP-related peripheral neuropathy.
The alpha and beta subunits of the mitochondrial trifunctional protein (MTP) are encoded by the HADHA and HADHB genes [26]. These subunits catalyze three steps in the beta-oxidation of fatty acids, including the long-chain 3-hydroxyacyl-CoA dehydrogenase step. Recessive HADHB mutation results in the dysfunction of the beta-oxidation of fatty acids, leading to MTP deficiency, characterized in a severe heterogeneous syndrome, such as cardiomyopathy, recurrent Leigh-like encephalopathy, hepatopathy, and neonatal or unexpected infant death [27]. In contrast, the milder form is characterized by later-onset progressive axonal peripheral neuropathy (approximately 50-80%) and myopathy with or without episodic myoglobinuria [17,26]. Phenylalanine 398 is located in the thiolase, C-terminal domain of HADHB protein (Fig. 1k), and is close to known mutations linked to the milder form [26]. After the genetic diagnosis, Patient 3 experienced recurrent episodes of brown urine after exercise and rhabdomyolysis, and his C 14 -OH level increased slightly; all these characteristic phenotypes were considered supportive evidence to his diagnosis. As described for Patient 3, peripheral neuropathy may be the main symptom in patients with HADHB mutations, or at least in certain stages of the disease course.
The SUCLA2 gene is involved in the citric acid cycle and mtDNA synthesis [28]. In the citric acid cycle, SUCLA2 catalyzes the reversible formation of succinate and ATP from succinyl-CoA and ADP. This protein also interacts with nucleoside diphosphate kinase, which is involved in mtDNA synthesis [28]. Recessive SUCLA2 mutation causes mtDNA depletion syndrome, which could cause mild methylmalonic aciduria, recurrent vomiting, hypotonia, dehydration, respiratory distress, neonatal encephalopathy, and progressive lethargy [16]. Peripheral neuropathy could also be observed, as described for Patient 4. We identified two heterozygous variants in SUCLA2 from Patient 4: frame shift (c.1300delG) and splice site (c.664-1G > A); the latter was found to produce a protein that is three amino acids smaller than normal (p.L222_Q224del). The c.1300delG variant, located close to the 3′-terminal of SUCLA2, is likely to produce a truncated protein (Fig. 1l). Taken together with the segregation analysis, we classified these bi-allelic variants as pathogenic/ probably pathogenic.
Peripheral nerves have long axons wrapped in myelin lamellae provided by Schwann cells and are highly dependent on energy metabolism, ATP synthesis might eventually be influenced by any abnormalities in energy production and  Table 2 Genetic pathophysiology, phenotype, inheritance pattern, and neuropathy type of the mitochondrial-related nuclear genes described in this report (upper  depletion of mitochondria, leading to peripheral neuropathy. Moreover, various mitochondrial dysfunctions, such as dysfunctions in mitochondrial energy production or PDC, or assembling mitochondrial respiratory chain complex have been reported to cause CMT (Table 2) [2,[29][30][31][32][33][34][35]. Pathological mechanisms of each gene remain unclear, because peripheral nerves have long axons wrapped in myelin lamellae provided by Schwann cells and are highly dependent on energy metabolism; however, ATP synthesis might eventually be influenced by any abnormalities in energy production and depletion of mitochondria, leading to peripheral neuropathy (Fig. 5). As observed in patients with bi-allelic variants in PDHB and SUCLA2, in mitochondrial disorders, multisystem manifestation representative of central nervous system involvement is a common feature suggestive of clinical diagnosis. Conversely, like MTPAP and HADHB, these genes could hardly result in central nervous system abnormalities; thus, more attention should be paid to clinical assessments, particularly for certain symptoms that transiently emerged in their disease course and naturally disappeared with age. Moreover, because of the clinical diversity of these patients, peripheral neuropathy has always been recognized as part of the mitochondrial disorder rather than of CMT. Heterogeneous phenotypes of both mitochondrial disorders and CMT make the clinical diagnosis of either difficult. However, the development of diagnostic genetics has facilitated diagnoses made on a genetic basis.
In this study of a large Japanese cohort of patients with clinically suspected pure or complex hereditary peripheral neuropathy, we identified novel likely pathogenic/pathogenic Fig. 5 Overview of the mechanisms underlying the main mitochondrial-related genes causing peripheral neuropathy and genes described in this report. Genes edged in purple are the main mitochondrial-related genes and those edged in red are the genes described in this report. PDHC comprises two alpha subunits (PDHA) and two beta subunits (PDHB). MTP comprises alpha subunits (HADHA) and beta subunits (HADHB). SUCL comprises an alpha subunit encoded by SUCLG1 and a beta subunit encoded by either SUCLA2 or SUCLG2. CoA coenzyme A, dCMP deoxycytosine monophosphate, dCTD deoxycytidine, dCTP deoxycytosine triphosphate, dNDPs deoxynucleoside diphosphates, mtDNA mitochondrial DNA, MTP mitochondrial trifunctional protein, NDPs nucleoside diphosphates, PDHC pyruvate dehydrogenase complex, SUCL succinyl-CoA ligase, TCA tricarboxylic acid, VLCAD very-long-chain acyl-CoA dehydrogenase. variants in four mitochondrial-related nuclear genes. Mitochondrial abnormalities should be considered as a differential diagnosis in cases of axonopathy with suggestive symptoms or other unexplainable multisystem manifestations. Considering the limited number of gene panels targeted in our study, the discovery of more mitochondrial-related nuclear genes leading to mitochondrial-related neuropathy is highly likely. Regarding treatment, early diagnosis would provide more effective and prompt therapy strategies and medicines for the improvement of mitochondrial function might one day be a common target for mitochondrial neuropathy.