Infantile SOD1 deficiency syndrome caused by a homozygous SOD1 variant with absence of enzyme activity

Abstract

Pathogenic variants in SOD1, encoding superoxide dismutase 1, are responsible for about 20% of all familial amyotrophic lateral sclerosis cases, through a gain-of-function mechanism. Recently, two reports showed that a specific homozygous SOD1 loss-of-function variant is associated with an infantile progressive motor-neurological syndrome.

Exome sequencing followed by molecular studies, including cDNA analysis, SOD1 protein levels and enzymatic activity, and plasma neurofilament light chain levels, were undertaken in an infant with severe global developmental delay, axial hypotonia and limb spasticity.

We identified a homozygous 3-bp in-frame deletion in SOD1. cDNA analysis predicted the loss of a single valine residue from a tandem pair (p.Val119/Val120) in the wild-type protein, yet expression levels and splicing were preserved. Analysis of SOD1 activity and protein levels in erythrocyte lysates showed essentially no enzymatic activity and undetectable SOD1 protein in the child, whereas the parents had ∼50% protein expression and activity relative to controls. Neurofilament light chain levels in plasma were elevated, implying ongoing axonal injury and neurodegeneration.

Thus, we provide confirmatory evidence of a second biallelic variant in an infant with a severe neurological syndrome and suggest that the in-frame deletion causes instability and subsequent degeneration of SOD1. We highlight the importance of the valine residues at positions V119-120, and suggest possible implications for future therapeutics research.

See Farrimond and Talbot (https://doi.org/10.1093/brain/awac063) for a scientific commentary on this article.

Introduction

The superoxide dismutase 1 (SOD1) protein, coded by the SOD1 gene, is a highly conserved, ubiquitously expressed antioxidant enzyme. It is a copper- and zinc-binding protein and is one of three isozymes of the SOD family in humans that converts superoxide radicals into molecular oxygen and hydrogen peroxide.^1-3,SOD1 heterozygous variants account for about 20% of all familial amyotrophic lateral sclerosis cases.⁴ Disease caused by most variants associated with amyotrophic lateral sclerosis is inherited as a Mendelian dominant trait. However, the most prevalent SOD1 pathogenic variant p.Asp91Ala causes recessively inherited amyotrophic lateral sclerosis with normal enzymatic activity.^5,6 Extensive research indicates SOD1 pathogenic variants cause amyotrophic lateral sclerosis through a toxic gain-of-function mechanism, with variants leading to protein-aggregation and sometimes increased oxidative damage.^7–9

In 2019, a novel homozygous frameshift SOD1 variant was reported in two apparently unrelated families of Afghan origin, with infantile-onset progressive motor function loss, axial hypotonia and spastic tetraparesis.^10,11 Both groups reported the same frameshift variant, c.335dupG; p.Cys112Trpfs*11, suggesting a founder variant. Analysis of SOD1 enzymatic activity in erythrocytes and leukocytes indicated that affected children lacked enzyme function, while their heterozygous parents had approximately half the normal SOD1 activity. The authors suggested that biallelic SOD1 loss-of-function results in a novel phenotype with a recessive mode of inheritance, distinct from the typical adult-onset amyotrophic lateral sclerosis. They highlighted that the loss of SOD1 free-radical scavenging activity primarily affects the nervous system and the implications of this for therapeutic trials aiming at reducing SOD1 in amyotrophic lateral sclerosis via antisense oligonucleotides, shRNA, miRNA or other compounds, calling for a cautious approach.^10,11

Here, we describe a female infant with global developmental delay, axial hypotonia and limb spasticity, in whom exome sequencing revealed a homozygous in-frame 3-base pair (bp) deletion variant in SOD1. This represents an affirmatory report of a second biallelic variant with a strikingly similar manifestation to the previous reports.^10,11

Materials and methods

Study participants

The study was approved by the local institutional review board (0306–10-HMO). Parents provided signed consent for exome sequencing and research participation.

Exome sequencing

Genomic DNA was extracted from peripheral blood samples of the proband and both parents. Exonic sequences were enriched in the DNA sample using the IDT xGen Exome Research Panel V2.0 capture (Integrated DNA Technologies), and sequenced on a NovaSeq 6000 sequencing system (Illumina) as 100-bp paired-end runs. Data analysis including read alignment and variant calling was performed with DNAnexus software (Palo Alto, CA, USA) using default parameters, with the human genome assembly hg19/GRCh37 as reference. Variants were filtered out if they were off-target (intronic variants >8 bp from splice junction), synonymous (unless <4 bp from the splice site) or had minor allele frequency (MAF) >0.01 in the Genome Aggregation Database (gnomAD) or in our in-house exome database.

RNA extraction and cDNA preparation

RNA was isolated from fresh lymphocytes of the proband and both parents by TRIzol reagent extraction. cDNA was prepared from 1 μg RNA using the qScript cDNA Synthesis Kit (Quantabio).

Polymerase chain reaction with reverse transcriptase

The SOD1 transcript was amplified by a polymerase chain reaction with reverse transcriptase using HS Taq Mix Red (PCRBIO). Primers are provided in Supplementary Table 1.

Sanger sequencing

Sanger sequencing of cDNA amplicons was performed using BigDye™ Terminator v.1.1 Cycle Sequencing Kit (ThermoFisher).

Enzymatic activity analysis

The SOD activity was determined in erythrocyte lysates derived from the proband and both parents. The SOD activity was analysed in triplicate using the direct kinetic assay with $O2.$⁻ derived from KO₂, and was corrected for haemoglobin concentration.^12,13

Western blotting

For western blot analysis, haemolysates were prepared on two different occasions and were stored at −80°C until use. Samples were analysed for haemoglobin (Hb) content and diluted in phosphate-buffered saline and sample buffer with beta-mercaptoethanol to a final concentration of 0.05 mg/ml haemoglobin. Then 10 μl of each sample preparation were loaded on Any kD Stain free Criterion TGX precast gels (Bio-Rad Laboratories). Gels were activated and protein content imaged using a criterion imaging apparatus, before blotting on nitrocellulose filters (Bio-Rad). Primary rabbit anti-human SOD1 antibodies raised against peptides corresponding to amino acids 25–40 (1 μg/ml) or amino acids 58–73 (1 μg/ml) were incubated overnight at 4°C. Horseradish peroxidase (HRP)-conjugated secondary anti-rabbit IgG (1:10 000, Dako) was incubated for 1 h at room temperature. ECL Select reagent (GE Healthcare Biosciences) was used to detect the signal. Images were acquired using a Chemidoc apparatus and analysed using ImageLab software (Bio-Rad Laboratories).

Neurofilament quantification

For the quantification of plasma neurofilaments, the ELLA microfluidic system (bio-techne) was used and measurements were performed according to the manufacturer’s instructions.¹⁴

3D modelling of protein structure

3D structure prediction of human SOD1 p.Val120del was performed by using the Tr-Rosetta software package.¹⁵

Data availability

The ClinVar accession number for the DNA variant data is SCV001774778.

Results

Clinical description

The proband (III-3 in Fig. 1A) was the first-born female child of consanguineous, first cousin parents of Arab Muslim origin. She was referred to the genetic department due to hypotonia and developmental delay. Pregnancy was reportedly uncomplicated until the eighth month of pregnancy, when foetal ultrasound suggested macrocephaly. The baby was delivered at term by C-section due to foetal distress, with a birthweight of 3500 g (+0.57 z). Apgar scores were 9 and 10 at 1 and 5 min, respectively. Developmental milestones were delayed; at 3 months of age, she did not support her head, was hypotonic and had decreased motor movements. She smiled at 4 months. At 9 months, she did not roll over and had a weak grasp. Parents noted regression around 10 months of age—she no longer reached for objects, and developed spasticity of the limbs. Formal neurodevelopmental evaluation at age 11 months yielded the following developmental quotient (DQ) scores: cognition DQ = 39, receptive speech DQ = 48, expressive speech DQ = 45, fine motor DQ = 42, gross motor DQ = 18. At her most recent evaluation at 25 months, there was further decline with cognition DQ = 16, receptive speech DQ = 17, expressive speech DQ = 16, fine motor DQ = 8 and gross motor DQ = 8.

At 17 months, she could not reach for objects, although she had reportedly been able to do so previously. She could not support her head nor roll over. She could smile, track and could say two words. She was fed by mouth, including solid foods. Growth parameters were as follows: weight 11 kg (+0.75 z, 54th percenttile), length 81.5 cm (+0.64 z, 70th percenttile), head circumference 51 cm (+3.58 z, >97th percentile). She was alert, had a social smile yet diminished facial expressions, and was macrocephalic with an open anterior fontanelle. Notably, her father’s head circumference was 61 cm, above the 97th percentile, possibly indicating familial macrocephaly. No dysmorphic facial features were noted, but she had a high arched palate, bilateral fifth finger clinodactyly, partial toe syndactyly of the second and third toes, and a single hyperpigmented macule. Exam was further positive for tongue fasciculations, axial hypotonia with limb spasticity (more pronounced in the lower limbs), ankle clonus and brisk patellar deep tendon reflexes. Ophthalmologic exam revealed suspected temporal pallor of the optic disc, with an otherwise normal retina and anterior chamber. Brain MRI (Fig. 1B) suggested benign enlargement of subarachnoid spaces (more in frontal and temporal regions), while spine MRI was normal. Plasma CPK levels were mostly normal (highest at 270 U/l, normal range 34–145 U/l), with mildly elevated AST and ALT levels. Plasma lactate and ammonia were normal, as was an abdominal ultrasound. Chromosomal microarray was normal. The mother was tested for spinal muscular atrophy carrier status with normal result. A primary genetic neurological disorder was suspected, and exome sequencing was undertaken.

Besides a family history for a cousin with oculomotor apraxia and nephronophthisis, in whom a homozygous NPHP4 intragenic deletion was identified, the parents reported no predisposition for a wasting neuromuscular disease. Supplementary Table 2 provides a comparison of the clinical features of previously reported patients and the current report.

Exome sequencing reveals a homozygous in-frame deletion in SOD1

Trio exome (proband and parents) was pursued, and yielded 57 million reads for the proband, with a mean coverage of 85.2 times and >96% over 20 times coverage. Following variant filtration, 14 rare homozygous variants were identified (Supplementary Table 3). Among these was a variant in DYNC2H1 [NM_001080463.2:c.11284A>G; p.(Met3762Val)] with conflicting evidence of pathogenicity in regarding its role in short-rib thoracic dysplasia (MIM 613091).¹⁶ The proband’s phenotype and chest X-rays were not classical for this diagnosis. No de novo variants were identified in coding regions, and no copy number variants were reported from read depth analysis. The prime variant of interest was a homozygous variant in SOD1: chr21:g.33039683[hg19]; NM_000454.5;c.357_357+2delGGT. This deletion affected three nucleotides (GGT) from amongst a tandem repeat (GGTGGT), including the splice junction (Fig. 2A and B). Thus, the variant was hypothesized to either remove a single amino acid or otherwise to interfere with the exon junction and to result in aberrant splicing.

To differentiate between these conditions, we isolated RNA from peripheral blood of the proband and both parents, generated cDNA and amplified exons 1–5. Sanger sequencing revealed no exon skipping, and showed that the patient was homozygous for a 3-bp deletion observed in the exome and that both parents were heterozygous. The deletion removed a single highly conserved hydrophobic valine residue, from either position 119 or 120 (Fig. 2A and B and Supplementary Fig. 1). No other variants nor missplicing were identified. There was no apparent downregulation of SOD1 expression in the blood of the patient, as evaluated by polymerase chain reaction with reverse transcriptase (Supplementary Fig. 2).

3D protein modelling suggests impaired copper binding

To predict the effect of a single amino acid deletion at p.Val119 or p.Val120, we performed in silico 3D protein modelling. The copper-binding site structure of one subunit of SOD1-wt (PDB code: 4A7V)¹⁷ was compared with the SOD1-V119del/V120del model. Absence of a single valine residue affected the binding pocket structure, and was predicted to prevent the binding of a copper ion (Fig. 2C).

SOD1 protein and activity are essentially absent in the proband

SOD activity, as measured in erythrocytes where SOD1 is the only SOD isoform, was 0.49 units/mg haemoglobin (U/mg Hb) in the affected child, and 27.9 and 30.6 U/mg Hb in the parents (normal values, 55.0 ± 6.5 U/mg Hb, age independent).¹³ Thus, the activities found in the heterozygous parents were reduced to approximately half of the controls while the affected child essentially lacked SOD1 enzymatic activity (Fig. 3A).

To determine whether the decreased SOD1 activity was attributed to a destabilization of the mutant protein levels and/or to altered function of a stable protein, we analysed the level of SOD1 protein in the proband, parents and unaffected control in haemolysates from erythrocytes. Protein levels were normalized to haemoglobin content. Antibodies to two different regions of SOD1 indicated that SOD1 expression was undetectable or below the detection limit in the proband, also after over-exposure. Parents expressed approximately half of the protein levels of the wild-type control (Fig. 3B and Supplementary Fig. 3), suggesting that the mutant protein is unstable and efficiently degraded.

Taken together, the cDNA sequencing, protein levels and the enzymatic activity analysis suggest that the sole 3-bp deletion in the SOD1 transcript is likely the primary cause for the observed phenotype in the affected child. While it does not downregulate mRNA expression nor alter splicing, it results in deletion of the highly conserved Val119 or Val120, presumably leading to destabilization of the mutant protein and therefore lack of SOD1 activity.

Neurofilament light chain levels are elevated in the proband

Neurofilament light chains (NfL) are a constituent of the cytoskeleton and are highly abundant in axons. Therefore, elevated plasma NfL levels have been used as a bio-marker of axonal damage and neurodegeneration.¹⁸ At 23 months of age, the proband had a serum NfL level of 74 pg/ml, which is a 3-fold increase above the upper limit of the range in healthy infants and children up to age 4 years.¹⁹ This suggested ongoing axonal degeneration in the patient. Both parents had serum NfL levels similar to healthy adults (6 and 3 pg/ml, respectively) (Fig. 3C).

Discussion

In this study, we describe an infant with a homozygous 3-bp deletion in SOD1 and essentially absent SOD1 free-radical scavenging enzymatic activity, resulting in a confirmatory report of homozygous LOF variants underlying a severe infantile neurological disease with apparently little affection of non-nervous tissue. The region of the 3ʹ end of exon 4 and the 5ʹ of exon 5 is known to contain many of the pathogenic variants in amyotrophic lateral sclerosis,²⁰ making it a hotspot within the SOD1 gene. p.Val120 has been computationally predicted to be involved in allosteric regulation of the catalytic site activity of SOD1.²¹ This was further supported by the 3D modelling, where the valine deletion seemed to disrupt the ligation of Cu²⁺ to p.His121 essential for enzymatic activity.²¹ The structural changes are expected to destabilize the mutant SOD1. Protein levels were dramatically reduced in patient erythrocytes leading to essentially no SOD1 activity. Similarly, the mRNA in the patient reported by Andersen et al.¹⁰ was readily detected in leukocytes, yet the protein could only be detected after proteasomal inhibition. The c.335dupG variant causes a frameshift at p.Cys112 and a neopeptide of 11 amino acids before truncation, also affecting the p.Val119-Val120-His121 residues. The infant described here had macrocephaly, which seemed to be familial rather than a phenotypic expansion of the SOD1-associated phenotype. No molecular diagnosis was identified as contributing to the macrocephaly on exome sequencing.

Notably, there are no reported cases of amyotrophic lateral sclerosis in the extended family of the proband (which presumably includes older heterozygous carriers of the variant). This resonates with the previous findings,^10,11 and lends support to the hypothesis that SOD1-associated amyotrophic lateral sclerosis is caused by a gain-of-function rather than loss-of-function mechanism.⁵ Additional studies in heterozygous carriers are called for to assess whether older heterozygous carriers are indeed free of neurological symptoms, or whether there are subclinical abnormalities in motor function or reduce disease penetrance for amyotrophic lateral sclerosis. Notably, NfL levels observed in both parents were low, suggesting that there was no current ongoing axonal damage.

The growing evidence for SOD1 toxicity through a gain-of-function as a major factor in amyotrophic lateral sclerosis pathophysiology gave rise to the concept of therapy through downregulation of SOD1.^8,22–25 Our findings along with the previously reported SOD1 truncating variant,^10,11 have implications regarding SOD1 reduction therapy. While downregulation of SOD1 can be effective, it should be carefully monitored so that the protein does not reach a critically-low expression level. Importantly, antisense oligonucleotides and siRNAs currently studied are non-specific and target both mutant and wild-type SOD1, and will hence result in a substantial reduction in superoxide free-radical scavenging in the cytoplasm. Further research is needed to assess the optimal levels of SOD1 reduction for amyotrophic lateral sclerosis patients.

As with the previous loss-of-function variant found in SOD1,^10,11 this case raises the question whether antioxidants, which have been shown to protect neurons by reducing reactive oxygen species,²⁶ might have a beneficial effect in affected individuals. External supplementation of antioxidant compounds can be beneficial; ascorbic acid treatment has been shown to improve the phenotype of SOD1-deficient mice.²⁷ Moreover, several SOD mimics—most of which are metal complexes, such as metalloporphyrins—have been shown to reduce oxidative stress.²⁸

There is an increasing number of genes known to cause different phenotypes, depending on the zygosity of the variant. Namely, a heterozygous (monoallelic) variant can increase the risk of an adult-onset disease while a homozygous (biallelic) variant in the same gene is causative of a more severe childhood disease. Such examples include BRCA2,^29,ATM,^30,ABC4,^31,GBA³² and others. Joined with previous reports of a SOD1 homozygous variant causing a severe phenotype at an early age,^10,11 we suggest SOD1 as a gene with similar duality.

In conclusion, our data confirm that homozygous SOD1 variants resulting in a loss of enzyme function cause a severe motor-neurological syndrome of infantile onset distinct from typical juvenile amyotrophic lateral sclerosis, highlight the importance of the p.Val119-120 residues, and call for antioxidant studies in affected individuals.

Acknowledgements

The authors wish to thank the family for their participation in this study.

Funding

This work was partly supported by the fund ‘Innovative Medical Research’ of the University of Münster Medical School (to J.H.P.), the Knut and Alice Wallenberg Foundation (grant no. 2020.0232 to P.M.A.), the JPND project Genfi-Prox and the ALS Association.

Competing interests

The authors report no competing interests.

Supplementary material

Supplementary material is available at Brain online.

References