Discussion
We report four independent affected individuals with an early-onset progressive axonal motor neuronopathy, resulting in severe weakness and respiratory impairment, in combination with cerebellar atrophy. All affected individuals harbor autosomal-recessive variants in EXOSC9, which encodes an exosomal protein. Individual 2:II-1 presented with congenital fractures and arthrogryposis at birth and subsequent symptom progression resulting in respiratory failure and death at 15 months of age. This individual carried the heterozygous null variant in combination with the missense variant, whereas the three other affected individuals carried this missense in homozygosity; this might explain the more severe clinical presentation in individual 2:II-1. Individual 1:II-1 and individual 4:II-1 showed milder phenotypes, starting with congenital esotropia and congenital nystagmus, respectively, and poor head control in a relatively normal neonatal course. Significant developmental delay, progressive muscle weakness, oculomotor dysfunction, and coordination difficulties became evident only after 6 months of age. In particular, the lower motor neuron symptoms subsequently progressed rapidly during the second year of life: these included limited spontaneous movement of the extremities, and individual 1:II-1 also required mechanical ventilation. Individual 3:II-1 had a more pronounced neonatal presentation with muscle hypotonia, weak cry, feeding difficulties, and early-onset seizures at the age of 5 months. The phenotype of individual 3:II-1 could be partially due to another, yet to be identified, recessive genetic disorder segregating in the family, given the parental consanguinity and the history of a sister who had severe spasticity and epilepsy and passed away at age 8 years. Unfortunately, DNA from the sister of individual 3:II-1 was not available for genetic testing.
The involvement of spinal motor neurons together with cerebellar atrophy, with or without pontine involvement, is a common feature in affected individuals with variants in different exosomal subunit (EXOSC3, EXOSC8, and EXOSC9) and belongs to the spectrum of PCH1-related disorders. Although these variants typically present with severe weakness precluding sitting in most cases, variants in EXOSC3 in particular are now recognized to also include a milder presentation. However, CNS hypomyelination has thus far been noted only in affected individuals with variants in EXOSC8. The presence of congenital symptoms in more severe exosomal protein defects suggests a developmental impairment; however, a progressive deterioration of motor skills in the first few months of life confirms that exosomal proteins are important not only for neuronal development but also for the survival of spinal motor and cerebellar neurons. Although the disease in our affected individuals fits into the PCH1 spectrum, the presence of congenital fractures in the severe case described here (individual 2:II-1) is reminiscent of congenital bone fractures with prenatal SMA caused by variants in subunits of the transcriptional coactivator complex, another disorder caused by abnormal RNA function.45 Although not all affected individuals present with obvious pontocerebellar hypoplasia at birth, all affected individuals show cerebellar atrophy and signs of motor neuronopathy with onset in early childhood, suggesting the very characteristic involvement of the cerebellum and spinal motor neurons. Similar to PCH2, which results from variants in genes encoding subunits of the tRNA-splicing endonuclease complex (TSEN),46 PCH1 has been also classified as PCH1A (VRK1 variants), PCH1B (EXOSC3 variants), and PCH1C (EXOSC8 variants). We propose to classify the disease caused by EXOSC9 variants as PCH1D. This would allow the characterization of the different phenotypes within each genotype.
In the four families reported here, we detected two different EXOSC9 variants (c.41T>C and c.481C>T). Interestingly, haplotype analysis revealed a common haplotype shared among all three affected individuals homozygous for the c.41T>C variant. Given the milder phenotype of affected individuals carrying this homozygous variant, we suspect that its loss of function is not as severe as in the case of the nonsense variant c.481C>T. The presence of two severe nonsense variants in one individual would most likely result in intrauterine lethality, which has also been proposed for other exosomal protein variants in EXOSC817 and EXOSC3.47 We noted that there has been no case of a complete loss of function in any of the exosomal components, thus reinforcing our hypothesis that such a defect would be cause intrauterine lethality.
We detected reduced levels of the protein EXOSC9 in two affected individuals, indicating a loss-of-function effect of the variants, as we have shown previously for EXOSC8 and RBM7 variants. Notably, the other subunits of the exosome (EXOSC3 and EXOSC8) were also reduced, suggesting that the exosome complex cannot be assembled or might be unstable if one of the subunits is primarily reduced. This notion was also supported by the detection of a lower total amount of the entire exosome complex by BN-PAGE in all cells from affected individuals with exosome variants. The individuals with EXOSC8 and EXOSC3 variants and severe phenotypes demonstrate the most pronounced reduction of the whole exosome and its subunits, whereas the reduction of the exosome was less pronounced in affected individuals with EXOSC9 (individual 1:II-1) and RBM7 variants. This pathophysiological concept would suggest that the disease manifestations and their severity are more the result of the degree of overall exosome dysfunction rather than of a dysfunction of individual exosome components. This would be consistent with the similar nature of the “exosomopathies” as a group.
We performed RNA-seq analysis in fibroblasts and skeletal muscle of affected individuals with EXOSC9 variants to investigate the effect of exosome impairment on the mRNA level. In fibroblasts, we detected significantly altered expression of genes involved in embryonic development of neurons (HOXB-A53, HOXB7, and HOXC8) and genes encoding extracellular matrix proteins and potassium channels, but we did not detect other neuron-specific gene alterations, probably because we studied fibroblasts and not neurons. RNA-seq in muscle showed altered gene expression of a high number of genes, suggesting that all tissues might be affected by the exosome defect. Expression of EXOSC9 and other genes encoding exosomal proteins was not different from control levels, suggesting that the downregulation of the exosome proteins is not due to reduced gene expression but is probably secondary to the instability of the holocomplex. Some of the genes contained AREs, suggesting that EXOSC9 variants affect the degradation of AU-rich transcripts, which is one of the best-known functions of the exosome in the cytosol. However, changes in gene expression of many non-AU-rich genes suggest a more widespread exosome function, potentially involving splicing and other forms of gene expression regulation. Further studies are needed to explore the mechanism of altered RNA metabolism in neurons.
We studied reduced exosc9 in an in vivo model by morpholino oligonucleotide knockdown and CRISPR/Cas9 mutagenesis in zebrafish embryos. Morpholino oligonucleotide knockdown is an established technique used for modeling diseases in zebrafish, although there has been recent criticism of the specificity of the morpholino oligonucleotides and suggestions of “off-target” effects.48 To address this potential problem, we complementarily used CRISPR/Cas9 mutagenesis to produce mosaic “crispant” zebrafish49 Knockdown and mutagenesis of exosc9 produced similar phenotypes in zebrafish, which suggests that both methods specifically target and reduce exosc9, although we could not find any antibody that successfully identified any of the exosome components in zebrafish to prove that there was indeed a reduction at the protein level.
The phenotype caused by the reduction of exosc9 in zebrafish is generally consistent with the phenotypic aspects (in particular, cerebellar defects and motor neuron pathology) seen in the four affected individuals. The phenotype of the exosc9 morphants and crispants was also similar to that observed in exosc3, exosc8, and rbm7 morphant zebrafish.16, 17, 19 Interestingly, other zebrafish models of cerebellar hypoplasia and atrophy caused by abnormal RNA processing have a phenotype similar to that of the exosc9 downregulated zebrafish. Morpholino oligonucleotide knockdown of Toe1 caused midbrain and hindbrain degeneration.50 TOE1 has roles in RNA degradation, and variants in TOE1 have been reported in affected individuals with PCH7. TSEN variants and the associated protein CLP1 have been associated with PCH in affected individuals.51, 52 Morpholino knockdown of these genes in zebrafish embryos also recapitulated the phenotype seen in affected individuals. Together, these studies highlight the consistency of zebrafish as a model of cerebellar hypoplasia and atrophy in particular in disorders of RNA processing.
Morpholino knockdown of exosc9 has previously been performed in Xenopus embryos.53 However, the Xenopus exosc9 morphants had defects in skin development but not the cerebellum. The discrepancy between zebrafish and Xenopus could be due to inherent differences in the models, or it could simply be that the cerebellum was not examined in the Xenopus morphants. Our results here show that zebrafish are a useful tool for generating rapid in vivo models of rare genetic diseases involving the development of the brain and spinal cord.
In summary, we have described four independent individuals who are affected by variants in EXOSC9 and who presented with motor axonopathy resembling SMA, cerebellar atrophy, and in one affected individual, multiple bone fractures. These EXOSC9-related phenotypes closely resemble the clinical spectrum of other exosomal defects, which could be referred to as “exosomopathies.” The clinically unique combination of a motor neuronopathy with cerebellar atrophy typical for the exosomopathies link the other two major clinical groups of the disorders of RNA processing, namely SMA without cerebellar involvement on the one end and the pontocerebellar hypoplasias without SMA on the other. Molecular studies indicate that the pathology is linked to a loss of function of the RNA processing by the exosome. Further studies on neuronal cells of affected individuals or zebrafish neurons might reveal neuron-specific alterations, which could be targeted in future interventions.