RTN2 deficiency results in an autosomal recessive distal motor neuropathy with lower limb spasticity

Abstract Heterozygous RTN2 variants have been previously identified in a limited cohort of families affected by autosomal dominant spastic paraplegia (SPG12-OMIM:604805) with a variable age of onset. Nevertheless, the definitive validity of SPG12 remains to be confidently confirmed due to the scarcity of supporting evidence. In this study, we identified and validated seven novel or ultra-rare homozygous loss-of-function RTN2 variants in 14 individuals from seven consanguineous families with distal hereditary motor neuropathy (dHMN) using exome, genome and Sanger sequencing coupled with deep-phenotyping. All affected individuals (seven males and seven females, aged 9–50 years) exhibited weakness in the distal upper and lower limbs, lower limb spasticity and hyperreflexia, with onset in the first decade of life. Nerve conduction studies revealed axonal motor neuropathy with neurogenic changes in the electromyography. Despite a slowly progressive disease course, all patients remained ambulatory over a mean disease duration of 19.71 ± 13.70 years. Characterization of Caenorhabditis elegans RTN2 homologous loss-of-function variants demonstrated morphological and behavioural differences compared with the parental strain. Treatment of the mutant with an endoplasmic/sarcoplasmic reticulum Ca2+ reuptake inhibitor (2,5-di-tert-butylhydroquinone) rescued key phenotypic differences, suggesting a potential therapeutic benefit for RTN2-disorder. Despite RTN2 being an endoplasmic reticulum (ER)-resident membrane shaping protein, our analysis of patient fibroblast cells did not find significant alterations in ER structure or the response to ER stress. Our findings delineate a distinct form of autosomal recessive dHMN with pyramidal features associated with RTN2 deficiency. This phenotype shares similarities with SIGMAR1-related dHMN and Silver-like syndromes, providing valuable insights into the clinical spectrum and potential therapeutic strategies for RTN2-related dHMN.


Introduction
RTN2 (HGNC:10468) encodes RTN2, an endoplasmic reticulum (ER)-shaping protein localized to the ER.In maize, it binds the autophagy protein ATG8a upon ER stress. 1 However, its role in ER structure is still unclear.Heterozygous RTN2 variants have previously been associated with progressive hereditary spastic paraplegia (HSP) (SPG12-MIM#604805). 2,3However, this association is yet to be confidently established.
To date, the empirical evidence includes information concerning two extended European families, encompassing a minimum of 23 individuals harbouring a segregating ultra-rare founder frameshift variant.Furthermore, there are reports of seven individuals from seven small families with four missense variants, one nonsense variant, one complete gene deletion and two frameshift variants.The reported phenotype is predominantly pure HSP with a varying age of onset ranging from 5 to 50 years old. 2,3hile the co-segregation of the founder RTN2 frameshift variant in two independent extended families strongly supports its association with the HSP phenotype on chromosome 19q13, it is important to clarify that this evidence supports the linkage rather than directly proving the pathogenicity of the variant itself.Furthermore, all the reported variants are found in very low frequency in large human genetic variant databases such as gnomAD, UK-Biobank and the UK's 100 000 Genomes Project.This may be attributed to incomplete or low penetrance of the variants or their lack of pathogenicity.
Here, we identified 14 individuals from seven independent consanguineous families with homozygous loss-of-function (LoF) RTN2 variants presenting with slowly progressive distal motor neuropathy with spasticity and hyperreflexia affecting the upper and lower limbs.Analyses of fibroblasts from a patient homozygous for the RTN2 p.Gly27Ter did not reveal obvious changes in ER structure or the response to ER stress.By generating a Caenorhabditis elegans disease model containing LoF of the worm RTN2 orthologue, ret-1, we identified key behavioural and phenotypic differences in the mutant, which are amenable to drug screening efforts, and confirmed that the inhibition of sarcoplasmic reticulum (SR)/ER Ca 2+ reuptake may have some therapeutic benefit in the treatment of diseases involving RTN2 variants.

Patient ascertainment, clinical assessments and genetic investigation
Using exome, genome, and Sanger sequencing, along with extensive data sharing with global clinical and research genetics laboratories, we identified 14 individuals (seven males and seven females) from seven consanguineous families harbouring homozygous LoF RTN2 variants (Fig. 1A).The age range of the affected individuals was 9-35 years for males and 11-50 years for females.Comprehensive data, including detailed clinical descriptions, electrodiagnostic results, brain and muscle imaging, as well as photographic and video documentation of the affected individuals, were collected.All participating families provided written informed consent.
For immunostaining, cells were seeded on 14 mm coverslips and fixed in 4% paraformaldehyde (PFA) for 15 min.Cells were washed three times, permeabilized with 0.25% Triton X-100 for 10 min and blocked using 5% normal goat serum (BIOZOL, S-1000) for 1 h at room temperature.Coverslips were incubated overnight with primary antibodies: sheep anti-CLIMP63 (1:1000, R&D Systems, AF-7355) and rabbit anti-NOGO (1:1000, Abcam, ab74085) diluted in blocking solution at 4°C.The next day the coverslips were washed three times and incubated with secondary antibodies conjugated with a fluorophore for 1 h at room temperature.After For ER stress induction and the cell viability count, cells were seeded in six-well plates and cultured to 70-80% confluency.Cells were washed with PBS and incubated with new medium supplemented with 5 µg/ml tunicamycin (Santa Cruz) without or in combination with 1 µM MG132 (Calbiochem, 474787-10MG) to inhibit the proteasome.After 48 h, the culture medium was removed, and cells were washed with PBS and trypsinized.All cells were pooled, centrifuged at 800 rpm for 5 min (Heraeus Sepatech Megafuge 2.0R) and resuspended in fresh medium.Cell viability was measured by trypan blue exclusion with an automatic counting device (Bio-Rad TC20 Automatic Cell Counter).For western blot analysis, cells from 10 cm plates were washed twice on ice in ice-cold PBS and lysed directly by adding radioimmunoprecipitation assay (RIPA) lysis buffer.Samples were prepared and subjected to sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and western blotting as described previously. 4

C. elegans mutant generation and drug response analysis
A homozygous C. elegans mutant strain, ret-(syb4955), containing LoF deletion (17.8 kbp) of the entire coding region of the worm RTN2 orthologue, ret-1, was generated by SunyBiotech (Fuzhou, Fujian, China) using clustered regularly interspaced short palindromic repeats (CRISPR) in an N2 background.Worms were cultured on nematode growth medium at 20°C with Escherichia coli strain OP50 following standard methods. 5Phenotypic differences in baseline morphology, posture and locomotion of ret-1(syb4955) and differences in behaviour as the result of treatment with a panel of bioactive molecules were identified using high-resolution worm tracking coupled with automated behavioural phenotyping.Age synchronized populations of young adult worms were reared and imaged using methods previously described in detail. 6For drug experiments, worms were exposed to compounds for 4 h prior to imaging (detailed protocol: dx.doi.org/10.17504/protocols.io.dm6gpj19 dgzp/v1).Videos were acquired and behavioural features extracted using a high-resolution multi-camera array tracker and Tierspy, following methods previously described in detail. 7,8

Ethics
The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board at University College London (#310045/1571740/37/598).All participating families provided written informed consent.

Clinical phenotyping and characterization
Affected individuals were from consanguineous Iranian (two families), Indian (two families), Egyptian, Tunisian and Pakistani families.One affected sibling of the affected individual from Family 3 did not give consent and was therefore excluded from the detailed analysis (Table 1 and Fig. 2).
Consistent features included lower limb spasticity, hyperreflexia, spastic gait and weakness of both the distal upper and lower limbs (lower > upper) with an early age of onset of between 1 and 6 years.Typically, affected individuals presented with early gait disturbances (tiptoe walking, frequent falls).Foot deformities were common and included pes cavus, hammertoes or pes planus.Atrophy of distal upper and lower limb muscles, along with weakness, prominent in finger extensors, was seen in nearly all individuals (Fig. 2B-P).Sensory deficits and cranial nerve involvement were not reported in any cases.Mild motor delay was reported in three individuals and urinary urgency in one.
Nerve conduction studies confirmed axonal motor neuropathy with neurogenic changes on electromyography.In two individuals, sensory nerve conduction in the lower limbs was also reported to be abnormal.Muscle MRI in two affected siblings of Family 1 performed at the ages of 17 and 27 years showed muscle atrophy, inflammation and fatty infiltration, particularly of the calf muscles (Fig. 2A).However, hand and thigh muscle MRI images were normal in two affected siblings from Family 6 at the ages of 9 and 11 years.Brain MRI images from Families 1 and 2 were reported to be normal, while one individual (Individual 13) from Family 7 showed some non-specific neuroimaging findings.
All participants had a slowly progressive disease course and remained ambulatory at the time of the investigation; the oldest participant is currently 50 years of age.None of the affected individuals died prematurely or had any vision, hearing, swallowing or respiratory problems.
While a comprehensive neurological examination was not conducted, it is noteworthy that all parents and carrier siblings exhibited no observable neurological symptoms.

Genetic findings
Exome and genome sequencing revealed the following homozygous LoF variants in RTN2 (NM_005619.Gly27Ter), in Family 6; and a frameshift variant, c.178dupC p.(Arg60ProfsTer11), in Family 7. Sanger sequencing segregation analysis demonstrated the complete co-segregation of all variants with the affected status within their respective families (Fig. 1A  and B).Notably, these variants were located within sizable regions of homozygosity.Furthermore, their presence was notably absent or occurred at extremely low allele frequencies across diverse genetic variant frequency databases, including but not limited to gnomAD (versions V2, V3 and V4), UK Biobank, Queen Square Genomics, Centogene and the 100 000 Genome Project.The aggregate number of alleles assessed across these databases exceeded 2 000 000 (Supplementary Table 1).

Functional characterization
Because of the prominent role of ER shaping proteins in ER structure, we stained fibroblasts from Patient XYZ, who was homozygous for the RTN2 p.G27Ter, and control fibroblasts for the ER sheet marker CLIMP63 and the ER tubule marker RTN4 (NOGO) 9 (Supplementary Fig. 2A-D).Surprisingly, both the CLIMP63-and the NOGO-labelled relative cell areas did not differ between the patient and control cells.In agreement, the abundance of CLIMP63 and NOGO did not differ in the protein lysates of patient and control fibroblasts (Supplementary Fig. 2E and F).Moreover, cell viability did not differ in steady state or upon induction of ER stress with tunicamycin (Supplementary Fig. 2G-I), suggesting that major ER functions remained intact without RTN2.
Deletion of the C. elegans RTN2 orthologue (ret-1) caused no developmental delay in worms (data not shown).Comparison of the ret-1 LoF mutant, ret-1(syb4955), and the parental strain (N2) revealed significant differences in 3345 morphological, postural and locomotive behavioural features (of 8829 in total) extracted from videos of worms crawling across an agar surface (Fig. 3).Key phenotypic behavioural differences between the strains were as follows: ret-1(syb4955) worms were longer, had decreased angular velocity of the head, increased angular velocity of the midbody and increased curvature across the body compared with N2.Although there was no significant difference in the overall baseline speed of ret-  primarily the head and midbody, exhibited increased speed in ret-1 LoF worms compared with N2.
We performed a phenotypic comparison of worms treated with a panel of 15 bioactive molecules known to be involved in intracellular calcium signalling pathways (Fig. 3).Hierarchical clustering the behavioural fingerprints of ret-1(syb4955) and N2 treated with the various compounds (all at a concentration of 100 µM) across the entire feature set revealed five distinct phenotypic clusters (Fig. 3A).There was a defined separation of the parental/mutant strains in phenomic space, although N2 treated with leflunomide or ret-1(syb4955) treated with cyclopiazonic acid (CPA) clustered as outliers.Treatment with 2-aminoethyl diphenylborinate (2-APB), CDN1163, egtazic acid (EGTA) or nicotinic acid adenine dinucleotide phosphate (NAADP) caused little change in the overall behaviour of wild-type/mutant worms.However, treatment with lansoprazole, leflunomide, roflumilast, teriflunomide, tolvaptan or istaroxime led to large behavioural differences, pushing N2 towards the mutant and ret-1(syb4955) further away from the parental strain in phenospace.
Importantly, we found that treating the ret-1 LoF mutant with 2,5-di-tert-butylhydroquinone (DTBHQ), an ER/SR Ca 2+ reuptake inhibitor, caused the strain to cluster closer to N2 (Fig. 3A), resulting in a reduced number of significant behavioural differences between the mutant and N2 (1072 out of 8289).This finding was further supported by the top principal components extracted from the behavioural feature set (Fig. 3B), where there was clear movement of the ret-1 LoF mutant towards untreated N2 with respect to both principal components.Furthermore, we note that treatment with DTHBQ rescued 3/5 of the key phenotypic differences observed between N2 and ret-1(syb4955) (Fig. 3C).We noted that treatment of N2 with DTHBQ caused significant attenuation of the photophobic escape response (defined as an increase in speed following exposure of worms to bursts of high intensity blue light), whereas there was no significant difference in the photophobic escape response of untreated N2/ret-1(syb4955) or the mutant strain treated with DTHBQ (Fig. 3D).We also noted that the ret-1 LoF mutant demonstrated resistance to severe attenuation of the photophobic escape response upon treatment with 2-APB [an inositol trisphosphate (IP3) receptor and transient receptor potential (TRP) channel inhibitor; Supplementary Fig. 1].Overall, these findings demonstrated clear movement of the disease model strain towards the healthy parental strain in overall phenospace upon treatment with DTHBQ.

Discussion
][12][13] This spectrum is proposed to exhibit a continuum rather than discrete entities, demonstrating both clinical and genetic overlap with axonal Charcot-Marie-Tooth disease (CMT2), juvenile amyotrophic lateral sclerosis, HSP, spinocerebellar ataxia and, in certain instances, distal myopathy.1][12] Monoallelic pathogenic variants in BSCL2 (HGNC:15832) encoding an ER membrane protein lead to a wide spectrum of highly variable neurological disorders ranging from pure dHMN type V without spasticity and CMT2 to Silver syndrome (SPG17-MIM#270685) and 'juvenile amyotrophic lateral sclerosis-mimics'. 14onoallelic pathogenic SETX (HGNC:445; MIM#608465) variants can also manifest as dHMN accompanied with pyramidal signs. 157][18] REEP1 (HGNC:25786) variants are found in patients who occasionally present with features overlapping with dHMN-V.In these cases, early and selective involvement of the thenar and dorsalis interosseus I muscle can be seen. 19pasticity can also rarely be observed in dominant dHMN caused by pathogenic variants of HSPB1 (HGNC:5246). 202][23] The phenotype associated with SIGMAR1, and one of the most frequent HSP subtypes, SPG11 (HGNC:11226), can also mimic juvenile amyotrophic lateral sclerosis and might be misdiagnosed as such. 21,24urthermore, biallelic COQ7 (HGNC:2244) variants were also recently described to cause dHMN with and without pyramidal signs. 25n this report, we have delineated a new subtype of recessive dHMN with pyramidal features associated with RTN2 deficiency.The distinct and unusual clinical feature of the novel RTN2disorder is finger extensor weakness, mimicking the phenotype of SIGMAR1-disorder.Furthermore, the phenotypic presentation mimics Silver-like diseases.However, in contrast to patients with biallelic variants in SIGMAR1, those with biallelic RTN2 pathogenic variants have a consistent spastic involvement of the legs in early childhood.Furthermore, unlike Silver-like syndromes, all affected individuals (apart from one) showed a predominantly distal lower limb weakness with early foot deformities and gait abnormalities at disease onset instead of early upper limb involvement.Interestingly, none of the heterozygous family members had any symptoms related to spasticity or neuropathy, which might be explained by incomplete or very low penetrance or a controversial role of the variants in the monoallelic state.
The muscle MRI findings in four individuals reported here, suggested an imaging phenotype consistent with length-dependent polyneuropathies in older individuals, specifically greater lower than upper limb involvement and more severe involvement distally.In the thighs, the hamstrings appeared to be involved first, while in the lower legs, the medial head of gastrocnemius and extensor digitorum longus were relatively spared.The imaging changes progressed with age and disease stage, with normal findings in the affected individuals aged 9 and 11 years from Family 6 and relatively mild changes and more severe changes in the individuals aged 17 years and 27 years from Family 1, respectively.However, confirmation of any phenotype/genotype correlation will require the analysis of MRI images from more patients.
The exact pathogenicity of RTN2-related disorder is poorly understood.RTN2 is known to be localized to the ER and involved in the dynamic remodelling of the ER. 1,4In addition, RTN2 interacts with FAM134B, a protein mediating ER-phagy.7][28] Our preliminary analysis, however, did not unveil major alterations to the ER structure or the response to ER stress as previously observed with FAM134B LoF. 4,27he ER plays critical roles in several cellular processes, including calcium homeostasis, and several monogenic neurodegenerative disorders, particularly those affecting upper or lower motor neurons with a variable degree of sensory involvement, have been associated with pathogenic variants in genes including ATL1, ATL3 (HGNC:24526), REEP1, REEP2 (HGNC:17975), SPAST, RTN2, ARL6IP1 (HGNC:697) and LNPK (HGNC:21610), which encode ER-resident membrane-shaping proteins characterized by reticulon homology domains. 4,29More recently, deletion of the Drosophila RTN2 orthologue was shown to alter ER organization and the function of presynaptic terminals significantly. 30Despite little change in resting Ca 2+ storage capacity, this variant led to major reductions in activity-evoked Ca 2+ fluxes in the cytosol, ER lumen and mitochondria.Interestingly, cytosolic and mitochondrial Ca 2+ responses in transheterozygous larvae were similar to those in larvae harbouring the wild-type RTN2 orthologue, further suggesting a pathogenic role for total loss of the protein.
Here, we generated a C. elegans RTN2 orthologous (ret-1) LoF mutant and characterized morphological and behavioural differences, which could be rescued via inhibition of ER/SR Ca 2+ reuptake.These data support findings from Drosophila functional studies that suggest changes in Ca 2+ signalling pathways contribute towards HSP pathology, 30 and as such we identified a potential benefit of inhibiting Ca 2+ reuptake in the ER/SR for the treatment of RTN2-related disorders.Moreover, characterization of the behavioural differences of C. elegans ret-1 LoF mutants means they are well suited for future high-throughput compound screens to further elucidate the molecular underpinnings of this rare neurological disorder and  following stimulation of the worms with three 10 s bursts of high-intensity blue light 100 s apart (denoted by blue shaded regions on the line plot), whereas the ret-1 LoF mutant showed resistance to the effects of DTHBQ.The mean peak speed (box plot, right) was calculated as an average of the maximum detected speed across the three independent pulses of blue light.Individual points marked on all box plots are average values from multiple worms (n = 3) in a single well.The different point colours indicate data from independent experimental days.P-vales were calculated using block permutation t-tests (n = 10 000 permutations).Permutations were shuffled randomly within, but not between, the independent days of image acquisition in order to control for day-to-day variation in the experiments and corrected for multiple comparisons using the Benjamini-Hochberg procedure to control the false discovery rate at 5%.P > 0.05 was considered not statistically significant (ns).identify additional candidate targets for treatment.Our findings delineate a new form of autosomal recessive dHMN with pyramidal features associated with RTN2 deficiency.

Figure 1
Figure 1 The illustration depicts family pedigrees and the RTN2 gene, highlighting the biallelic and monoallelic variants identified in this study and previous studies respectively.(A) Pedigrees of 14 individuals from seven consanguineous families with segregating homozygous RTN2 variants are shown.(B) The localization of the presented variants across the gene is displayed.With homozygous loss-of-function variant (black) identified in this study shown above the gene schematic, and previously reported heterozygous variants (grey) shown below the gene.

Figure 2
Figure 2 Images displaying distal limbs and muscle MRI from families with RTN2-related distal hereditary motor neuropathy.Muscle weakness and wasting of the distal lower > upper limbs in affected individuals from Families 1 (A-F and M-R), 2 (G) and 3 (H-L).Axial fat T1-weighted images of the lower limbs in a 17-year-old Iranian boy [A(i and ii)] and his 27-year-old sister from Family 1 [A(iii and iv)].The thigh muscles of the boy are normal [A(i)].There is fatty infiltration of his leg muscles with relative sparing of the medial head of gastrocnemius and extensor digitorum longus [A(ii)].Changes are more severe in his older sister with generalized loss of muscle bulk and fatty infiltration of the hamstrings [A(iii)].There is more extensive change in her legs, with relative sparing of the medial head of the gastrocnemius, tibialis anterior, extensor digitorum longus and peroneus longus [A(iv)].

Figure 3
Figure 3 Loss-of-function ret-1 (RTN2 homologue) Caenorhabditis elegans phenotyping and drug screen data.(A) Hierarchical clustering of behavioural fingerprints of untreated wild-type (N2) and C. elegans RTN2 orthologue, ret-1, loss-of-function (LoF) strains, alongside treatment with the panel of 15 bioactive compounds.The heatmap shows the entire set of 8289 behavioural features extracted by Tierpsy for untreated worms and worms exposed to 100 μM of each compound for 4 h.The top dendrogram shows the relationship of the individual extracted features within the entire feature set, with the tracking period bar denoting when during image acquisition the feature was extracted: pre-stimulation (pink), during stimulation with blue light (blue) and post-stimulation (green).The left dendrogram shows the phenotypic similarity of the worms and the colour map (top left) represents the normalized z-score of the features.Despite the strains generally clustering separately from each other, treatment of ret-1 LoF mutant with 2,5-di-tert-butylhydroquinone (DTHBQ) caused this strain to cluster alongside N2, suggesting phenotypic rescue of the mutant.(B) Position of N2 and ret-1 LoF worms in phenospace with respect to the top principal components in the behavioural feature set upon drug treatment.Untreated worms are denoted by stars and treated N2 or treated ret-1(syb4955).Error bars represent the standard error of the mean.Treatment of the ret-1 LoF strain with DTHBQ moved the mutant towards the untreated wild-type in phenomic space.(C) Key morphological, postural and locomotive features were significantly different between the untreated ret-1 LoF mutant (black boxes) and untreated N2 (grey boxes).From left to right: ret-1 mutants are longer; have decreased angular velocity of the head; increased angular velocity of the midbody; decreased midbody curvature; and increased midbody speed.White boxes show ret-1(syb4955) exposed to 100 μM DTHBQ for 4 h, rescuing of 3/5 of the key behavioural features.(D) Overall speed of ret-1 LoF mutant and N2 worms after 4 h exposure to DTHBQ.Treatment of N2 with DTHBQ resulted in an attenuation of the photophobic escape response (increased speed)following stimulation of the worms with three 10 s bursts of high-intensity blue light 100 s apart (denoted by blue shaded regions on the line plot), whereas the ret-1 LoF mutant showed resistance to the effects of DTHBQ.The mean peak speed (box plot, right) was calculated as an average of the maximum detected speed across the three independent pulses of blue light.Individual points marked on all box plots are average values from multiple worms (n = 3) in a single well.The different point colours indicate data from independent experimental days.P-vales were calculated using block permutation t-tests (n = 10 000 permutations).Permutations were shuffled randomly within, but not between, the independent days of image acquisition in order to control for day-to-day variation in the experiments and corrected for multiple comparisons using the Benjamini-Hochberg procedure to control the false discovery rate at 5%.P > 0.05 was considered not statistically significant (ns).

Table 1 Clinical features of affected individuals with RTN2-related distal hereditary motor neuropathy
1(syb4955)and N2 (data not shown), individual body parts,