A novel, rapidly progressive ataxia due to a spontaneous Myo5a mutation in mice impairs transport proteins and alters mitochondria

Abstract

Spontaneous mouse mutants have helped define genetic contributions to many phenotypes. Here we report a spontaneous Novel Ataxic Phenotype in mice. Ataxia findings were evident at postnatal day 11 in NAP mice and rapidly worsened, resulting in preweaning lethality. Using genome sequencing and genome-wide mapping, we identified a 3’ donor splice variant in exon 14 of Myo5a, encoding an actin-based motor protein. The variant in Myo5a (c.1752g>a) excises exon 14 and ablates MYO5A protein expression, which is implicated in intracellular transport and Griscelli syndrome type I in humans. NAP mice displayed expansion of PAX6-positive cells in the external granule layer of the cerebellum, and mass spectrometry analysis of cerebellar extracts uncovered differentially abundant proteins involved in short range organelle transport, and specifically proteins implicated with early endosomes. Using cerebellar lysates and primary neurons, we provide evidence for an interaction between MYO5A and ANKFY1, a known effector for the endosomal protein, RAB5A. We also found neurons from NAP mice had elongated mitochondria, linking MYO5A mitochondrial homeostasis. This allele provides new insight for Myo5a function in developmental neuropathology.

Graphical Abstract

A novel ataxic phenotype (NAP) spontaneouly appeared on the MRL/MpJ mouse background. Through genome sequencing, this recessive phenotype was identified as a novel mutation in Myo5a, leading to missplicing and loss of MYO5A protein. The loss of MYO5A resulted in aberrant expression of endosomal proteins in the cerebellum, and elongated mitochondria in neurons.

Plain Language Summary:

New spontaneous mutations occur in mice and humans. We uncovered a new mutation in mice in the gene Myo5a, which codes for a motor protein. In humans, mutations in MYO5A cause Griscelli syndrome, a syndrome with neurodegeneration. In mice, we found the spontaneous Myo5a mutation that leads to rapidly progressive movement disorders and death by 3 weeks of age. Proteins important for intracellular transport were altered, and the mitochondria in Myo5a null neurons were longer. These findings extend what is known about motor proteins and the means by which cargo move in neuronal cells.

INTRODUCTION

Spontaneous phenotypes have been a valuable source of information linking genotype and phenotype. In mice, spontaneous phenotypes have been useful in modeling both human diseases and gaining insight into new biological processes. When arising on specific genetic backgrounds, phenotypic expression can be altered, and background strain differences can also be used to identify genetic modifiers (1–3). The Murphy Roths Large (MRL/MpJ) strain is a mouse background strain which displays accelerated wound repair (4, 5). First discovered for their ability to more rapidly reseal ear punches, the MRL’s healing properties have been demonstrated to enhance recovery from injury in many acute and chronic settings (6–10), with a range of genetic regions implicated in these properties.

In mice, some of the earliest reported spontaneous phenotypes were in coat color variation (11, 12). A subset of coat color mutants displayed accompanying neurological phenotypes. The dilute lethal strains were identified by lighter coat color, with some alleles also displaying severe ataxia and early lethality (13). A small group of the dilute lethal series are defined by large chromosomal structural variants or even insertion of retroviral sequences (14, 15). Many of the dilute lethal allelic series were defined only through phenotypic characterization and breeding, lacking molecular definition.

The Myo5a gene was first linked to the dilute lethal phenotype in 1991 (16). Myo5a encodes a large unconventional myosin with a typical motor domain at its N-terminal head and a globular tail at its C-terminus. Like other class V myosins, the MYO5A protein mediates intracellular, actin-based transport and secretion, and the globular domain can inhibit ATPase activity of the head region (17). A chromosomal rearrangement that creates a fusion protein between the first two exons of the Gnb5 gene, encoding guanine nucleotide binding protein beta 5, and the C-terminus of Myo5a produces the flailer phenotype, which displays seizures and mild ataxia, surviving to adulthood with normal reproduction (18, 19). In cellular models, transgenic expression the MYO5A tail without the head in neurons is sufficient to inhibit intracellular transport (20).

Myo5a mutations produce coat color defects by inhibiting melanosome capture and exocytosis in melanocytes (21). In melanocytes, MYO5A binds melanosomes via a tripartite complex involving MYO5A, melanophilin (encoded by Mlph), and the small GTPase protein RAB27A (22). In mice, loss of function mutations in either Mlph or Rab27a phenocopy Myo5a coat color findings, but do not produce neurological phenotypes (23, 24). In humans, Griscelli syndrome is a recessive disorder associated with partial albinism and bone marrow defects including neutropenia and thrombocytopenia (25). Recessive mutations in MYO5A lead to Griscelli syndrome type 1 where patients have been described as having silvery hair with large pigment clumps found in the hair shafts (26). Type 1 Griscelli syndrome patients have hypotonia, and developmental and motor delay.

In the mouse central nervous system (CNS), loss of MYO5A was shown to hinder Purkinje cell long-term depression response (27). The loss of MYO5A was associated with reduced smooth endoplasmic reticulum (ER) in dendritic spines of Purkinje cells (28). The reduction of the smooth ER in dendritic spines disrupts local Ca²⁺ release which is required for glutamate receptor endocytosis. An excess of glutamatergic receptors in the post synapse prevents the dampening long-term depression response in Purkinje cells (29). In other neuronal cell types, MYO5A has been associated with synaptic signaling, specifically in tethering a pool of refilling vesicles following stimulation to the plasma membrane (30, 31). Additionally, MYO5A has been shown to directly interact with RAB3A containing compartments and the syntaxin-binding protein 1 (STXBP1), both of which are found on synaptic vesicles (32, 33). MYO5A has also been implicated in receptor recycling, myelination, mRNA transport, mitochondrial movement, and morphology, and the neuromuscular junction (34–39).

We now characterized a spontaneous Novel Ataxic Phenotype (40) in mice, which arose in a backcross of an unrelated neuromuscular defect due to a mutation in Sgcg, the gene for γ-sarcoglycan, in the background of the MRL strain. Through breeding, we separated the NAP phenotype from the Sgcg^−/− allele, and we demonstrated NAP’s autosomal recessive inheritance. We used genome-wide mapping and genome sequencing and found Myo5a (c.1752g>a), which results in the excision of exon 14 from the Myo5a transcript and ablates MYO5A protein expression. We identified an increase in granule cells in the external granule layer of the cerebellum in NAP mice, and tandem mass tag (TMT)-based quantitative proteomic analysis revealed a reduction for the level of proteins involved in short range organelle trafficking in NAP mice compared to their WT counterparts. We provide evidence that MYO5A and ANKFY1 interact, consistent with a role for MYO5A in the endosomal system, and we found that NAP primary neurons have elongated mitochondria, implicating MYO5A in mitochondrial homeostasis.

MATERIALS AND METHODS

Approvals

All studies were conducted under approval from the Animal Care and Use Committee at Northwestern University.

Animals

Sgcg^−/− null mice were previously described and bred into the 129T2/SvEmsJ (129T2) mice (stock 002065, Jackson Laboratories) (1, 41). These mice were backcrossed to the MRL/Mpj strain (MRL), (stock 000486, Jackson Laboratory). Mice were bred and housed in a specific pathogen–free facility on a 12-hour light/dark cycle and fed ad libitum. Male and female mice were used for all experiments. Sex differences were not assessed in these experiments. All mice were age matched per assay (see specific assay for exact age).

Gait Analysis

Gait analysis was conducted on P17 mice using an automated Digigait system (42). Briefly, mice are individually placed into a glass rectangular enclosure with a transparent treadmill as the floor. Gait was captured for ~5 sec intervals using a highspeed camera positioned below the treadmill, which was set at a specific speed of either 2 cm s⁻¹ or 5 cm s⁻¹.

DNA isolation

DNA was isolated using a Gentra Puregene Tissue Kit (43 158667, Qiagen). Briefly, 5 mm of tail tissue was used per mouse to isolate genomic DNA. The tail was lysed, and DNA was precipitated using the Puregene Protein Percipitation Solution (43 158667, Qiagen). The DNA was washed with ethanol, air dried and then resuspended. The concentration and quality of the DNA was determined using a nanodrop (43 ND-2000, Thermo Fisher). DNA from every sample was diluted to equal concentrations prior to all experiments.

Whole Genome Sequencing

Genomic DNA was extracted from tail snips. Whole genome sequencing (WGS) was performed using PCR-free libraries on an Illumina Genome Analyzer II system (parental backcross strains) or an Illumina XTen machine yielding (parental and affected mice) yielding 30X coverage across the genome. WGS was performed on two parents who had produced NAP offspring and two affected NAP mice. One of the affected NAP mice was a direct offspring of the sequenced parents and the other was the product of a different parents who had also produced NAP offspring. Additionally, a single mouse from each parental line used in the backcross (MRL and 129T2) was also sequenced. Reads were aligned to the mouse reference genome sequence GRCm38/mm10 using the Burrows-Wheeler Aligner (BWA) and variants were called using the Genome Analaysis Tool Kit (GATK) (44, 45). The MegaSeq pipeline was used for analysis (46). Variants were annotated using snpEff (47). Variant filtration and analysis were performed with in house scripts using Perl 5 or R 3.3.0.

Myo5a genotyping

Genotyping for Myo5a (c.1752g>a) using the following genomic primer sequences: forward primer: 5′ ATGTGATTTCAGGTGGAGTAC 3′, and reverse primer: 5′ ATAGGACCTGACACGTGTTAC 3′. Products were amplified by PCR using Phusion High-Fidelity DNA Polymerase (New England Biolabs) with the following cycle conditions: initial denaturation 98°C, 30 seconds followed by 98°C, 10 seconds; 58°C, 30 seconds; 72°C 60 seconds for 34 cycles, and a final extension 72°C for 5 minutes. Products were run on 2% agarose gels with ethidium bromide. For Sanger sequencing, chromatograms were visualized using 4Peaks Software (Nucleobytes). A related protocol was transferred to Transnetyx for colony maintenance genotyping.

LOD Score Determination

A Giga Mouse Universal Genotyping Array (GigaMUGA, cat # 550, Neogen) was used to obtain a SNP marker map with approximately 140,000 SNPs per mouse (n=24; 12 NAP, 6 Parents, 2 129 Mice, 2 MRL mice and 2 B6 mice). We then genotyped selected mice from this SNP array for Myo5a (c.1752g>a) (n=20; 8 NAP, 6 Parents, 2 MRL, 2 129, 2 B6). The Myo5a (c.1752g>a) genotype was incorporated into the SNP array for a subsequent LOD analysis focusing on the chromosome 9 interval containing Myo5a. Parametric linkage analysis used MERLIN software (48).

Gene Expression Analysis

Hemi-forebrains from P18 mice were isolated and snap frozen. RNA was isolated from these hemi-forebrains using TRIzol^™ Reagent (43 15596026, Thermo Fisher Scientific). cDNA was synthesized using qScript cDNA SuperMix (43 95047–100, QuantaBio). Relative fluorescence was captured using iTaq Universal SYBR Green Supermix (43 1725120, BIO-RAD) and measured with CFX96 Touch Deep Well Real-Time PCR System (BIO-RAD). The expression of Myo5a (forward primer: 5’ CTTTGAGAAGCCCCGCAT 3’, and reverse primer: 5’ GGGACTGATGGCCTTCTCAT 3’) was probed and analyzed relative to an internal control Rn45s (forward primer: 5’ GTAACCCGTTGAACCCCATT 3’, and reverse primer: 5’ CCATCCAATCGGTAGTAGCG 3’) using an unpaired parametric two tailed t-test.

Immunoblotting

Total protein was isolated from snap frozen, P18 hemi-forebrains or cerebella using a Whole Tissue Lysis Buffer (50 mM HEPES (pH 7.5), 150 mM NaCl, 2 mM EDTA, 10 mM NaF, 10 mM Na-pyrophosphate, 10% glycerol, 1% Triton X (100x)) buffer supplemented with 1× cOmplete Protease Inhibitor Cocktail (cat # 11697498001 CO-RO; Roche), PhosSTOP phosphatase inhibitor (43 4906837001, Millipore) and quantitated with the Pierce^™ BCA Protein Assay kit (43 23225, Thermo Fisher). Either 10, 14, or 25μg lysates from forebrain or cerebellum were incubated at RT with 2× Laemmli Sample Buffer (Bio-Rad) and 2-mercaptoethanol (Millipore-Sigma) followed by 10 min at 99°C, and then separated on 4–15% Mini-Protean TGX Stain-Free Protein Gel (Bio-Rad). Separated proteins were transferred to Immobilon-P PVDF membrane and incubated with anti-MYO5A antibodies (cat # 3402S, Cell Signal and Myo5a antibody (49) diluted at 1:1000 in StartingBlock T20 Blocking buffer (43 n37543, Thermo Fisher). Goat anti-Rabbit IgG secondary (43 31460, Thermo Fischer) conjugated to horseradish peroxidase was used at 1:2500 (cat # 111–035-003; Jackson ImmunoResearch Laboratories). SuperSignal West Pico Chemiluminescent Substrate and SuperSignal West Femto Maximum Sensitivity Substrate (cat # 34080 and 34096; Thermo Fisher Scientific) were applied, and membranes were visualized using an Invitrogen iBright CL1000 Imaging System (cat # A32749; Thermo Fisher). Pierce Reversible Protein Stain Kit for PVDF Membranes (MemCode; cat # 24585, Thermo Fisher) was used to stain the entire blot to ensure complete transfer and equal loading. Immunoblot bands were quantified using ImageJ gel analysis tools (NIH).

Histology

Whole brains from P18 mice were isolated and fixed in 10% formalin. Whole brains were sagittally sectioned at 15 μm thickness. Sections were stained with hematoxylin and eosin (43 12013B and 1070C, Newcomer Supply) per manufacturers’ protocol. All images were acquired using a Keyence BZ-X810 microscope at the same exposure. The same cerebellar lobule was compared for all mice across both groups. Image analysis was done using ImageJ (NIH).

Immunofluorescence microscopy

P18 mice were euthanized and transcardially perfused with ice-cold PBS containing both protease and the phosphatase inhibitors cOmplete Protease Inhibitor Cocktail (cat #11697498001 CO-RO, Roche) and PhosSTOP phosphatase inhibitor (43 4906837001, Millipore). Whole brains were isolated and submerged in 4% PFA at 4°C overnight. Brains were then cryopreserved in 30% w/v sucrose in PBS until sectioning. The cerebellum was removed and sectioned separate from the forebrain. Fifteen μm sagittal cerebellar slices were directly mounted onto slides (Fisherbrand Superfrost Plus Microscope Slides (cat # 12–550-15, Fisher) and stored in −80°C until staining. For staining, slides were removed from the −80°C and brought to RT. Mounted sections were then washed with PBS 3x for 5 minutes then submerged in a Coplin jar containing Citrate Buffer, Antigen Retriever (43 C9999, Sigma). The Coplin jar was sealed and placed in a beaker containing boiling water for 10 minutes. Slides were then removed, cooled to RT, and then washed with PBS 3x for 5 minutes. The sections were blocked and permeabilized for 1 hour at RT with 5% BSA, 1% donkey serum and 0.1% Triton X-100 in PBS sterilely filtered through a 0.2 μm disc filter. PAX6 antibody (43 GT9412, GeneTex) or calbindin antibody (43 NBP2–50048, Novus) was diluted in blocking buffer (1:100) and applied to the sections overnight at 4°C. The following day sections were washed with PBS 3x for 10 minutes and then the secondary Alexa Fluor 594 donkey anti-rabbit (43 A32754, Thermo Fischer) or Alexa Fluor 488 donkey anti-mouse (43 A32766, Thermo Fischer) was diluted into blocking buffer (1:2000) and applied to sections for one hour at RT protected from light. Sections were then washed with PBS 3x for 5 minutes. Slides were mounted with Prolong Gold Antifade (43 P36934, Thermo Fischer), and coverslips were placed and sealed with nail polish.

Primary Neuronal Isolation

Pre-coated poly-D-lysine 1.5 cm coverslips (43 P35GC-1.5–14-C, Mattek) were incubated overnight with 10 μg/μl of laminin (43 L2020, Sigma) diluted in Neurobasal A Medium (43 10888022, Thermo Fischer) at 37°C in 95% O2, 5% CO2. The following day, P4-P6 mice were euthanized, brains were removed, cerebella were dissected into ice cold PBS, and then placed into preheated papain buffer (43 LK003176, Worthington) supplemented with 50 μg/ml of DNase (43 11284932001, Sigma) at 37°C for 15 minutes. After 5 minutes, the tissue was gently triturated with a P1000 pipette tip. After 10 minutes, the slurry was further triturated with a fire polished Pasteur pipette. Ovomucoid was added (10 mg/ml Ovomucoid (cat # LS003085, Worthington); 10 mg/ml Bovine Serum Albumin; 10 μg/ml DNase in 1x PBS). This slurry was strained using a 40 μm mesh filter and then centrifuged at 4°C for 5 min at 300 g. The supernatant was removed, the cells were applied to a density gradient using the Ovomucoid buffer, and centrifuged at 4°C at 100g for 5 minutes. The Neuron Isolation Kit, mouse (43 130–115-389, Miltenyi) was used to deplete all non-neuronal cells. Primary neurons were plated on laminin coated coverslips at 100,000 cells per 1.5 cm². Culture media was added (Neurobasal A-Medium, 100x Glutamax I (cat # A1286001, Thermo Fisher), 100x Pen/Strep, 50x B-27 (serum free) (43 17504044, Gibco), 250 μM KCl) and cells were placed in 37°C in 95% O2, 5% CO2. The next day the media was fully changed. Then one half of the volume of culture media added per coverslip was replaced every 4 days until cells were used for assays.

Neuronal Immunofluorescence Microscopy

Cells were washed with 1x PBS and fixed in 4% paraformaldehyde for 10 minutes at RT, then washed with PBS 3x for 5 minutes. Cells were permeabilized with 0.1% Triton X-100 in PBS for 10 minutes at RT. Cells were blocked with 10% donkey serum in PBS for 1 hour at RT. Antibodies (ANKFY1 (cat # sc-39335, Santa Cruz (1:50)), MYO5A (43 3402S, Cell Signal (1:100)) and RAB5A (43 2143, Cell Signal (1:150)) were diluted into blocking buffer and applied to cells overnight at 4°C. The following day, cells were washed with PBS 3x for 5 minutes. Secondary antibodies (Alexa Fluor 488 donkey anti-mouse (cat # A32766TR, Thermo Fischer), Alexa Fluor 594 donkey anti-rabbit (43 A32754, Thermo Fischer) and Alexa Fluor 647 Donkey anti-Rabbit (43 A-31573, Thermo Fischer) were diluted into block (1:2000) and applied to cells for 1 hour at RT covered from light. Cells were then washed, Prolong Gold Antifade (43 P36930, Thermo Fischer) was applied. Images were acquired on either a Nikon W1 Dual CAM spinning disk confocal microscope with a 60× objective, using NIS-Elements software or a Keyence BZ-X810 microscope with a 100x objective. All images acquisition settings were the same across samples.

Immunoprecipitation (IP)

Dynabeads Protein G (43 10003D, Invitrogen) were used following the manufacture’s protocol with modifications. Briefly, 30 μl of the Dynabead Protein G slurry was used per IP and washed once with 1x PBS prior to antibody binding. Both anti-Myo5a antibody (43 3402S, Cell Signal) at 1:50 and a control antibody, rabbit gamma globulin (43 31887, Thermo Fischer) at 5 μg/μl were diluted into 200 μl of PBS with 0.1% Tween^®20 and added to the washed beads for 30 minutes at 4°C with constant rotation. One mg of cerebellar lysate was added to each bead-antibody tube and incubated overnight at 4°C with constant rotation. Bead-antibody-antigen complexes were washed three times in PBS with 0.1% Tween and then washed three times with Whole Tissue Lysis Buffer (as described above). The final complex was resuspended into 35 μl of Laemmli Sample Buffer (43 1610737, Biorad) and heated at 99°C for 10 minutes. Thirty μl was separated by gel electrophoresis and used in immunoblotting. Input comparisons used 10 μg of either cerebellar or forebrain lysates. Membranes were incubated with the following antibodies: anti-Ankfy1 (43 A305–512A, Thermo Fischer), anti-MYO5A (43 3402S, Cell Signal) and anti-RAB5A (43 2143, Cell Signal) diluted at 1:1000.

Mitochondrial imaging

MitoTracker^™ Deep Red FM (43 M22426, Thermo Fischer) was used per manufacture’s protocol. Briefly, 50 μg of Mitotracker was reconstituted into DMSO and diluted in neuronal culture media at a working concentration of 250 nM and applied to primary neuronal cells for 30 minutes. Cells were immediately fixed in 4% PFA and mounted with Prolong Gold Antifade.

Tandem Mass Tag Mass Spectrometry (TMT-MS)

TMT-MS sample preparation was performed as previously described (50, 51). In short, 200 μg of homogenized cerebellar extracts were methanol-chloroform precipitated. Extracted protein was resuspended in 6M guanidine in 100 mM triethylammonium bicarbonate and further reduced of disulfide bonds with DTT, followed by alkylation of cysteine residues with iodoacetamide. Proteins were digested overnight at 37°C with 3 μg Trypsin/LysC (Promega). The digest was then acidified with formic acid and desalted (C18 HyperSep columns, cat # 30108, Thermo Fischer). Peptides were resuspended in 100mM triethylammonium bicarbonate and 100μg for used for each respective isobaric TMT tag. After a 75 min incubation at RT, the reaction was quenched with 5% (v/v) hydroxylamine to 0.3%. Isobarically labeled samples were then combined 1:1:1:1:1:1:1:1:1:1 and subsequently desalted. The sample was then fractionated using high pH reversed-Phase columns (Pierce) and dried before reconstituted in LC-MS Buffer A (5% acetonitrile, 0.125% formic acid) for LC-MS/MS analysis.

TMT-MS analysis was performed as previously described (50, 51). Samples were resuspended in 20 μl Buffer A (5% acetonitrile, 0.125% formic acid), and 3μg of each fraction was loaded for LC-MS analysis. Orbitrap Fusion was used to generate MS data. The chromatographic run was performed with a 4 hour gradient as previously described (50, 51). In MultiNotch MS3, the top ten precursor peptides were selected for analysis then were fragmented using 65% HCD before orbitrap detection (52). A precursor selection range of 400–1200 m/z was chosen with mass range tolerance. An exclusion mass width was set to 18 ppm on the low and 5 ppm on the high. Isobaric tag loss exclusion was set to TMT reagent. Additional MS3 settings include an isolation window = 2, orbitrap resolution = 60 K, scan range = 120 – 500 m/z, AGC target = 6*105, max injection time = 120 ms, microscans = 1, and datatype = profile.

TMT-MS data analysis was performed as previously described with quantitative changes to calculate protein turnover with TMT-MS (50, 51). Protein identification, TMT quantification, and analysis were performed with The Integrated Proteomics Pipeline-IP2 (Integrated Proteomics Applications, Inc., http://www.integratedproteomics.com/). Proteomic results were analyzed with ProLuCID, DTASelect2, Census, and QuantCompare. MS1, MS2, and MS3 spectrum raw files were extracted using RawExtract 1.9.9 software (http://fields.scripps.edu/downloads.php). Fully and half-tryptic peptide candidates were included in search space, all that fell within the mass tolerance window with no miscleavage constraint, assembled and filtered with DTASelect2 (ver. 2.1.3). Static modifications at 57.02146 C and 229.1629 K at N-term were included. The target-decoy strategy was used to verify peptide probabilities and false discovery ratios (52). Minimum peptide length of six was set for the process of each protein identification and each dataset included a 1% FDR rate at the protein level based on the target-decoy strategy and Isobaric labeling analysis was established with Census 2 with no intensity threshold applied.

Data Availability

The mass spectrometry proteomics data have been deposited to the MassIVE repository with the identifier MSV000095598.

Graphics

The Myosin Va protein graphic depicted in Figure 3E, 4F and 6F were created with BioRender.com

Figure 3. A homozygous splice variant in exon 14 of Myo5a in NAP mice.

(A) PCR of genomic DNA from mice who produced NAP offspring and NAP offspring demonstrated Myo5a (c.1752g>a) in the heterozygous state in parents and in the homozygous state in NAP mice (NAP mice (n=8), parents (n=6)). Unaffected littermates were either heterozygous or wild type for Myo5a (c.1752g>a) (n=6). (B) LOD score calculations derived from genotypes determined from a mouse universal genotyping array are shown from chromosome 9. The position of the Myo5a (c.1752g>a) is indicated with the red line and yields a LOD score=3.52 (n=20 mice; 8 NAP, 6 parents, 2 MRL, 2 129, 2 B6). (C) Across 13 litters (n=95) NAP mice were born in the expected Mendelian ratio, indicating no significant prenatal loss (P=0.84, Chi-squared test). (D) Myo5a (c.1752g>a) maps to the 3’ boundary of Myo5a exon 14 (of 41 exons). (E) The position of Myo5a (c.1752g>a) (green arrow) is within the N-terminal, “head” region of MYO5A.

Figure 4. In frame skipping of exon 14 in Myo5a causes loss of Myosin Va protein.

(A) Primers directed at exonic sequences were used to amplify exons 13 to 15 using RNA from mouse forebrains. The WT band was seen at 193bp, and products from NAP brains were ~80 base pairs smaller than the WT product, consistent with loss of exon 14. (B) Sanger sequencing of the amplified products confirmed loss of exon 14 in the NAP Myo5a transcript. (C) Quantitative RT-PCR showed a nonsignificant trend towards lower Myo5a transcript in forebrains at P18 (P=0.08, n=5 per group). (D) Immunoblotting of lysates from P18 forebrains did not detect MYO5A protein using an antibody to the C-terminus. (E) Quantification from panel D showed a significant reduction in MYO5A (****P=<0.0001, n=5 per group). (F) The position of the antibody epitope is shown. Comparisons made with unpaired parametric two tailed t-test; data represent mean±SD.

Figure 6. Differential protein expression in P18 NAP vs WT cerebella supports defects in short range organelle transport.

(A) Cerebellar Tandem Mass Tag-Mass Spectrometry (TMT-MS) quantified 4,497 unique proteins, and the volcano plot displays the 273 statistically differential proteins (164 significantly decreased, 109 significantly increased, n=4 per group). Myo5a was the most significantly reduced (P=0.0002). (B) Gene Ontology (GO) analysis for significantly decreased proteins included Synaptic Vesicle Cycle (GO:0099504), Actin Cytoskeleton (GO:0015629), Pre and Postsynapse (GO:009793 and GO:0098794, respectively). (C) A heatmap displays the significantly reduced terms associated with the synapse, either pre-, post- or both. (D) Shown are z-scores derived from the TMT intensities for the 156 proteins present with in the dataset that are present in the Early Endosome (GO:0005769) term. NAP mice showed a reduction in early endosome proteins (****P<0.0001). (E) TMT intensities for specific proteins within the Early Endosome GO term implicate Rab5a (*P=0.049), and two RAB5A effectors, APPL1 (*P=0.013,) ANKFY1 (P=0.070). (F) Graphical hypothesis for a MYO5A:ANKFY1:RAB5A tripartite complex implicating early endosomes. Comparisons made using unpaired parametric two tailed t-test. Data represent mean±SD.

Statistical Analysis

TMT-MS data was visualized using R 3.3.0 software. Data was analyzed using Prism 10. All datasets were analyzed by means testing via unpaired parametric two tailed Students t-test. In all cases, P<0.05 was defined as statistically significant. Statistical data are reported as mean ±SD Confidence intervals are reported as 95% (95% c.i.).

RESULTS

A spontaneously arising early onset, novel ataxic phenotype in mice

In the course of analyzing the effect of genetic background on the unrelated muscular dystrophy phenotype caused by loss of Sgcg (γ-sarcoglycan), we observed mice with an early onset ataxic phenotype. The ataxia phenotype appeared while conducting a backcross of heterozygous Sgcg^+/− mice in the 129T2/SvEmsJ (129) strain with the white coat color MRL/MpJ strain (MRL) (5, 41). At the fifth generation, interbreeding heterozygous Sgcg^+/− produced a small number of mice with a wobbly gait and a rapidly progressive neurodegenerative process resulting in natural death in the first few weeks of life, or requiring sacrifice for humane reasons (Fig 1A). We termed these animals Novel Ataxic Phenotype (40) mice. Genotyping excluded involvement of the Sgcg locus on chromosome 14, and through breeding, the Sgcg locus was removed from the crosses while the NAP findings remained, confirming the NAP locus as genetically distinct from Sgcg (Fig 1A).

Figure 1. A “Novel Ataxic Phenotype” (40) spontaneously arose in a backcross.

(A) A backcross was conducted between the MRL/MpJ (MRL) and 129T2/SvEmsJ (129) carrying an Sgcg null mutation. Interbreeding the 5^th generation revealed a distinct novel ataxic phenotype (40) (red), which was genetically distinct from the Sgcg null mutation. (B) At P18, NAP mice demonstrated hindlimb collapse (Top Row). At P18, NAP mice were smaller than unaffected age matched siblings (***P=0.0001, n=25 per group) (Bottom Row). (C) At P17, NAP mice were unable to maintain a walking speed of 5 cm s⁻¹ speed compared to WT counterparts. At a slower speed of 2 cm s⁻¹, NAP mice made contact to walls of the apparatus more frequently (P=0.06) and spent more time leaning against the wall of the apparatus compared to WT (**P=0.003, n=3 per group). (D) The stride of the mice was defined by the time the foot kept contact to the surface. P17 NAP mice displayed a significant increase in stride in three out of 4 limbs as compared to WT (P=0.1, *P=0.03, *P=0.03, *P=0.03 respectively, n=6 WT, n=5 NAP). Comparisons made using unpaired parametric two tailed t-test. Data shown as mean±SD.

NAP mice had symptom onset beginning at approximately postnatal day 13 (P13) with rapid worsening of gait defects (Video S1). By P17, NAP mice were unable to sustain forward movement without falling to the side (Video S2), and NAP mice displayed a prolonged righting reflex when placed in a supine position (Video S3). At P18, NAP mice display a hindlimb collapse when suspended from the tail (Fig 1B, top). At P18, NAP mice were smaller than their unaffected siblings (***P=0.0001) (Fig 1B, bottom). Gait analysis using a Digigait apparatus was conducted at P17 (Fig 1C). Although wildtype (WT) mice could walk at 5 cm s⁻¹, the NAP mice were unable to walk at that same belt speed of 5 cm s⁻¹. At the slower 2 cm s⁻¹ belt speed, video analysis showed the NAP mice were unable to walk without using the side wall for support, measured as the time spent leaning on a side wall during movement (**P=0.003, unpaired parametric two tailed t-test) and a trend in touching the side wall per run (P=0.06). At P17, NAP mice also displayed an increase in stride duration as defined by the time the foot kept contact to the surface in 3 of 4 limbs (*P=0.03, *P=0.03, *P=0.03, P=0.1 respectively, unpaired parametric two tailed t-test) (Fig 1D).

Whole Genome Sequencing (WGS) identified potential pathogenic variants

We assessed heritability of NAP through breeding using data from multiple litters. From four litters (n=22 mice), 15 were unaffected (68%) and 7 were NAP (32%) with 3 males and 4 females, consistent with autosomal recessive inheritance (Fig 2A). Whole genome sequencing (WGS) was conducted to identify pathogenic variants. WGS was performed on 6 mice including parents of NAP mice (n=2), a NAP mouse from these parents and an additional NAP produced from two other parents. We also included one mouse each from both parental lines used in the backcross, MRL and 129T2. When aligned with GRCm38/mm10, the average number of variants per mouse was ~7.4 million (Fig 2B). We applied a systematic filtering strategy to reduce the list of variants to those potentially pathogenic, evaluating variants that fit autosomal recessive inheritance. This filtering step reduced the list of potentially causal variants to 4,243. Given the spontaneous presentation, we also assumed the pathogenic variant would be de novo with respect to the parental lines. This reduced the total variant count to 2,345. We then expanded the de novo analysis to 36 distinct inbred strains of laboratory mice, made available through the Sanger Institute (53). This reduced the list of potentially causal variants to 1,300. Of these, only two variants mapped to coding regions, and only one of these was in a validated gene (Fig 2C). The coding variant was a synonymous variant in the Myo5a gene, and this variant disrupted a canonical splicing site.

Figure 2. Whole Genome Sequencing (WGS) of NAP mice, parents, and parental strains identified potential pathogenic variants.

(A) Across 4 litters (n=22), NAP mice appeared 7 times (~32%), with males and females comparably represented. (B) WGS was conducted on two phenotypically confirmed parents and one of its NAP offspring, and a more distantly related NAP mouse. WGS was also carried out on single animals from the parental MRL/MpJ (MRL) and 129T2/SvEmsJ (129) strains, producing ~7.4 million variants per mouse. (C) Based on the preliminary autosomal recessive inheritance pattern, we applied a systematic filter to these variants per mouse, which reduced the number of variants to 4,243. Second, assuming a new spontaneous mutation, variants were removed that were present in the 129 or MRL background strain, reducing the number of variants to 2,345. We then expanded the analysis to 36 additional strains available through the Sanger Institute, again assuming a new spontaneous mutation, leaving 1,300 total variants. Only one of these variants were within a coding region, a synonymous SNP in Myo5a, which encodes the Myosin Va protein (MYO5A).

A homozygous splice variant in exon 14 in the Myo5a gene in NAP mice

The Myo5a (c.1752g>a) (chr9:75156264 (GRCm38/mm10)) falls at the 3’ end of exon 14 (Fig 3A, top). PCR of genomic DNA from NAP mice, parents, and unaffected siblings amplified the Myo5a (c.1752g>a) region. Sanger sequencing confirmed heterozygosity in parents and homozygosity in NAP mice (Fig 3A, bottom). For mapping, a subset of parents and homozygous offspring were genotyped using a mouse universal genotyping array (54). We added the genotype for Myo5a (c.1752g>a) to the SNPs on chromosome 9 in the vicinity of Myo5a and performed a parametric linkage analysis. The computed LOD score for Myo5a (c.1752g>a) was 3.52 (n=20; 8 NAP, 6 parents, 2 MRL, 2 129T2, 2 B6) (Fig 3B), providing statistical evidence for Myo5a (c.1752g>a) as causing the ataxic phenotype. We genotyped Myo5a (c.1752g>a) across 13 litters (Fig 3C) and confirmed autosomal recessive inheritance with no deviation from Mendelian ratios (P=0.84, Chi-squared test). Myo5a (c.1752g>a) maps to the last base pair in exon 14 in the transcript (Fig 3D). Myo5a (c.1752g>a) locates to the N-terminal head domain MYO5A protein (Fig 3E), upstream of the C-terminal tail domain that binds specific organelles to facilitate short-range trafficking.

Molecular characterization of the Myo5a splice variant

Myo5a is broadly expressed across cell types, and is abundantly present in both melanocytes and the CNS (55), where it has previously been linked to the dilute lethal phenotype (16, 56). We amplified the regions surrounding Myo5a (c.1752g>a) using RNA from forebrains of P18 WT and NAP mice and confirmed a deletion in the Myo5a transcript consistent with the loss of exon 14 (Fig 4A). Sanger sequencing of these RT-PCR products demonstrated exon 14 excision in all NAP mice (Fig 4B) (n=3 per group). To assess Myo5a expression, we performed quantitative RT-PCR (qPCR) using RNA from P18 forebrains (Fig 4C). NAP mice showed a nonsignificant reduction in Myo5a expression (P=0.08). We assessed MYO5A protein level in forebrain lysates from P18 mice using immunoblotting. Using an antibody to the C-terminus of the protein, MYO5A protein was undetectable in NAP mice while it was readily detected at the appropriate size in WT samples (Fig 4DEF). The loss of MYO5A in the cerebellum was confirmed with two separate antibodies (Fig S1). These data indicate the in-frame loss of exon 14 destabilizes MYO5A protein, accounting for the findings in NAP mice.

NAP mice show increased PAX6-positive granule cells in the external granule layer (57) at P18.

Since cerebellar function is critical for normal gait and often implicated in ataxia phenotypes, we examined NAP and WT cerebella (58). Cerebellar sagittal sections from P18 NAP mice appeared grossly normal (Fig 5A, top). To further characterize cerebellar tissue, we quantified the number of Purkinje cells within the Purkinje cell layer normalized to a specified distance (Fig 5A, bottom left, dotted white line). The NAP mice had comparable Purkinje cell counts compared to WT (P=0.28) (Fig 5A, top right). We also assessed the number of cells within the molecular layer, normalized for area, and found that NAP mice had similar cell counts to WT (P=0.24) (Fig 5A, bottom right). Because a Myo5a^−/− rat was previously reported to have an increased cell number in the EGL, we evaluated NAP mice for this defect (59). We stained P18 sagittal cerebellar slices with the granule cell marker PAX6 (Fig 5B top left), and we assessed the number of PAX6 positive cells normalized for area in the EGL by capturing the “base” of each EGL within five different sulci regions in each sample. An example is highlighted by the 40x image of PAX6-positive cells within the EGL for sulci IX/X (Fig 5B, bottom, white rectangle). Five separate sulci were analyzed per sample (4 samples per group, except for WT sulci II/III and III/IV which only had 3 samples analyzed). Combining the 5 EGLs captured per sample, the NAP mice displayed an increase in the number of PAX6-positive cells relative to WT at P18 (P=0.0001) (Fig 5B, right). This data is consistent with the Myo5a^−/− rat model confirming a role for MYO5A in post-natal cerebellar development. Lastly, we stained P18 sagittal cerebellar sections for the Purkinje cell marker calbindin, and we observed comparable dendritic complexity between groups (Fig S2).

Figure 5. PAX6-positive cells were increased in the external granule layer (57) of P18 NAP mice.

(A) H&E staining of WT and NAP P18 sagittal cerebellar sections. The number of Purkinje cells along the Purkinje cell layer, normalized for distance was comparable between WT & NAP mice (P=0.28, n=4). The white dashed line is an example distance used to count Purkinje cells. White arrows indicate example Purkinje cells. The number of cells within the molecular layer was quantified, normalized for area, and was similar between groups (P=0.24, n=4). Scale bars: 4x=500μm, 20x=100μm. (B) PAX6 staining was used to mark granule cells in P18 sagittal cerebellar slices. The number of PAX6-positive granule cells in the EGL per sulcus were counted and normalized to area (Bottom row, white rectangles). The PAX6-positive cell count in the EGL of NAP mice was greater (***P=0.0001, n=4 per group). Scale bars: 4x=500μm, 40x=50μm. Comparisons made with unpaired parametric two tailed t-test; data represent mean±SD.

Tandem Mass Tag Mass Spectrometry (TMT-MS)-based quantitative proteomic analysis of whole cerebellar extracts reveals NAP deficits in short range organelle transport

TMT-MS proteomic analysis was carried out using cerebellar extracts from P18 NAP and WT mice (n=4 biological replicates per group). Given MYO5A’s role in cellular transport, we hypothesized that a loss of MYO5A would alter protein content. MS analysis quantified 4,497 unique proteins, and Fig 6A displays a plot of the 164 proteins that were significantly reduced and the 109 that were significantly increased (unpaired two tailed t-test). Of the 273 differentially expressed proteins, MYO5A was the most significantly reduced (Fig S3A) and had the largest absolute Log2 ratio between NAP and WT (−1.99). In the TMT-MS data, there were few MYO5A peptides in NAP cerebella, while MYO5A was abundantly represented in the WT sample (Fig S3B). The distribution of MYO5A peptides along the length of the protein was similar between WT and NAP, suggesting no excess of N-terminal peptides in the NAP sample. Of other myosins, MYH9 trended towards being increased (Fig S3C). Gene ontology (GO) enrichment analysis of proteins with reduced levels in the cerebellum of NAP mice revealed terms such as Synaptic Vesicle Cycle (GO:0099504), Pre- and Postsynapse (GO:009793 and GO:0098794, respectively), and Actin Cytoskeleton (GO:0015629) (Fig 6B). These findings are consistent with MYO5A’s role in synaptic transmission (32, 60).

The synaptic terms were then subdivided as specific to either pre-synaptic, post-synaptic or both (Fig 6C). MYO5A facilitates short range trafficking in part through its interaction with Rab proteins, where MYO5A can directly interact with specific Rab proteins or requires intermediary effector proteins forming tripartite complexes (22, 61). We identified three Rab proteins significantly reduced in NAP cerebella, including RAB3A and RAB3C, both of which are implicated in synaptic vesicle function (62). MYO5A is known to directly interact with RAB3A (32). The third Rab protein that was reduced was RAB5A, which is also known to contribute to synaptic function, but RAB5A is also known to have a critical role in the early endosomal system (63, 64). To assess for a proteomic signature of early endosome deficit, we analyzed all 436 proteins associated with the GO Term Early Endocytosis (GO:0005769), of which 156 were present in the dataset. We computed the z-score for each of the 156 proteins and found NAP cerebella had a significant reduction in the abundance for proteins involved in the early endosomal system (P<0.0001). TMT intensities for proteins within the list implicated RAB5A (P=0.049) and two RAB5A effectors, APPL1 (P=0.013) and ANKFY1 (P=0.070) (Fig 6E). A prior report links MYO6 to RAB5A and APPL1, providing support for this analysis (65). MYO5A is not thought to directly interact with RAB5A (61), so based on these data, we hypothesized MYO5A may use ANKFY1 as an intermediary adaptor for its interaction with RAB5A, which would implicate Myo5a with early endosomal function (Fig 6F).

Colocalization and immunoprecipitation suggest MYO5A interacts ANKFY1, a RAB5A effector, and NAP neurons have elongated mitochondria.

Primary neurons from P4-P6 cerebella were isolated since this age permits ready isolation of primary neurons. Neuronal identity was confirmed by both distinct morphology and neuron specific staining of β3-tubulin (Fig S4). Primary neurons were stained with antibodies to MYO5A and ANKFY1, demonstrating colocalization in several neurons across all WT samples (Fig 7A). Primary cerebellar neurons from NAP mice lacked MYO5A immunoreactivity as expected (Fig S5). Co-immunoprecipitation (co-IP) was carried out using lysates from native P18 WT cerebellum. An anti-MYO5A antibody immunoprecipitated endogenous MYO5A. A weak interaction with ANKFY1 was detected consistent with the partial co-localization seen by immunofluorescence microscopy (IFM) (Fig 7B Top). IFM of primary neuronal cells showed a similar co-localization with ANKFY1 and RAB5A (Fig 7C). There was colocalization in several neurons within samples as well as across all WT samples (n=3 per group). Co-IP using an antibody to RAB5A demonstrated a weak interaction with ANKFY1, consistent with the observed partial co-localization (Fig 7D Top, Fig S6).

Figure 7. MYO5A and ANKFY1 colocalize and interact, ANKFY1 and RAB5A colocalize and interact and loss of MYO5A associates with longer mitochondria.

(A) Primary neurons from the cerebellum of P4-P6 WT mice were isolated and stained for MYO5A and ANKFY1 and showed colocalized puncta (n=3 WT animals). (B) Co-immunoprecipitation (Co-IP) with an anti-MYO5A antibody from P18 WT cerebellar lysates detected a faint band when probed for ANKFY1. The bottom image depicts the membrane probed for MYO5A (n=2). (C) Primary cerebellar neurons were stained for ANKFY1 and RAB5A and showed colocalized puncta (n=3 WT animals). (D) (Top) Co-IP using a RAB5A antibody in P18 WT cerebellar lysates detected a faint band when probed for ANKFY1. The membrane was stripped and probed for RAB5A on the bottom image. (E) TMT-MS showed mitofusin-1 (MFN1) was increased (*P=0.014, n=4 per group). (F) Primary cerebellar neurons were isolated from P4-P6 WT and NAP mice (n=3 per group) and stained with Mitotracker. (G) The NAP neurons showed an increase in the overall length of the mitochondria within neuronal processes (****P<0.0001, n=3 per group). Total mitochondria measured in each group: n=289 WT, n=321 NAP. All scale bars=5μm. All comparisons used unpaired parametric two tailed t-test. Data represent the mean±SD.

An interaction between early endosomes and mitochondria is thought to be vital for normal organelle function (66–68). Furthermore, reducing Ankfy1 results in elongated mitochondria (69), and there is evidence to support that unconventional motors such as MYO5A can alter mitochondrial function (38). We observed an increase in mitofusin 1 (MFN1) protein by TMT-MS in NAP cerebella (P=0.014) (Fig 7E), corroborated by a similar trend by immunoblotting for MFN1 (Fig S7). Mitochondrial length was significantly greater in NAP neurons compared to WT (P<0.0001) (Fig 7G). In analyzing the GO term Macroautophagy, a process that includes mitophagy, we observed a significant reduction in the TMT-MS data in NAP cerebella compared to WT (Fig S8). Together, these data support a MYO5A:ANKFY1 interaction, which may affect early endosomal function, and that loss of MYO5A affects mitochondrial morphology.

DISCUSSION

A spontaneous mutation in Myo5a produces an ataxia phenotype

Reduced or defective MYO5A is thought to underlie most alleles of the dilute lethal series, linking this locus to ataxia and coat color defects. Like others in this allelic series, the limited lifespan of NAP mice, where death typically occurred by 18–21 days of life, suggested a critical role in early post-natal development. From 0–14 days of life, NAP mice appeared similar to littermate controls, and the profound loss of motor function occurred over a relatively short window of 7–9 days. The NAP mouse arose on the MRL background, which is described for its role in modulating enhanced wound healing in many different forms of injury (5, 8, 70). Because NAP mice die before weaning, the MRL background is not dramatically extending lifespan (56, 71). Most spontaneous dilute lethal mice were discovered because of their coat color defects. Had the Myo5a (c.1752g>a) arose on the 129 background the dilute phenotype would presumably be present, however at the 5^th generation of interbreeding, the NAP mice were genetically ~97% of the albino, MRL background, thus masking the potential dilute phenotype.

The excision of exon 14 by Myo5a (c.1752g>a) resulted in an in-frame deletion, yet MYO5A protein was greatly reduced as to be undetectable by immunoblotting. The TMT-MS data showed very minimal reporter ion intensity from peptides mapping to Myo5a in NAP mice, consistent with greatly reduced protein. The identified peptides were sparse and distributed throughout the protein, both N- and C-terminal to the position disrupted by loss of exon 14. Similar spontaneous phenotypes in rats due to Myo5a 5’ splice variant in exon 4 or the 141 in-frame base pair deletion in the head region of Myo5a also resulted in undetectable MYO5A protein expression (72, 73). These findings are consistent with internal truncations within MYO5A’s N-terminal head region producing an unstable protein.

The effect of Myo5a loss on the cerebellum

In quantifying the number of Purkinje cells present in NAP mice, we observed comparable abundance between mutant and WT controls, which is consistent with some previous reports (73). However, other studies describe Purkinje cell loss in the dilute lethal model (74). We note that Sawada et al. evaluated dilute lethal mice at P21, which is three days beyond our analytical time point. Additionally, their analysis was in the C57BL/6 strain versus the MRL/MpJ strain that was used here, and either of these features may contribute to differences. Furthermore, at P18, we saw no difference in the overall number of cells present within the molecular layer. In assessing granule cells by Pax6 staining, we did observe an increase in granule cells in the EGL of NAP mice at P18. Multiple genes including Pax8, Atoh1, Neurod1, Cntn2, Girk2 and Sema6a have been implicated granule cell development (75, 76). While it is possible that MYO5A interacts with a critical protein required for granule cell development, another possibility stems from the loss of MYO5A resulting in hypothyroidism in rats. Hypothyroid rats and mice have showed delayed granule cell migration and maturation (59, 77), although the mechanism by which reduction in thyroid hormone affects granule cell development may be developmental stage specific through either cell intrinsic or extrinsic properties.

Sloane and Vartanian showed a reduction in myelin proteins at P15 in Myo5a null mice (35). We observed no differences in the abundance of proteins implicated in myelination at P18 (data not shown). Both the day of analysis and strain must be noted as we assessed 3 days past Sloane and Vartanian’s study and their dilute lethal model was on a C57BL/Gr background. Our results do not preclude the possibility that myelin is produced but delayed in development or that myelin is produced but unable to localize and compact for proper function. Nonetheless, we observed a severe ataxic phenotype in presence of normal myelination, suggesting myelination was not a driver of the ataxia findings.

Proteomic insights from MYO5A loss

TMT-MS analysis identified over 250 differentially abundant proteins in NAP mice compared to WT controls, several of which corroborate known MYO5A functions, most notably in synaptic transmission. Nearly a quarter of the 164 significantly reduced proteins were represented by the GO term Synapse (GO:0045202). Interestingly, the presynaptic protein SV2A was significantly reduced, which is target for the anti-epileptic drug, levetiracetam (78). Additionally, the list included PPP3CA which is implicated in epileptic encephalopathy (79) and APBA1, which is implicated in Alzheimer’s disease (80). There is well-established literature linking synaptic defects to neuropathology (43, 81), and the loss of MYO5A likely affects synapse function through a broad range of aberrant signaling. Myosins are known to have redundancy, and we observed a trend toward increased MYH9, and MYH9 was recently linked to mitochondrial aggregation and interaction with the ER (82, 83).

The MS analysis showed a reduction in the abundance of RAB5A, and two RAB5A effectors, APPL1 and ANKFY1, as well as a decrease in the exchange factor RABGEF1 that activates GTP-RAB5A. In bone marrow dendritic cells from dilute lethal mice, perinuclear accumulation of endosomal vesicles was seen (84), suggesting a “tethering” role for MYO5A. Early endosomes, marked by RAB5A, are sorted for recycling back to the plasma membrane, the endolysosomal system for clearance or to the trans Golgi network (85–87). The reduction of critical endosomal proteins supports involvement of MYO5A in this system. We provide evidence for an interaction between MYO5A and ANKFY1, which may be transient, and because ANKFY1 is a known effector for RAB5A, this may implicate MYO5A with endosomes. Further work is needed to solidify this claim.

Mitochondrial Elongation

The NAP mice showed increase levels of the mitofusin 1 (MFN1) protein as detected by TMT-MS, which is involved in mediating mitochondrial fusion, a process that counteracts mitochondrial fission (88). We also observed an increase in the length of mitochondria in NAP primary cerebellar neuronal processes. Differences in mitochondrial morphology can result from a variety of cellular responses, such as macroautophagy, where mitochondria elongate as a protective mechanism against degradation (89). How a loss of MYO5A results in elongated mitochondria is not known. Mitochondria endoplasmic reticulum contact sites, referred to as MERCs (90, 91), are likely critical to mitochondrial homeostasis, and there is a role for MYO19 in mitochondrial movement and MERCs (92, 93). Given MYO5A’s role in ER movement, in Purkinje cell dendritic spines MYO5A could contributes to MERC’s, but additional investigation is needed to support this concept. Additionally, there is evidence that blocking nonmuscle myosin II with blebbistatin or siRNA-mediated reduction inhibits mitochondrial fission. The nonmuscle myosin II contribution to mitochondrial fission is thought to act upstream of the dynamin related protein 1 (Drp1) which is has well-known role in mitochondria fission (94, 95). Lastly, there is a growing literature for endosomal proteins affecting mitochondrial fission (69), specifically contact sites between endosomes and mitochondria have been described (96). These contact sites implicate mitophagy and requires mitochondrial fission to “pinch-off” portioned mitochondria via Drp1 (67, 97). Mitochondria fragments are then transported to Rab5a-containing endosomes for clearance (67). In addition to RAB5A endosomes affecting mitochondrial fission, other endosomal proteins implicate fission. These include: Vps35 which is a core component of the retromer, EHD1 and Ankfy1 (69, 98, 99). These data are consistent with a potential MYO5A-endosomal protein interaction possibly affecting mitochondrial homeostasis.

Summary Statement:

A recessive ataxic phenotype spontaneously arose caused by a de novo Myo5a splice variant affecting granule cell development and proteins involved in short range organelle trafficking and mitochondrial fission.

DATA AVAILABILITY

Primary data is available from the authors. The proteomic data is deposited under MassIVE MSV000095598.