The secreted MSP domain of C. elegans VAPB homolog VPR-1 patterns the adult striated muscle mitochondrial reticulum via SMN-1

The major sperm protein domain (MSPd) has an extracellular signaling function implicated in amyotrophic lateral sclerosis. Secreted MSPds derived from the C. elegans VAPB homolog VPR-1 promote mitochondrial localization to actin-rich I-bands in body wall muscle. Here we show that the nervous system and germ line are key MSPd secretion tissues. MSPd signals are transduced through the CLR-1 Lar-like tyrosine phosphatase receptor. We show that CLR-1 is expressed throughout the muscle plasma membrane, where it is accessible to MSPd within the pseudocoelomic fluid. MSPd signaling is sufficient to remodel the muscle mitochondrial reticulum during adulthood. An RNAi suppressor screen identified survival of motor neuron 1 (SMN-1) as a downstream effector. SMN-1 acts in muscle, where it colocalizes at myofilaments with ARX-2, a component of the Arp2/3 actin-nucleation complex. Genetic studies suggest that SMN-1 promotes Arp2/3 activity important for localizing mitochondria to I-bands. Our results support the model that VAPB homologs are circulating hormones that pattern the striated muscle mitochondrial reticulum. This function is crucial in adults and requires SMN-1 in muscle, likely independent of its role in pre-mRNA splicing. Highlighted Article: Secreted MSPds promote the localization of mitochondria within the body wall muscle during development, with implications for ALS. See also the companion paper by Cottee et al.


INTRODUCTION
VAMP/synaptobrevin-associated proteins (VAPs) comprise an evolutionarily conserved protein family with an N-terminal major sperm protein domain (MSPd), coiled-coil motif, and transmembrane region (Lev et al., 2008). They are synthesized as type II integral membrane proteins with the MSPd residing in the cytosol. The MSPd is named after nematode major sperm proteins, which function as sperm cytoskeletal elements and secreted signaling molecules (Bottino et al., 2002;Miller et al., 2001;Roberts and Stewart, 2012). In animals, VAPs have two diverse biochemical functions. VAPs act in a cell-autonomous fashion as scaffolding components at intracellular membrane contact sites (Lev et al., 2008;Stefan et al., 2011). In this capacity, there is evidence for roles in lipid transport, Ca 2+ homeostasis, the unfolded protein response and other processes. VAPs also have a non-cellautonomous signaling function (Han et al., 2012(Han et al., , 2013Tsuda et al., 2008). In neurons and other cells, the MSPd is liberated from the transmembrane domain and unconventionally secreted into the extracellular environment. The MSPd signals through Eph and Larlike receptors that modulate the actin cytoskeleton. An important MSPd target is striated muscle, where signaling regulates mitochondrial morphology and localization (Han et al., 2012(Han et al., , 2013. Humans have two VAP paralogs called VAPA and VAPB, which have broad, largely overlapping expression patterns (Gkogkas et al., 2008;Larroquette et al., 2015;Lev et al., 2008). A P56S substitution in the VAPB MSPd is associated with amyotrophic lateral sclerosis (ALS) and spinal muscular atrophy (SMA) (Di et al., 2016;Nishimura et al., 2004). ALS clinical symptoms often emerge when the patient is in their fifties and are characterized by progressive muscle weakness, atrophy, and spasticity, resulting from degeneration of upper and lower motor neurons (Rowland, 1998). Most ALS cases occur sporadically without a clear family history. However, mutations in over 20 genes, including VAPB, are associated with familial ALS forms (Peters et al., 2015).
The MSPd P56S mutation also causes a late-onset form of SMA, which is characterized by lower motor neuron degeneration (Burghes and Beattie, 2009). Although SMA is more common in infants and children, rare adult-onset cases do occur (Nishimura et al., 2004;Tiziano et al., 2013). Reduced survival of motor neuron 1 (SMN-1) function causes ∼95% of all SMA cases (Burghes and Beattie, 2009;Lefebvre et al., 1995). SMN-1 is part of a protein complex that controls the assembly of small nuclear ribonucleoproteins (snRNPs) essential for pre-mRNA splicing (Fallini et al., 2012;Liu et al., 1997). It is not clear whether this function or an alternative function is crucial for SMA pathogenesis (Cauchi, 2010). For instance, SMN-1 localizes to myofilaments in Drosophila flight muscles, where it regulates actin dynamics (Rajendra et al., 2007).
Evidence is accumulating that MSPd signaling may be important in sporadic ALS cases. The VAP MSPd is found in human blood and cerebrospinal fluid (CSF), although its circulating function is not understood (Deidda et al., 2014;Tsuda et al., 2008). In an Italian cohort, a majority of sporadic ALS patients had undetectable VAPB MSPd levels in CSF (Deidda et al., 2014). The pathogenic P56S mutation prevents MSPd secretion in cultured cells and animal tissues (Han et al., 2012;Tsuda et al., 2008). EphA4, an ephrin receptor that also interacts with the VAPB MSPd (Lua et al., 2011;Tsuda et al., 2008), modifies pathogenesis in ALS patients and in a zebrafish model (Van Hoecke et al., 2012). Eph and Lar-related receptors are expressed in motor neurons and striated muscles. While both cell types are implicated in ALS, their respective roles are not well delineated (Dupuis et al., 2011;Turner et al., 2013;Zhou et al., 2010). Familial ALS patients carry the pathogenic mutation throughout their lives. Disease-causing mutant proteins tend to be expressed early and ubiquitously, potentially triggering secondary effects and compensatory mechanisms that mask the primary pathological event. Unfortunately, defining early pathogenic processes has proven challenging. A better understanding of MSPd function might provide insight into these processes.
C. elegans and Drosophila VAPs have an important signaling function that impacts striated muscle mitochondria (Han et al., 2012(Han et al., , 2013Tsuda et al., 2008). MSPd signaling to C. elegans body wall muscle remodels the actin cytoskeleton, thereby docking mitochondria to myofilaments, altering fission/fusion balance and promoting energy metabolism (Han et al., 2012). MSPd antagonizes signaling via the CLR-1 Lar-related tyrosine phosphatase receptor. Excess CLR-1 activity promotes actin filament formation in the muscle belly, displacing mitochondria from I-bands. In aging worms, muscle cytoskeletal or mitochondrial abnormalities induce elevated Forkhead Box O (FoxO) transcription factor activity (Han et al., 2013). FoxO promotes muscle triacylglycerol (TAG) accumulation, alters ATP metabolism, and extends lifespan, despite reduced mitochondria electron transport chain activity. Vapb knockout mice also exhibit abnormal muscular FoxO metabolic gene regulation (Han et al., 2013). These data support the model that the MSPd promotes striated muscle energy metabolism.
Here we use C. elegans to further investigate the VAP-related 1 (VPR-1) signaling mechanism. Our results support the model that neurons and germ cells secrete the MSPd into the pseudocoelom, where it acts on CLR-1 receptors expressed throughout the muscle plasma membrane. Although vpr-1 mutant muscle mitochondrial defects initiate early in larval development, MSPd-to-CLR-1 signaling is sufficient during the L4 stage and adulthood to localize mitochondria to I-bands. In a suppressor screen, we identified SMN-1 as a crucial MSPd downstream mediator in muscle, where it regulates mitochondrial morphology and localization. We propose that VAPB homologs have an evolutionarily conserved signaling function to pattern the mitochondrial reticulum in striated muscle. This signaling activity is essential during adulthood and requires SMN-1 in muscle.

Muscle mitochondrial defects in vpr-1 mutants emerge in larval development
In adult central body wall muscles, mitochondrial tubules lie in parallel arrays on top of (or beneath, depending on dorsal or ventral orientation) dense bodies along myofilament I-bands (Fig. 1A,B). Muscle mitochondria are visualized using a mitoGFP reporter expressed under the muscle-specific myo-3 promoter (Fig. 1B), as well as dyes such as Rhodamine 6G and MitoTracker CMXRos (Han et al., 2012(Han et al., , 2013. Mitochondria localize along actin-rich thin filaments (Fig. 1C), where they undergo fission and fusion with adjacent tubules (Han et al., 2012). Myofilaments appear normal in vpr-1 mutants, but ectopic Arp2/3-dependent actin network reorganization in the adult muscle belly displaces most mitochondria from I-bands (Han et al., 2012(Han et al., , 2013. To investigate the origin of vpr-1 mutant mitochondrial defects, we conducted a developmental timecourse starting at the L1 larval stage. Shortly after hatching, mitochondria in wild-type body wall muscle are predominantly peri-nuclear and extend thin tubules into the muscle cytoplasm or belly (Fig. 1D). As larvae develop, muscle mitochondria become progressively associated with I-bands, forming parallel arrays of mitochondrial tubules. By the early adult stage, most tubules are closely associated with I-bands ( Fig. 1D) (Han et al., 2012). Thus, mitochondria in muscle develop their stereotypical positioning during larval development.
At hatching, muscle mitochondria in vpr-1(tm1411) mutants look similar to those in the wild type. The tm1411 allele is a molecular null mutation that deletes the translational start site and entire MSPd (Han et al., 2012(Han et al., , 2013Tsuda et al., 2008). vpr-1 mutant mitochondria are predominantly peri-nuclear with tubules branching into the cytoplasm (Fig. 1E). However, mitochondria in vpr-1 mutant muscles fail to target I-bands during larval development and adulthood. They largely remain in the muscle belly, forming branched networks as the muscle grows in size (Fig. 1E). These mitochondrial networks increase in complexity as the worm ages. Rhodamine 6G and MitoTracker CMXRos dyes stain vpr-1(tm1411) mutant muscle mitochondria well during early larval development. Staining is less efficient at the L4 and adult stages, perhaps reflecting changes in transmembrane potential or tubule architecture (Han et al., 2012). During adulthood, mitochondrial tubules are hyperfused and fat droplets accumulate in the belly (Han et al., 2012(Han et al., , 2013. Consistent with fission/fusion imbalance, transgenic lines expressing the fission mediator DRP-1:: mCherry and fusion mediator FZO-1::mCherry show abnormal localization in adult vpr-1 mutant muscle (Fig. S1). We conclude that the vpr-1 mutant muscle mitochondrial defects initiate in larval development, resulting in abnormal mitochondrial fission/fusion dynamics in adults.
In neurons, synapses have a high energy demand that depends on closely associated mitochondria (Hollenbeck and Saxton, 2005). To determine if vpr-1 loss affects neuronal mitochondrial localization, we generated wild-type and vpr-1(tm1411) transgenic lines expressing mitoGFP and the synaptic vesicle marker mCherry:: RAB-3 (Ding et al., 2007;Nonet et al., 1997) in motor neurons. There was no statistical difference between the two lines in the percentage of mitochondria associated with synapses ( Fig. S2). We observed a possible increase in mCherry::RAB-3 puncta size in vpr-1(tm1441) mutants, suggesting that synaptic size or vesicle density is increased. Therefore, vpr-1 loss causes mitochondrial localization defects in larval and adult muscles, but not in motor neurons. Whether vpr-1 loss causes functional or subtle trafficking defects in neuronal mitochondria is not clear with the present data.
The nervous system and germ line are major origins of VPR-1 signaling activity Previous experiments showed that driving vpr-1 cDNA expression pan-neuronally using the unc-119 promoter rescued ∼30-40% of the muscle mitochondrial defects in vpr-1(tm1411) mutants (Han et al., 2012). We found that vpr-1 genomic sequence, including the 3′ UTR, is more efficient than the cDNA with the unc-54 3′ UTR in rescuing the vpr-1 mutant gonadogenesis defect (Cottee et al., 2017). Transgenes containing vpr-1 genomic DNA driven by pan-neuronal (unc-119p), GABA motor neuron (unc-25p), cholinergic motor neuron (unc-17p), head interneuron (glr-5p) or sensory neuron (osm-6p) specific promoters rescue the vpr-1 mutant muscle mitochondrial defects in about half the muscles, with variation apparent among animals ( Fig. 2; data not shown). By contrast, the endogenous vpr-1 promoter and genomic locus provide complete rescue (Han et al., 2013). These results indicate that vpr-1 expression in diverse neuron classes is sufficient to promote muscle mitochondrial localization, but additional sources of VPR-1 might be involved.
Genetic mosaic analysis using the vpr-1 genomic locus indicates that vpr-1 expression is essential in the nervous system and germ line but not in muscle to promote muscle metabolism (Han et al., 2013). The germ line is a source of maternal mRNAs provided to the embryo, as well as zygotic gene expression in developing and adult gonads. The germ line and muscle can exchange signaling molecules and other factors through the pseudocoelom ( Fig. 2A). Maternal vpr-1 mRNA is sufficient to weakly rescue the vpr-1 (tm1411) mitochondrial defects in adult muscle (Fig. 2B). To investigate zygotic germline expression, we generated single-copy integrated transgenic lines that express vpr-1 under the pie-1 germline promoter (Seydoux and Dunn, 1997). Two independent integrated transgenes were crossed into the vpr-1(tm1411) background. Both transgenes completely rescued the muscle mitochondria ( Fig. 2B) and gonad development defects (Cottee et al., 2017). We used two strategies to evaluate zygotic germline vpr-1 expression (see Materials and Methods). These vpr-1 mutant hermaphrodites lack maternal vpr-1 mRNA and contain a single copy of the pie-1p::vpr-1 transgene provided in the paternal genome. In both experiments, the vpr-1 mutant muscle mitochondrial defects were rescued in about half the muscle, with variability among animals (Fig. 2B). Therefore, zygotic germline vpr-1 expression is sufficient to promote body wall muscle mitochondrial localization. In summary, the results of these genetic mosaic and transgenic expression studies are consistent with the nervous system and germ line acting together as major sources of VPR-1 MSPd signaling activity to body wall muscle.
expression pattern should provide clues as to where and when MSPd signal transduction initiates. To determine endogenous CLR-1 expression, we used Cas9 to fuse tdTomato to the clr-1 genomic locus, creating a C-terminal fusion protein (Fig. 3A). Reduced clr-1 function in the hypodermis causes fluid to accumulate throughout the pseudocoelom, arresting development and causing gonad degeneration (Huang and Stern, 2004;Kokel et al., 1998). tdTomato insertion did not disrupt CLR-1 function, as shown by growth to adulthood, fertility, and the absence of fluid accumulation. We observe endogenous CLR-1::tdTomato expression in muscle and a variety of other cell types, including somatic gonad, hypodermis and neurons ( Fig. 3B and Fig. S3). In larval and adult body wall muscle, CLR-1 is expressed throughout the plasma membrane, called the sarcolemma, and in puncta within the muscle cytoplasm (Fig. 3B). We did not detect enrichment or absence at post-synaptic plasma membrane sites near the nerve cord, showing that CLR-1 is uniformly expressed. Cell surface CLR-1 expression did not significantly change in the vpr-1(tm1411) background (Fig. S4). Therefore, the CLR-1 extracellular domain is accessible to MSPd signals secreted from adjacent motor neurons or from more distant neurons and germ cells via the pseudocoelom.
In 1-day adult vpr-1 mutants, the muscle mitochondrial reticulum is strongly disrupted due to excess CLR-1 activity (Fig. 1E) (Han et al., 2012). To test whether this reticulum can be remodeled, we shifted 1-day adult worms to the restrictive temperature and observed their mitochondria 24, 48 and 72 h later. After 24 h, the mitochondria are still largely disorganized and surround fat droplets in the muscle belly (Fig. 4D). Mitochondria disorganization and increased fat droplets are both due to abnormal Arp2/3 activity (Han et al., 2012(Han et al., , 2013. Shifting for 24 h is sufficient to induce fluid accumulation in nontransgenic mutants, indicating that CLR-1 function is compromised. At 48 h, most mitochondrial tubules are aligned at myofilaments, although morphology is still abnormal (Fig. 4E). By 72 h, mitochondria are positioned correctly at I-bands with morphology similar to those in wild-type muscle (Fig. 4F). Very few fat droplets are observed in the cytoplasm. These data suggest that MSPd signals continuously instruct muscle mitochondria to remodel via an active process throughout adulthood.
A vpr-1 mutant mitochondrial suppressor screen identifies smn-1 We next sought to understand how MSPd signals are transduced in muscle. To identify downstream mediators, we developed a vpr-1 (tm1411) RNAi suppressor screen based on prior work using arx-2 and clr-1. arx-2 (also known as arp-2) encodes a component of the Arp2/3 complex (Roh-Johnson and Goldstein, 2009). Arp2/3 loss in vpr-1 mutants suppresses the muscle mitochondrial defects (Han et al., 2012). Transgenic lines that express functional ARX-2:: mCherry (see below) in wild-type muscle show punctate localization between thin filaments, slightly above mitochondrial tubules (Fig. 5A,B). Little ARX-2::mCherry is found in the muscle belly (Fig. 5B). In vpr-1 mutants, ARX-2::mCherry is observed at myofilaments and throughout the belly (Fig. 5B), where actin filaments and most mitochondria exist. These data support the model that excess CLR-1 signaling promotes Arp2/3 activity and/or localization in the muscle cytoplasm, preventing mitochondria from targeting I-bands. Disrupting this pathway suppresses the vpr-1 (tm1411) muscle mitochondrial defects, which is likely to be because a redundant mechanism localizes mitochondria to I-bands.
Our suppressor screen should identify those gene products specifically required for abnormal mitochondrial localization caused by MSPd deficiency. As a pilot investigation, we tested 31 RNAi clones corresponding to C. elegans homologs of genes implicated in ALS and SMA (Table S1). Three potential suppressors were identified: the nuclear export receptor xpo-1, the Gemin3 homolog mel-48, and the survival of motor neuron 1 gene smn-1. In this paper, we focus on smn-1. Similar to inactivation of clr-1 or arx-2 ( Fig. 4D and Fig. 5C), smn-1 RNAi largely suppresses the muscle mitochondrial defects in vpr-1 mutant animals (Fig. 5C). smn-1 RNAi initiated in the parental generation often causes larval arrest in vpr-1 mutant progeny and muscle mitochondria to localize to I-bands with more globular morphology. smn-1 RNAi initiated in L2-L3 larva is sufficient to partially suppress the vpr-1 mutant mitochondrial defect, suggesting that smn-1 activity is essential Fig. 4. Temporally controlled MSPd activity in vpr-1 null mutants. The clr-1 (e1745ts) temperature-sensitive mutation was used to inactivate CLR-1 (Kokel et al., 1998), thereby simulating MSPd signaling. The rol-6p::clr-1 transgene drives clr-1 expression specifically in the hypodermis to alleviate fluid accumulation (Huang and Stern, 2004). myo-3p::mitoGFP was used to visualize muscle mitochondria. The diagram summarizes the restrictive temperature shift initiation period according to developmental stage. E, embryonic stage. (A) vpr-1(tm1411);clr-1(e1745ts) adult worm grown at the permissive temperature. Mitochondrial tubules are found throughout the muscle belly, along with fat droplets (arrows) (Han et al., 2013). In some animals, the mitochondria appear fragmented. (B) vpr-1(tm1411);clr-1 (e1745ts) adult worm shifted to restrictive temperature as embryos. The mitochondria largely align at I-bands, but are often fragmented and poorly organized. Few fat droplets are seen. Similar results are observed when shifting at L1 and L2 stages. (C) vpr-1(tm1411);clr-1(e1745ts) adult worm shifted to restrictive temperature for 3 days starting as an L4. (D-F) Timecourse showing vpr-1(tm1411);clr-1(e1745ts) adult worm shifted to the restrictive temperature for (D) 24 h, (E) 48 h and (F) 72 h. Asterisk, nucleus. Scale bar: 10 µm. during late larval development and adulthood. smn-1 RNAi also partially suppresses vpr-1 null mutant muscle fat droplet accumulation, as measured using fluorescent BODIPY fatty acid analogs (Fig. S6). We did not observe mitochondrial or fat droplet suppression following RNAi of C. elegans homologs of Gemin2, Gemin6, Gemin7 or other genes crucial for pre-mRNA splicing (Table S1 and Fig. S7). Hence, smn-1 function may be independent of its role in spliceosome assembly.
To test whether smn-1 is necessary in muscle, we specifically depleted smn-1 in body wall muscle of wild-type animals using an RNAi mosaic strategy (Durieux et al., 2011;Esposito et al., 2007). sid-1 (systemic RNA interference deficiency-1) mutants are defective for siRNA transport between cells, thereby preventing systemic RNAi effects. However, producing siRNA within tissues still induces cell-autonomous RNAi (Winston et al., 2002). We expressed smn-1 sense and antisense RNAs in body wall muscle of sid-1( pk3321) mutants. Muscle smn-1 RNAi causes mitochondrial morphological defects similar to those seen in smn-1 mutants or systemic RNAi hermaphrodites (Fig. 6A,C). smn-1 reduction of function suppresses the vpr-1 null muscle mitochondrial defects, whereas smn-1 loss throughout development causes more globular mitochondrial morphology. In summary, a vpr-1 mitochondrial suppressor screen identified smn-1, which is necessary and sufficient in muscle to control mitochondrial morphology.

DISCUSSION
The results presented here, together with those in prior studies, suggest that VAPB proteins have a noncanonical function as an endocrine factor to pattern the striated muscle mitochondrial reticulum. The primary mechanism hinges on a secreted VAP proteolytic fragment, the MSPd (Tsuda et al., 2008). In C. elegans, the secreted MSPd antagonizes the CLR-1 Lar-like phosphatase receptor, triggering changes in muscle mitochondrial localization, fission/fusion and function (Han et al., 2012). Disrupting MSPd signaling causes energy deficit in adult muscle, along with compensatory metabolic changes during aging (Han et al., 2013). Here we show that muscle mitochondria target myofilament I-bands during late larval development and adulthood. This mitochondrial patterning mechanism fails in vpr-1 null mutants. Our results support the model that neurons and germ cells secrete the MSPd into the pseudocoelomic fluid, where it interacts with CLR-1 throughout the muscle plasma membrane. Downstream in the muscle cytoplasm, actin cytoskeletal reorganization events requiring the Arp2/3 complex and SMN-1 shift mitochondria from the cytoplasm to I-bands, where they form parallel arrays that fuse and divide. Continuous MSPd signaling during adulthood is required to maintain mitochondrial localization. Below, we further discuss the model, focusing on its developmental origin and implications for motor neuron diseases.
Genetic mosaic analysis demonstrates that vpr-1 loss in the nervous system or germ line, but not in muscle, causes the muscle metabolic abnormalities (Han et al., 2013). Expressing vpr-1 specifically in germ cells or neuron subsets is sufficient to rescue the vpr-1 mutant muscle and gonad phenotypes, although not in all animals. Expression from multiple cell types provides complete rescue. The data point to neurons and germ cells as primary MSPd secretion sites, which act collectively. Secreted MSPds are likely to enter the pseudocoelom, a primitive circulatory system that bathes the body wall muscles and other internal organs (Hall et al., 1999). We show that endogenous CLR-1 receptor is expressed throughout the muscle plasma membrane in larval and adult worms, where it is accessible to the pseudocoelomic fluid. These results are consistent with the MSPd having endocrine and paracrine signaling activity to muscle.
A temperature-sensitive clr-1 mutation (Kokel et al., 1998) was used to temporally induce MSPd signaling in vpr-1 null animals. MSPd signaling at the L4 stage or during adulthood is sufficient to localize mitochondrial tubules to muscle I-bands. Earlier inductions promote mitochondrial localization, but morphology is abnormal. It is possible that MSPd levels rise throughout larval development as germ cells increase in number, further antagonizing CLR-1 signaling. In wild-type animals, reducing muscle clr-1 function throughout larval development causes globular instead of tubular mitochondrial morphology (Han et al., 2012). A balance in CLR-1 signaling might be important early in development, whereas CLR-1 is largely inactive during adulthood. Muscle mitochondrial networks are already disorganized in young adult vpr-1 mutants, yet CLR-1 inactivation at this time induces remodeling events that target mitochondria to I-bands within 48 to 72 h. Therefore, MSPd activity on muscle appears to be instructive.
In muscle, CLR-1 activity triggers actin remodeling in the muscle belly that is dependent on the Arp2/3 complex (Han et al., 2012). These actin networks prevent mitochondria from associating with actin-rich I-bands, where ATP and Ca 2+ levels fluctuate (Moerman and Williams, 2006). The MSPd attenuates CLR-1 signaling, thereby restricting Arp2/3 activity to the I-band. Mitochondria then align along the I-band, alter their fission/fusion properties, and alter function (Han et al., 2013). Whether Arp2/3 restriction occurs by controlling localization or actin nucleation activity is not clear. We identified smn-1 in an RNAi screen for vpr-1 mutant mitochondrial suppressors. SMN-1 acts in muscle, where it colocalizes with the Arp2/3 complex. Genetic studies are consistent with SMN-1 promoting Arp2/3 activity. This function is sensitive to gene dosage, suggesting that the SMN-1 expression level is important. It might also require the Gemin3 homolog MEL-46, but appears independent of other gemins, SM proteins, and other proteins involved in snRNP assembly. Prior studies have implicated SMN-1 in mRNA transport and actin remodeling independent of its snRNP role, although the mechanism is not well understood (Burghes and Beattie, 2009;Fallini et al., 2012;Rajendra et al., 2007).
In mammalian fast-twitch muscle fibers, mitochondrial doublets encircle myofibers at I-bands, where they couple to the sarcotubular system (Boncompagni et al., 2009;Pham et al., 2012). Similar to C. elegans, mitochondria acquire this I-band positioning during postnatal development (Boncompagni et al., 2009). These data raise an intriguing model. Reduced VAPB or SMN-1 function might perturb a mitochondrial transition in the postnatal neuromuscular system. Motor neurons innervating fast-twitch fibers are the first to degenerate in ALS mouse models (Van Hoecke et al., 2012;Vinsant et al., 2013a,b). Either muscle mitochondrial dysfunction or compensatory mechanisms could predispose these motor neurons to degeneration during aging. An important implication is that motor neuron disease may initiate many years before clinical symptoms emerge. Therefore, a much larger window could exist for therapeutic interventions and biomarker development to track disease progression.

Statistical tests
Two-tailed Student's t-tests were computed using Excel 2013 (Microsoft) without the assumption of equal variance.

Imaging
Confocal images were taken with a Nikon 2000 U inverted microscope, fitted with a PerkinElmer UltraVIEW ERS 6FE-US spinning disk laser apparatus. Confocal images were processed with ImageJ version 1.48 (NIH). All other worm images were taken by a motorized Zeiss Axioskop equipped with epifluorescence and AxioVision software version 4.8.
Muscle mitochondria were visualized using the myo-3p::mitoGFP transgene, Rhodamine 6G dye, or MitoTracker CMXRos dye (Han et al., 2012(Han et al., , 2013). An advantage of the transgene is that muscle mitochondria are specifically labeled, but a disadvantage is that mitoGFP overexpression can cause abnormal mitochondrial morphology and location. The dyes do not appreciably affect mitochondrial morphology or location, although they stain mitochondria in most cells.

RNA-mediated interference (RNAi)
RNAi was performed by the feeding method (Timmons and Fire, 1998). HT115(DE3) bacterial feeding strains were obtained from the genome-wide library (Kamath et al., 2003). PCR and sequencing (UAB Heflin Center for Genomics Sciences) were used to confirm that strains contained the correct clones. RNAi phenotypes were compared with those of null mutants to determine effectiveness.