SMN function is essential for proper recovery after exhaustion
The C. elegans genome contains a single ortholog of SMN, encoded by the smn-1 gene. The severe loss of function allele, smn-1(ok355), has impaired neuromuscular function, and animals usually die during larval development, despite the fact that premature motor neuron death is not observed [31, 32]. A less severe allele, smn-1(cb131), is viable, fertile, and has only subtle defects. smn-1(cb131) is a D27N amino acid substitution, analogous to a D44V allele found in a type III SMA patient [33, 34]. The locomotion of smn-1(cb131) animals is normal on solid surfaces [35], and we observed no locomotion defects in animals swimming in liquid (assessed as body bends per minute in developmental stage matched animals, Fig. 1b pre-exposure time point, p = 0.746). However, smn-1(cb131) animals are resistant to immobilization by pyridostigmine bromide, an acetylcholinesterase inhibitor, suggesting latent defects in cholinergic neuromuscular junction (NMJ) synapse function [35].
Liewald and colleagues showed that channelrhodopsin (ChR2) stimulation of C. elegans cholinergic motor neurons depletes synaptic vesicle pools and results in transiently decreased locomotion post-stimulation [36, 37]. Using this paradigm (Fig. 1a and detailed description in the “Methods” section), we challenged NMJ function in smn-1(cb131) animals. As a result, cholinergic motor neuron stimulation in smn-1(cb131) animals revealed a locomotion defect during recovery after exhaustion, only in the presence of retinal. Before the blue light activation of cholinergic ChR2, locomotion rates of smn-1(cb131) and control animals were indistinguishable. After ChR2 activation for 20 s, locomotion rates decreased in both smn-1(cb131) and control animals. However, the post-exhaustion decrease was more profound in smn-1(cb131) animals (locomotion rates were 58% of controls, Fig. 1b, p = 0.001). smn-1(cb131) locomotion was restored to control levels 20 min after stimulation (Fig. 1b, p = 0.869), arguing against neuron or synapse damage caused by ChR2 activation and more likely a delayed recovery after exhaustion in these animals. Thus, we conclude that a latent defect in recovery after exhaustion is revealed in smn-1(cb131) by ChR2 stimulation of cholinergic neuron activity.
To test if smn-1(cb131) defects are caused by diminished SMN-1 levels, we undertook transgenic rescue studies. A single copy of the C. elegans smn-1 gene, including promoter, exons, introns, and 3′ untranslated sequences, was inserted into another chromosome by site-directed homologous recombination, creating rtSi10[smn-1p::smn-1] [8]. This transgene restored normal recovery rates after exhaustion in smn-1(cb131);ChR2 animals (smn-1(cb131);rtSi10 versus smn-1(cb131), Fig. 1c, p = 0.043). Therefore, diminished levels of SMN-1 impair recovery after exhaustion in smn-1(cb131) animals. Next, we used this ChR2 exhaustion assay to assess the impact of PLS3, a previously identified genetic modifier [12, 16].
Increased expression of plastin in neurons suppresses smn-1(cb131) locomotion defects
PLS3 is a protective modifier of SMA; increased PLS3 levels suppress defects in humans and SMA models in zebrafish, flies, and mice [9, 12, 16, 38,39,40]. We determined whether increasing PLS3 levels would ameliorate various functional defects in C. elegans SMA models.
First, we assessed locomotion after ChR2 exhaustion to determine if the introduction of human PLS3 could restore locomotion when SMN-1 function is decreased. We introduced two transgenes; a single-copy transgene (rtSi27[dpy-30p::PLS3]) or a multi-copy transgene (rtIs59 [dpy-30p::PLS3]) was crossed onto the smn-1(cb131);ChR2 background. In both cases, human PLS3 was expressed ubiquitously in somatic tissues using the dpy-30 promoter [41]. We found that PLS3 overexpression restored post-exhaustion locomotion rates to normal levels in smn-1(cb131) animals (after cholinergic ChR2 neuron stimulation, Fig. 2a, p = 0.002 and 0.003). Therefore, we conclude that expression of human PLS3 ameliorates the functional defects caused by decreased SMN-1 in this exhaustion paradigm.
To test for cross-species conservation of plastin function, we also examined the impact of overexpressing the C. elegans PLS3 ortholog, plst-1. Loss of plst-1 exacerbates smn-1 neuromuscular defects in other paradigms [39]. We assessed locomotion in smn-1 mutant animals that ubiquitously express C. elegans plst-1 mRNA at increased levels (rtEx850 [dpy-30p::plst-1]). The multi-copy plst-1 transgene restored normal locomotion rates after exhaustion in smn-1(cb131);ChR2 animals (Fig. 2a, p = 0.006). Combined with previous loss of function studies [39], our results are consistent with the hypothesis that Plastin 3 and its C. elegans ortholog are cross-species modifiers of locomotion defects in SMA models.
SMN is broadly expressed, but its function is required in motor neurons for normal NMJ activity, based on work in mammalian [42] and C. elegans SMA models [31]. Plastin 3 is also broadly expressed in neuronal and non-neuronal tissues [43]. To determine where PLS3 function is required in C. elegans SMA models, we expressed PLS3 in either body wall muscle or neurons of smn-1(cb131) animals. Expression of human PLS3 in neurons (rtEx852[unc-119p::PLS3]), but not in muscles (rtEx851[myo-3p::PLS3]), restored normal recovery of locomotion rates after exhaustion in smn-1(cb131);ChR2 animals (Fig. 2b, p = 0.05 and p = 0.237, versus no PLS3, respectively). This is consistent with studies of PLS3 and SMN in vertebrate models, where increasing PLS3 in neurons was also sufficient to ameliorate defects when SMN levels decreased [16].
Finally, we examined the impact of increased PLS3 activity in otherwise normal animals. A single-copy insertion of the ubiquitously expression dpy-30p::PLS3 transgene had no impact on basal locomotion; activity in liquid was indistinguishable from control animals. However, further increase in PLS3 dosage did alter locomotion. A transgene generated at higher concentrations of dpy-30p::PLS3, rtIs59, increased basal locomotion rates by 38% (Fig. 2c, p = 0.003). However, it is unclear whether the increased locomotion of these high copy PLS3 overexpression animals is pertinent to the impact of PLS3 in SMA models. We cannot rule out that a high dosage of PLS3 is detrimental because it has been shown that high levels of the analogous actin-bundling protein Sac6 cause lethality in yeast [44]. Therefore, we avoided use of this high copy PLS3 transgene in the rest of this study, unless explicitly noted.
Increasing plastin levels ameliorates NMJ and pharyngeal pumping defects caused by decreased smn-1 function
smn-1(cb131) animals are resistant to paralysis by the acetylcholinesterase inhibitor, pyridostigmine bromide, suggesting reduced cholinergic NMJ function [35]. We found that diminished SMN-1 function also resulted in decreased sensitivity to aldicarb, another acetylcholinesterase inhibitor. Prolonged exposure to aldicarb gradually induces paralysis, due to acetylcholine accumulation in the synaptic cleft. smn-1(cb131) animals paralyze more slowly on aldicarb than control animals (Fig. 2d, log rank p = 0.009). Introduction of the human PLS3 multi-copy transgene into smn-1(cb131) animals restored normal aldicarb sensitivity (Fig. 2d, p = 0.009 versus transgenic control lacking PLS3). These results indicate that ubiquitous PLS3 expression improves cholinergic NMJ function when SMN-1 is impaired.
Various studies show that the effects of PLS3 may depend on the levels of SMN present. In models where SMN levels are substantially decreased, increased PLS3 insufficiently suppresses neuromuscular defects [15, 16]. However, to test if the beneficial impact of PLS3 is not limited to smn-1(cb131) animals, we also tested smn-1(ok355) severe loss of function animals, which contain even lower levels of SMN-1 activity compared to smn-1(cb131). smn-1(ok355) animals have decreased pharyngeal pumping rates, a behavior frequently used to assess neuromuscular function. C. elegans feed on bacteria using a dedicated set of muscles and neurons in the pharynx. Pharyngeal pumping rates average roughly 250 pumps per minute in the presence of food; late larval stage smn-1(ok355) animals have dramatically reduced pharyngeal pumping rates [31, 39]. In control animals with a wild-type smn-1 gene, ubiquitous expression of human PLS3 in C. elegans slightly decreased pumping rates (single-copy transgene, rtSi27, − 11% versus wild type, Additional file 1: Fig. S1, p = 0.002). However, the introduction of the single-copy PLS3 transgene into smn-1(ok355) animals increased pharyngeal pumping rates by 50% (Additional file 1: Fig. S1, versus ok355 p = 0.011). Although pumping rates were not restored to wild-type levels, we conclude that in severe loss of function smn-1 animals, increased PLS3 partially ameliorates neuromuscular defects. Using two different alleles and testing multiple defects, an overexpression of PLS3 proves beneficial when SMN levels are decreased in C. elegans. This concurs with the partial amelioration observed in particular type IIIb SMA patients that carry two SMN2 copies and have elevated levels of PLS3 [12]. Taken together, our results demonstrate that plastin is a conserved genetic modifier of neuromuscular function.
Identification of sym-2 as a genetic modifier of both PLS3 and smn-1
PLS3, SMN, and actin can be co-precipitated from HEK293T cells and murine spinal cord, but SMN does not directly interact with PLS3. The identity of other proteins in this complex is unknown [12]. Delineating other proteins in the PLS3/SMN complex would lead to a better understanding of SMA and the protective mechanism engaged by increased PLS3 levels. To identify these other proteins, we undertook a small, targeted screen starting with results from a previous large-scale proteomic study undertaken in Drosophila S2 cells [40]. From those results, we assembled a list of proteins that pulled down either with Drosophila SMN or with Drosophila Fimbrin, the PLS3 ortholog. We prioritized proteins found in both lists, based on frequency of pull-down; protein similarity between D. melanogaster, H. sapiens, and C. elegans; and the availability of C. elegans loss of function (lf) alleles. From this, eleven C. elegans candidate genes were targeted for functional analysis (Fig. 3a). We hypothesized that proteins in the same protein complex as SMN and PLS3 might show functional interactions, and we interrogated these putative interactions by testing if C. elegans candidate gene loss of function suppressed PLS3- and SMN-associated defects. High copy PLS3 expression results in aberrantly high locomotion levels in liquid (Figs. 2c and 3b). We determined if candidate gene loss of function alleles could suppress locomotion changes caused by increased PLS3.We also determined if candidate gene loss of function could suppress the defect in smn-1(cb131) recovery from ChR2-induced exhaustion. Only sym-2 loss of function ameliorated defects in these assays (Fig. 3b, p = 0.007, and c, p = 0.006). RNAi knockdown of sym-2 also suppressed smn-1(cb131) ChR2-induced exhaustion recovery defects (Additional file 1: Fig. S2, p = 0.02, smn-1(cb131) control versus knockdown of sym-2), but knockdown of genes encoding two related C. elegans proteins had no impact (hrpf-1 and hrpf-2, Additional file 1: Fig. S2). Glorund was identified from the proteomic Drosophila screen, and BLAST results list the orthologs sym-2, hrpf-1, and hrpf-2 in C. elegans. BLAST results also show the human orthologs to glorund as hNRNP F, H1, H2, GRSF1, and ESRP 1 and 2. Based on predicted functional domains and expression patterns, the human proteins most closely related to glorund and SYM-2 are hnRNP F, hnRNP H1, and hnRNP H2 [46]. Members of this heterogeneous nuclear ribonucleoprotein family have roles in pre-mRNA processing [42, 47, 48]. Given the results above, we chose to focus on sym-2 for the remainder of this study.
sym-2(mn617) is a Y163N missense allele that likely decreases SYM-2 function [46]. As described above, introduction of sym-2(mn617) into the PLS3 overexpression background returned locomotion rates to normal (Fig. 3b, p = 0.007). And, introduction of sym-2(mn617) ameliorated smn-1(cb131) post-exhaustion locomotion defects in double mutant animals (Fig. 3c smn-1(cb131); sym-2(mn617) versus smn-1(cb131), p = 0.006). Taken together, these results demonstrate that decreased sym-2 function suppresses defects caused by decreased smn-1 activity or increased PLS3 function.
PLS3 colocalizes with SMN and SYM-2 orthologs in vertebrate neurons
SMN-1, PLS3, and SYM-2 may act in a common functional pathway, but C. elegans genetic studies cannot determine if these proteins act in a conserved protein complex in mammalian neurons. Previously, SMN protein was found in RNP granules with various hnRNPs in neuronal processes, but not with hnRNP F/H orthologs [48,49,50]. While the majority of studies focus on the nuclear role in splicing of hnRNP F/H, one cannot exclude a possible role outside the nucleus [51, 52]. Therefore, a series of immunohistochemical assays were done to look for an association of hnRNP F/H with PLS3 and SMN that might be pertinent in cells and occur outside the nucleus.
First, we examined localization of PLS3, SMN, hnRNP F, and hnRNP H1/2 in murine fibroblasts. Primary cultures were generated from mice overexpressing a V5-tagged PLS3 [16]. As expected, all four proteins were detected in fibroblast nuclei using immunohistochemistry. Interestingly, all four proteins are also present within cytoplasmic filopodia, suggesting they function in the same cellular compartment (Fig. 4a–d).
Next, we examined localization of hnRNP F, hnRNP H1/2, PLS3, and SMN in the processes of cultured mouse motor neurons. Embryonic stem cells were used to generate neuronal cultures from mice expressing motor neuron GFP (under the control of the Hb9 promoter). After 13 days of growth in culture, endogenous SMN, PLS3, hnRNP F, and hnRNP H1/2 proteins were detected within neuronal processes using immunohistochemistry. Each protein was distributed within motor neuron processes. To quantitatively assess protein localization within these structures, we used semi-automated image analysis software (Columbus, Perkin Elmer). By defining protein presence based on size, this analysis showed that SMN independently localized to the same region with all three proteins of interest in these structures (Fig. 4e–g), allowing us to conclude that hnRNP F and H1/2 are found in fibroblasts and in neuronal processes of motor neurons and that these proteins frequently localize to the same region with SMN. Collectively, these cellular data suggest that an association may happen between SMN, PLS3, hnRNP F, and H1/2 in vivo.
SMN, PLS3, and SYM-2 orthologs co-exist in vertebrate protein complexes
The localization analysis suggests that these proteins might be found in a similar cellular compartment and therefore could be in a complex together, a question we addressed using co-immunoprecipitation studies. HEK293T cells were transfected with tagged versions of all four proteins. Immunoprecipitation of GFP-tagged hnRNP F or hnRNP H1/2 revealed that these proteins are associated with both SMN and PLS3 (Fig. 5a). The interaction was not RNA-dependent, as associations were maintained after RNase treatment (Additional file 1: Fig. S3). These results are consistent with the localization analysis that show an interaction between the proteins.
Co-immunoprecipitation from mouse brain lysates confirmed an endogenous interaction between SMN and PLS3, hnRNP F, or hnRNP H1/2 (Fig. 5b bracket) in neuronal tissues. Reciprocally, endogenous PLS3 co-immunoprecipitated with SMN, hnRNP F, and hnRNP H1/2 (Fig. 5c). We conclude that SMN and PLS3 are in protein complexes that also contain hnRNP F and/or hnRNP H1/2 in the mouse brain. However, from this evidence alone, one cannot conclude that hnRNP F or hnRNP H1/2 are in a complex simultaneously containing both SMN and PLS3.
To test if these proteins are found in a complex together, we undertook size fractionation of whole cell lysates using FPLC columns and HEK293T cells expressing tagged versions of PLS3 and hnRNP F. In complexes of less than 669 kDa and greater than 158 kDa, four proteins—PLS3, SMN, actin, and hnRNP F—were found (Fig. 5d). And, in the smaller complexes within this range, hnRNP H1/2 also were found. Note that most of the PLS3, actin, and hnRNP F/H1/2 proteins are found in even smaller complexes, consistent with previously described interactions of these proteins. Taken together, size fractionation, co-immunoprecipitation, and localization results suggest that a rare previously unidentified complex may exist that contains actin, SMN, and PLS3 as well as hnRNP F and/or hnRNP H1/2.
SYM-2 loss rescues smn-1 endocytosis defects
Fluid-phase endocytosis is perturbed in C. elegans with decreased smn-1 function [8]. For mechanistic insight in how sym-2 loss suppresses smn-1 defects, we used a classical C. elegans assay which assesses endocytosis in coelomocyte cells. Six coelomocyte cells are found in the body cavity (pseudocoelom) of adult C. elegans. These cells play a critical role in the active removal of proteins by endocytosis [53, 54]. When muscle cells in transgenic animals secrete soluble GFP into the pseudocoelom, GFP is taken up by coelomocytes through fluid-phase endocytosis and is sent for degradation. This assay has been used extensively to characterize endocytic pathways [55]. RNAi knockdown of smn-1 in coelomocytes impairs GFP uptake and GFP clearance from the body cavity, consistent with endocytic pathway defects [8]. We found that either PLS3 overexpression or coelomocyte-specific RNAi knockdown of sym-2 ameliorated endocytic defects in smn-1(RNAi) animals. Overexpression of PLS3 reduced the number of animals with GFP accumulation in the pseudocoelom (Fig. 6, p = 0.039), consistent with the previously demonstrated role of PLS3 in endocytic pathways [9]. Moreover, RNAi knockdown of sym-2 dramatically improved GFP clearance from the pseudocoelom of smn-1(RNAi) mutant animals (Fig. 6, p = 0.1 × 10−5). These results demonstrate that the endocytic pathway defects observed when smn-1 levels drop are counteracted by knockdown of sym-2 in a cell autonomous manner. Combined with previous work, these results suggest that SMN, PLS3, and hnRNP F/H act in a protein complex, pertinent to SMA, whose function is required for normal function of endocytic pathways.
PLS3 expression suppresses defects in a C. elegans ALS model
Increasing PLS3 levels accelerates neurite outgrowth in vertebrate motor neurons, regardless of SMN status, suggesting that increased PLS3 might impact general function and survival of motor neurons [12]. With the notion that common mechanisms may underlie neurodegenerative diseases, increased PLS3 expression may modify defects associated with other neurodegenerative diseases besides SMA [16, 20, 56]. Based on this hypothesis, we determined if increasing PLS3 levels would modify defects in an established C. elegans model of ALS [30].
In the previously characterized C. elegans ALS model, wild-type human SOD1 (SOD1-WT) or mutant human SOD1 carrying the patient amino acid change G85R (SOD1-G85R) are expressed at high levels using a pan-neuronal C. elegans promoter (from the snb-1 gene). Mutant animals at young adult stage showed decreased locomotion in liquid, decreased pharyngeal pumping rates, and resistance to aldicarb, compared to their corresponding control animals (Additional file 1: Fig. S4, p = 6 × 10−17, p = 1.1 × 10−8, p = 6.6 × 10−15, p = 3.7 × 10−10, p = 4.7 × 10–4, p = 1.41 × 10−8) [30]. We introduced the single-copy PLS3 transgene rtSi27 into both SOD1-WT and SOD1-G85R to determine if these defects would be suppressed.
Expression of PLS3 in SOD1-G85R animals proved to be beneficial. Locomotion rates increased 61% compared to SOD1-G85R animals lacking PLS3 (Additional file 1: Fig. S4a, p = 3 × 10−6). Pharyngeal pumping defects were similarly ameliorated; PLS3 expression increased pumping rates by 35% in SOD1-G85R animals, restoring pumping to normal levels (versus SOD1-G85R without PLS3, Additional file 1: Fig. S4b, p = 4.1 × 10−8). Sensitivity to paralysis by aldicarb was also restored in SOD1-G85R animals expressing additional PLS3, compared to SOD1-G85R animals without PLS3 (Additional file 1: Fig. S4c, p = 0.002). Overexpression of PLS3 does not indiscriminately increase locomotion or pumping, nor does it always change aldicarb sensitivity. SOD1-WT animals expressing PLS3 showed decreased locomotion rates, compared to SOD1-WT animals without PLS3 (Additional file 1: Fig. S4a, p = 1.7 × 10−5). PLS3 had no impact on pharyngeal pumping rates or aldicarb sensitivity in SOD1-WT animals (Additional file 1: Fig. S4b and 4c). Overexpression of SOD1-WT results in motor dysfunction, specifically in locomotion assays [57]. While we observed this defect as well, it is unclear why overexpression of PLS3 is detrimental to SOD1-WT animals when assessing locomotion but not for other behaviors. However, due to the consistent beneficial effects in mutant SOD1-G85R animals, we conclude that increased plastin levels can suppress defects in a C. elegans model of ALS.
To determine if PLS3 is beneficial outside of neurodegenerative disease models, we examined the impact of overexpression on animals with decreased locomotion due to synaptic defects. The unc-25 gene encodes the C. elegans ortholog of mammalian GAD1 and is required for ɣ-aminobutyric acid (GABA) synthesis [58]. The behavioral defects of these animals include a 65% decrease in locomotion rates in liquid (Additional file 1: Fig. S5, p = 5 × 10−25 unc-25(e156) versus wild type). PLS3 expression did not increase locomotion rates (p = 0.576 versus unc-25). The unc-13 and unc-57 genes encode C. elegans orthologs of mammalian UNC13 proteins and Endophilin A protein, which are critical for presynaptic vesicular release and endocytosis, respectively. These mutant animals move in an uncoordinated manner compared to wild-type animals. PLS3 expression did not increase locomotion of animals lacking unc-13 function, but did increase locomotion in animals lacking unc-57 function (Additional file 1: Fig. S5, p = 0.02 unc-57 versus unc-57 with PLS3oe). Taken together, these results indicate that PLS3 overexpression is not generically beneficial and ameliorating endocytic pathways may improve function in C. elegans models of motor neuron disease.
SYM-2 loss also improves locomotion in a C. elegans ALS model
Considering that PLS3 overexpression suppressed defects in a C. elegans disease model of ALS and the relationship between PLS3 and SYM-2 described above, it seemed plausible that sym-2 might also be a modifier in an ALS model. We assessed the impact of sym-2 on locomotion in the C. elegans SOD1 overexpression model. Using RNAi to knockdown sym-2 levels, SOD1-G85R animals had increased locomotion rates with sym-2(RNAi) compared to empty vector RNAi feeding (Additional file 1: Fig. S4d, versus control (RNAi), p = 3.7 × 10−8). These results suggest that, reminiscent of PLS3, the beneficial impact of hnRNP F/H loss and endocytic pathway changes may extend to other motor neuron diseases.