Neuromuscular effects of G93A-SOD1 expression in zebrafish

Amyotrophic lateral sclerosis (ALS) is an adult-onset, relentlessly progressive, and ultimately lethal neuromuscular disease involving the degeneration and loss of motor neurons (MNs). Effective treatments for ALS have been elusive due to the fact that the mechanisms of MN loss in ALS are unknown. It is likely that multiple cell types within the MN microenvironment and multiple mechanisms contribute to ALS pathogenesis [1, 2]. While MN death is associated with caspase-mediated programmed cell death, neuromuscular junction (NMJ) defects and axonal degeneration are evident early in the disease course before symptom onset and MN loss [3, 4]. Understanding the causes of NMJ and axonal degeneration in ALS and identifying therapies that protect against these early insults represent the best strategy for preventing the irreversible loss of MNs.

The majority of ALS is sporadic with unknown etiology (90-95%); however, the remaining cases are familial (fALS) and have been linked to genetic mutations in genes such as Cu²⁺/Zn²⁺ superoxide dismutase (SOD1), Alsin (Als2), dynactin, angiogenin, senataxin, vesicle-associated membrane protein/synaptobrevin-associated membrane protein B, TAR DNA binding protein 43 (TDP43), fused in sarcoma (FUS), optineurin, and ubiquilin 2, or a hexanucleotide repeat expansion in C9ORF72[5‐18]. SOD1 mutations are associated with 10-20% of fALS cases and currently more than 150 missense mutations in SOD1 are reported, all of which are associated with a toxic gain of unknown function. Many in vitro and in vivo models expressing mutant SOD1 have been developed to study disease pathogenesis [2, 19]. G93A-SOD1 is the most thoroughly characterized ALS mutation and expression in primary MN cultures results in caspase activation and MN death [20, 21]. In vivo ALS models, including transgenic G93A-SOD1 mice, exhibit MN degeneration, muscle weakness and early lethality, resembling the human ALS disease course [22, 23]. G93A-SOD1 mice also show denervation at NMJs, defects in axonal transport, and axonal degeneration well before the onset of clinical symptoms, which supports a “dying-back” mechanism of MN degeneration [24, 25]. Thus, understanding and targeting the distal degeneration that occurs prior to the irreversible loss of MNs is imperative for developing effective therapies for ALS.

Recently, zebrafish have been gaining momentum as an emerging technology for the study of neurodegenerative diseases [26, 27]. Zebrafish are ideal for the development of novel strategies to understand ALS pathogenesis and screen potential therapies due to the fact that they develop quickly, have large numbers of offspring, are less expensive than rodent models, and they are easily amenable to genetic manipulation. Zebrafish neuromuscular physiology is also similar to humans, neuromuscular development is well characterized and rapid, and many tools exist to closely examine MN and muscle pathology in zebrafish. Furthermore, mechanistic insight into ALS pathogenesis could also be possible through combination with various tissue- and organelle-specific transgenic reporter zebrafish lines available [28‐31], and some neurodegenerative disease models have already been developed. Spinal muscular atrophy, a neurodegenerative disease involving MNs, has been successfully modeled in zebrafish [32‐34], and transient expression of select ALS mutations has also been reported. Transient overexpression of mutant SOD1 or mutant FUS induces an axonopathy phenotype at 30 hours post-fertilization (hpf) characterized by decreased axon length and aberrant branching [35, 36], as does knockdown of fus, als2, and elp3[35, 37, 38]. Similarly, knockdown of tdp43 or expression of mutant TDP43 in zebrafish embryos results in an axonal phenotype early in development, which is further evidenced in swimming deficits at 48 hpf [39]. While these results reflect an effect of mutant ALS proteins on developing MN axons, the transient expression in these models limits the amount of insight that can be gained regarding the mechanisms of progressive neurodegeneration that is characteristic of ALS.

To investigate ALS disease pathogenesis and progression in zebrafish, we generated novel stable transgenic zebrafish ubiquitously expressing G93A-SOD1. At the onset of this project, no transgenic ALS zebrafish models were established to our knowledge; however, in 2010 the Beattie laboratory reported the generation and characterization of transgenic zebrafish lines ubiquitously expressing zebrafish G93R-sod1[40]. This model exhibits motor deficits, MN degeneration and loss, and effects on survival [40] that are characteristic of ALS, validating our approach to generate and utilize the transgenic ALS zebrafish model described in this report. Our initial experiments verified the potential of human G93A-SOD1 to induce the previously reported motor axon defects, and we also demonstrated attenuation of these MN axonal defects with a known neuroprotective factor, insulin-like growth factor-I (IGF-I). We then established stable transgenic zebrafish lines expressing GFP-tagged human G93A-SOD1 using Tol2-mediated transgenesis. Characterization of our stable transgenic G93A-SOD1-GFP zebrafish revealed motor deficits that coincide with defects at the NMJ and subsequent alterations in neuromuscular innervation. We also observe MN loss later in the disease course, which supports a sequence of events that resemble human ALS pathogenesis. Overall, this report describes our detailed examination of MN axonal pathology, NMJ integrity, and MN loss in the context of disease onset and progression in a novel in vivo zebrafish ALS model that will be instrumental in future studies of ALS disease mechanisms and therapeutic development.

ALS is a relentlessly progressive neuromuscular disease with no effective treatments. While the cause of MN degeneration in ALS is unknown, multiple lines of evidence support a “dying-back” phenomenon that is characterized by defects at the NMJ and in distal MN axons which is then followed by subsequent MN loss [3, 4, 24, 42]. In vitro studies demonstrate protective effects of caspase inhibition and growth factor therapies on MN survival [21, 41]. However, studies in in vivo models indicate that treatments targeted at MN loss have limited, if any, impact on disease symptoms or survival [20, 43, 44] and that neuroprotective strategies targeted at muscle and NMJs such as growth factor therapies improve symptom onset and severity [24, 41, 45, 46]. Clinical translation of such strategies, however, has not yet been realized and improved treatments for ALS are required. Because defects in distal axons and at NMJs occur before symptom onset and MN loss in ALS, comprehending the mechanisms of ALS neuromuscular pathogenesis is essential for the development of therapies targeted at events prior to the irreversible loss of MNs. Detailed characterization of neuromuscular morphological alterations is possible in zebrafish; therefore, we investigated the effects of G93A-SOD1 expression on zebrafish MNs and NMJs.

In recent years, a handful of studies have demonstrated the consequences of transient expression or knockdown of select ALS mutant proteins in zebrafish embryos [35‐39]. Expression of mutant forms of SOD1, for example, induce early defects in MNs including decreased axon length and aberrant branching at 30 hpf, which are attenuated or exacerbated by vascular endothelial growth factor upregulation or knockdown, respectively [36]. Therefore, we first examined the effects of transient human G93A-SOD1 on zebrafish embryo MNs. Because RNA was generated from a pCS2-based construct known to elicit efficient high-level expression in zebrafish, we utilized a dose range that was between 10- to 100-fold lower than the previous report. As seen in Figure 1, our findings confirm the previous axonopathy findings, and further demonstrate that significant effects can be observed even with low levels of G93A-SOD1 RNA. We then examined the therapeutic efficacy of IGF-I upregulation by co-injecting igf-I RNA with G93A-SOD1 RNA and assessing axonal development. Despite the fact that clinical translation of IGF-I therapies for ALS have been slow and not yielded the expected outcomes, IGF-I has been extensively studied for ALS. In vitro and in vivo studies demonstrate that IGF-I activates neuroprotective signaling pathways and provides protection to MNs and at NMJs [41]. Along those lines, IGF-I is capable of preventing the decreased axon length observed with G93A-SOD1 expression in zebrafish embryos (Figure 1C). Together with the previous report [36], these data demonstrate an effect of low levels of human G93A-SOD1 on zebrafish MNs that is amenable to neuroprotective intervention, providing rationale for the generation of stable transgenic G93A-SOD1 zebrafish lines to examine neuromuscular pathogenesis in ALS.

Zebrafish are easily amenable to genetic manipulation and multiple approaches exist for the generation of stable transgenic disease models. The transgenic zebrafish lines described in this study (Figure 2) were developed using Tol2-mediated transgenesis. Briefly, constructs were generated by multisite Gateway cloning which provides a simple approach to generate Tol2-flanked constructs with various promoters and fusion elements that are amenable to random genomic insertion in the presence of transposase [47]. The CMV promoter we used is derived from the pCS2+ vector used in our transient studies, and a C-terminal GFP-fusion allows for easy detection of germline transmission and confirmation of cellular expression patterns. Confirmation that utilization of a GFP-fusion construct does note exacerbate the phenotype was supported by preliminary studies comparing transient expression of G93A-SOD1 and G93A-SOD1-GFP in embryos (data not shown). Other reports also demonstrate that GFP expression does not affect the mutant SOD1 phenotype in embryos [36], and the use of GFP in transgenic tissue-specific reporter zebrafish lines like HB9:mGFP zebrafish that express GFP in MNs [28] provide further evidence supporting our approach. The Tol2-based transposon system supports dramatically increased transgenesis efficiencies [48], and using this approach, we were able to establish 6 independent transgenic G93A-SOD1-GFP zebrafish lines. The G1 and G21 lines characterized for these studies exhibit ubiquitous expression of the transgene that persists throughout the zebrafish lifespan and through multiple generations.

Transgenic zebrafish are an emerging model for the study of neurodegenerative diseases and at the initiation of this project, long-term data was not available on how chronic G93A-SOD1 expression, or late-onset neurodegeneration in general, would affect zebrafish. Therefore, we first examined swimming behavior as a correlate to muscle weakness observed in ALS patients using the Noldus Larvae Activity Monitoring System. Spontaneous swimming behavior was regularly monitored for quantification of swim speed and duration of movement during a defined interval. Linear regression analysis of AB control and transgenic G93A-SOD1-GFP zebrafish swimming velocity between 10 and 60 weeks of age (Figure 3C) did not reveal a significant difference in overall swim speed change, or slope, between the 2 lines. What is important to note, however, is that examination of the individual time points reveals similar swim speeds at 10 weeks of age, whereas apparent differences in the degree of movement emerge after 20 weeks of age between the 2 lines, with the transgenic G93A-SOD1-GFP zebrafish consistently exhibiting a trend of slightly slower swim speeds. Transgenic G93A-SOD1-GFP zebrafish, however, do tend to spend more time resting at advanced time points compared to controls (Figure 3D). These findings may indicate that, although there are no striking changes in locomotor potential between 10-60 weeks of age in either line, the increased time resting could reflect a higher tendency for fatigue in the transgenic G93A-SOD1-GFP zebrafish that may precede and potentially predict later motor phenotypes. Given the drastic motor deficits that appear in rodent ALS models [22, 49], these modest effects were initially surprising, but during the course of these studies the Beattie laboratory reported the first generation and characterization of a transgenic ALS zebrafish model [40]. In that study, the transgenic zebrafish overexpress zebrafish G93R-sod1 and exhibit decreased swimming endurance at 12 and 16 mo of age compared to non-transgenic controls, as measured by calculating critical swimming speeds against an increasing current flow [40]. These findings suggest that more drastic effects on movement behavior in our model may not be apparent within the 60 week time frame examined and support the continued evaluation of swimming behavior in our transgenic lines, and also suggest the possibility that muscle strength defects in our model may become apparent by exploring additional assays for motor function.

As previously described, neuromuscular defects are apparent in ALS rodent models and patients well before symptom onset and MN loss [3, 4]; therefore, we next performed a detailed pathological characterization of NMJs and MNs in control and transgenic G93A-SOD1-GFP zebrafish. Techniques to quantify NMJ innervation by MNs are well-established in rodents [25, 45, 50, 51] and larval zebrafish [52‐55], however, only limited information regarding visualization and quantification of juvenile and adult zebrafish NMJ integrity are available [32, 56]. For the current studies, we labeled presynaptic MNs and postsynaptic AChR clusters with SV2/NF and αBTX, respectively, and quantified the number of intact NMJs in control AB and transgenic G93A-SOD1-GFP zebrafish between 10 and 60 weeks of age (Figure 4). Transgenic G93A-SOD1-GFP zebrafish exhibited a significant loss of intact NMJs at 20 weeks of age that was consistent across the remaining time points. These effects at the NMJ even in the absence of significant symptomatic affects are consistent with those observed in rodent ALS models. Transgenic G93A-SOD1 mice exhibit a 40% reduction in the number of innervated NMJs at 47 days of age, while symptomatic onset does not occur until around 80 days of age [3]. The importance of NMJ maintenance is further supported by a study where muscle-specific expression of mutant SOD1 in mice promoted NMJ defects and distal axonal degeneration [57]. Interestingly, we noticed that this early decrease in NMJ integrity at 20 weeks of age was subsequently followed by alterations in the innervation patterns of the muscle that became apparent at the 30 week time point. While imaging at earlier time points revealed long axons that synapse along longitudinal sections of myofibers, later time points exhibited more branched nerve fibers that synapsed across multiple myofibers (Figure 5). These changes are significant in transgenic G93A-SOD1-GFP zebrafish beginning at 30 weeks of age, and may reflect compensatory reinnervation of myofibers as previously observed in transgenic ALS mice [24, 42]. MN loss follows NMJ denervation and distal axonal degeneration in patients and rodent ALS models [3, 42]; therefore, we also quantified MN numbers in transgenic G93A-SOD1-GFP zebrafish between 30-60 weeks of age. We observed an approximate 50% reduction in the number of large MN cell bodies in the ventral horn of transgenic G93A-SOD1-GFP zebrafish spinal cords at 40 weeks of age compared to age-matched controls (Figure 6). This coincides with the sequence of events in ALS rodent models, as MN loss is evident around 100 days in G93A-SOD1 mice, relative to initial NMJ defects as early as 47 days, and death at 131 days [3]. A timeline summarizing the sequence of observed morphological alterations in our transgenic G93A-SOD1-GFP zebrafish is depicted in Figure 7.

This characterization of our novel transgenic G93A-SOD1-GFP zebrafish supports and builds upon the previous findings in transgenic G93R-sod1 zebrafish [40]. The transgenic G93R-sod1 zebrafish exhibit a small but significant reduction in NMJ innervation at 11 days post-fertilization (dpf) in transgenic larvae, and a more significant loss of NMJ innervation at 12 months of age [40]. While intermediate time points were not described, these findings coincide with our later observations of NMJ loss between 40-60 weeks of age, and reflect the fact that significant NMJ defects are evident as early as 20 weeks of age in our model. MN loss is also evident in both models, with our transgenic G93A-SOD1 zebrafish exhibiting an approximate 50% reduction in MN numbers by 40 weeks of age, and the transgenic G93R-sod1 zebrafish having a 38% reduction in MN numbers at end-stage disease. In addition, transgenic G93R-sod1 zebrafish lines exhibit reductions in survival that become evident around 18-22 months of age [40]. While preliminary findings in our transgenic G93A-SOD1-GFP zebrafish did not reveal any noticeable differences in survival (data not shown), it is possible based on the findings in the other model that the 60 week time frame for the current study was too short to detect quantifiable differences, and studies are currently ongoing to determine when or if survival is affected in our model. Taken together, both models depict a sequence of events that supports the “dying-back” theory in ALS, where defects at the NMJ are the first insult and then lead to progressive distal axonal degeneration and subsequent MN loss [3, 4, 24, 42].

This newly established ALS model provides an additional system that will supplement our understanding of ALS neuromuscular pathology in the context of disease progression. While we observe significant neuromuscular defects that reflect a characteristic distal-to-proximal ALS disease course, there are still unanswered questions that should be addressed to understand why the disease course is not as severe in zebrafish as it is in G93A-SOD1 rodents, and why drastic locomotor deficits are not observed even in the context of MN loss. A possible explanation for the reduced disease severity could relate to the relative expression levels of mutant SOD1 in the transgenic models. Our data indicate that G93A-SOD1-GFP expression levels were approximately 64% that of the endogenous zebrafish wtSOD1 (Figure 2C), whereas transgenic G93A-SOD1 rodent models exhibit robust levels of overexpression [22]. It is also possible that the disease course is affected by the regenerative ability of zebrafish; while neuronal regeneration is not prevalent in human or rodent spinal cords, evidence of MN regeneration has been reported in zebrafish [58]. The observed alterations in NMJ innervation patterns (Figure 4) support the possibility of compensatory reinnervation, and although we observe a decrease in MN numbers in transgenic G93A-SOD1-GFP zebrafish, regenerative markers must be examined in spinal cords to understand what role regeneration plays in disease pathogenesis in this model. Non-cell autonomous effects including astrocytes and glial activation also have a significant impact on disease progression in rodent ALS models [1, 2]; therefore, the contributions of these cell types to disease pathogenesis should also be investigated. The impact of G93A-SOD1-GFP expression on other neuronal populations is also of interest, although no effects were seen on the commissural axon of the Mauthner neuron in the hindbrain, on sensory neurons of the lateral line, or on Rohon-Beard sensory neurons in the transient model described by Lemmens et al [36]. Finally, a detailed analysis of MN axons using electron microscopy and IHC with advanced imaging techniques to comprehend the mechanisms of axonal degeneration throughout the disease course will provide important insight into the “dying-back” mechanisms of MN degeneration in ALS [3, 4, 24, 42].

In summary, the novel transgenic G93A-SOD1-GFP zebrafish exhibit defects at the NMJ around 20 weeks of age, subsequent alterations in innervation patterns at 30 weeks of age, and evidence of MN loss around 40 weeks of age (Figure 7). These neuromuscular findings support the utilization of zebrafish as a model organism to investigate early events in ALS disease pathogenesis for therapeutic development. Furthermore, the generation of transgenic ALS zebrafish models expressing other familial ALS mutations such as TDP43 should be considered, as they have the potential to provide additional insight into the mechanisms of ALS onset and progression.