Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disorder that, for ∼80% of patients, is fatal within five years of diagnosis. To better understand ALS, animal models have been essential; however, only rodent models of ALS exhibit the major hallmarks of the disease. Here, we report the generation of transgenic zebrafish overexpressing mutant Sod1. The construct used to generate these lines contained the zebrafish sod1 gene and ∼16 kb of flanking sequences. We generated lines expressing the G93R mutation, as well as lines expressing wild-type Sod1. Focusing on two G93R lines, we found that they displayed the major phenotypes of ALS. Changes at the neuromuscular junction were observed at larval and adult stages. In adulthood the G93R mutants exhibited decreased endurance in a swim tunnel test. An analysis of muscle revealed normal muscle force, however, at the end stage the fish exhibited motoneuron loss, muscle atrophy, paralysis and premature death. These phenotypes were more severe in lines expressing higher levels of mutant Sod1 and were absent in lines overexpressing wild-type Sod1. Thus, we have generated a vertebrate model of ALS to complement existing mammal models.
Amyotrophic lateral sclerosis (ALS) is a devastating neurodegenerative disease that is characterized by progressive motor neuron loss in the brain and spinal cord, leading to paralysis and death. The incidence of ALS is 1.6 per 100,000 and approximately 5000 Americans are diagnosed with ALS each year (Hirtz et al., 2007), with only 20% of patients living for five years or more after diagnosis. Thus far, riluzole is the only FDA-approved treatment for the disease and extends life by only a few months (Bensimon et al., 1994). Approximately 10% of ALS cases are caused by mutations in specific genes, whereas the remaining 90% of ALS cases are sporadic, with no known cause. Sporadic and familial ALS (FALS) share several clinically and pathologically indistinguishable disease features, including progressive upper and lower motor neuron loss, muscle atrophy, paralysis and early death. However, it has been reported recently that TDP43-containing aggregates are present in sporadic ALS but not FALS (Mackenzie et al., 2007). Modeling ALS using identified genes, however, has the potential to lend insight into common disease mechanisms. Mutations in the superoxide dismutase (SOD1) gene account for 20% of FALS cases (Deng et al., 1993; Rosen et al., 1993; Valdmanis and Rouleau, 2008). To date, over 150 mutations in SOD1 have been discovered, including the point mutation G93R (Elshafey et al., 1994; Turner and Talbot, 2008).
To gain an understanding of how mutations in SOD1 cause ALS, mouse models were generated and have been used extensively to study ALS onset, progression and therapeutics (Turner and Talbot, 2008). These animals exhibit hindlimb tremor, progressive weakness, locomotor deficits, paralysis and early death (Bruijn et al., 1997; Gurney et al., 1994; Wong et al., 1995). As useful as these have been, mouse models do have some drawbacks: the mice are highly inbred, which may affect the disease progression and phenotypes; some lines express very high levels of mutant protein, which may not accurately reflect the human situation; and it is expensive to perform drug screens in mice. Drosophila and C. elegans that overexpress human SOD1 mutations have also been generated (Mockett et al., 2003; Oeda et al., 2001; Watson et al., 2008). These animals do recapitulate some ALS phenotypes in that they show motoneuron damage (Watson et al., 2008), however, they do not display motoneuron loss, paralysis or premature death. To complement these models and to generate a vertebrate model of ALS that may be more amenable to cellular manipulation and to drug or genetic screening, we have generated zebrafish that carry mutant forms of Sod1.
The zebrafish is an excellent organism for modeling neurological diseases owing to its conserved, yet simplified, vertebrate nervous system, the ability to make transgenic and targeted knockout animals, and the ease of making genetic mosaics (Doyon et al., 2008; Meng et al., 2008; Lieschke and Currie, 2007; Carmany-Rampey and Moens, 2006). Furthermore, for the study of motoneuron diseases, the zebrafish offers accessible motoneurons that can be manipulated in vivo (McWhorter et al., 2003; Lemmens et al., 2007; Boon et al., 2009; Kabashi et al., 2010), making them highly relevant for performing cell-autonomy studies, electrophysiology and imaging. We generated transgenic zebrafish expressing mutant zebrafish Sod1 at moderate levels. We show that these fish recapitulate the major phenotypes of ALS including neuromuscular junction defects, decreased endurance, motoneuron loss and muscle pathology. Moreover, we find that the onset and progression of the disease phenotypes is variable, which may reflect a more natural state of the disease, as seen in humans.
Generation of transgenic sod1 zebrafish
To generate zebrafish expressing transgenic mutant Sod1, we chose to use a zebrafish genomic region containing the endogenous sod1 promoter and sod1 gene. Because zebrafish live at a lower temperature (28°C) than mammals (37°C), we thought it would be more prudent to use the fish sod1 gene and regulatory regions to guard against any temperature sensitivity. SOD1 is highly conserved between human, mouse and zebrafish (supplementary material Fig. S1), with 77% identity between the zebrafish and human amino acid sequence. Furthermore, of the 72 amino acids within SOD1 that are mutated in FALS, 88% (63 out of 72) are conserved in zebrafish, compared with 90% (65 out of 72) in the mouse and 72% (52 out of 72) in the fly. We mutated Sod1 by changing glycine 93 to arginine (G93R); this mutation affects a conserved amino acid that is often mutated in FALS (http://alsod.iop.kcl.ac.uk) (Elshafey et al., 1994).
To generate transgenic fish expressing mutant Sod1, we identified a bacterial artificial chromosome (BAC) containing the zebrafish sod1 gene (accession number AL837524). Since this BAC contained other genes, we isolated a 21-kb region containing the 5-kb sod1 gene and flanking sequences (11.7 kb upstream and 4.5 kb downstream). This amount of upstream sequence is consistent with mouse and rat transgenic Sod1 models (11.5–14.5 kb) (Gurney et al., 1994; Nagai et al., 2001). The point mutation G93R was generated and the wild-type sod1 or mutant sod1 21-kb genomic region were recombineered into a BAC vector backbone (pBAC5). To track transgene expression, we utilized the zebrafish heat shock protein 70 (hsp70) promoter (Halloran et al., 2000), which we used to drive the fluorescent protein DsRed. We added the hsp70 that was driving DsRed to the engineered pBAC5, generating the final construct: Tg(sod1:sod1G93R or WT; hsp70:DsRed) (Fig. 1A). The DNA construct was injected into early one-cell stage embryos. At 48 hours post-fertilization (hpf), the injected embryos (F0) were heat shocked and those with high mosaic DsRed expression at 72 hpf were kept and grown to adulthood. Since integration into the germline is random, these F0 fish were raised to adulthood and outcrossed to identify those that incorporated the construct into their germline. F1 larvae from these outcrosses were heat shocked and analyzed for DsRed expression. F1 animals carrying the transgene, as identified by DsRed expression (supplementary material Fig. S2), were grown to adulthood and used to establish lines (F2 generation). Tg(sod1:sod1;hsp70:DsRed) fish will hereafter be referred to as either WT (transgenic wild-type Sod1) or G93R (transgenic mutant Sod1 with a point mutation at G93R), followed by the line designation [e.g. Ohio State (os) #].
We next characterized copy number and Sod1 protein levels in two WT lines and two G93R lines (Fig. 1B–D). Quantitative PCR (qPCR) was performed to determine sod1 gene dosage in comparison to non-transgenic fish, and western blots were performed to examine Sod1 steady-state protein levels. One founder line, G93R os10, with a high transgene copy number (∼165 copies) showed a threefold increase in steady-state Sod1 protein levels in the brain and a fourfold increase in the spinal cord at one year of age compared with non-transgenic, wild-type siblings (nTg) (Fig. 1B–D). Another G93R line, os6, had approximately nine copies of the transgene and Sod1 protein levels were 2.5-fold higher than in non-transgenics (Fig. 1B,D). In WT lines, the number of copies of the transgene expressing wild-type Sod1 varied from two to 42, and the increase in protein levels varied from 1.9- to 3.3-fold. Thus, WT os4 protein levels were similar to those in G93R os10 animals, and WT os3 protein levels were similar to those in G93R os6 animals. The finding that the G93R os10 line shows increased Sod1 protein in the spinal cord, brain, muscle and liver (Fig. 1C) is consistent with the ubiquitous expression of Sod1 and indicates that the transgene is expressed in a manner consistent with endogenous sod1. The disease characterization that is discussed here was performed most extensively on G93R os10 fish.
G93R larvae and adult fish have abnormal neuromuscular junctions
One of the earliest hallmarks of disease in mouse models of ALS is neuromuscular junction (NMJ) defects that are seen as early as one to two months of age, before overt behavioral phenotypes (Frey et al., 2000; Schaefer et al., 2005; Fischer et al., 2004). We first examined motor axon outgrowth in G93R os10 embryos and found this to be indistinguishable from motor axon outgrowth in nTgs (supplementary material Fig. S3). An examination of NMJs in larval and adult fish, however, did reveal abnormalities (Fig. 2). Using SV2 as a pre-synaptic marker and α-bungarotoxin (BTX) as a post-synaptic marker, whole-mount immunostaining and confocal microscopy was performed on 11 days post-fertilization (dpf) larvae. A small but significant reduction in SV2 and BTX colocalization was observed using Pearson’s correlation coefficients. Furthermore, a significant reduction in the area occupied by BTX (post-synaptic volume) in G93R os10 larvae compared with nTg controls was observed.
Despite larval NMJ alterations, these fish survived and grew to adulthood. In 12-month-old fish, numerous NMJ abnormalities were observed. Not only was colocalization of SV2 and BTX reduced significantly, but there was also a reduction in the area of the NMJ (NMJ volume). These quantitative measures are consistent with observations that NMJ synaptic boutons were shorter and more punctate in G93R os10 adults compared with nTg controls, which showed longer stretches of boutons (Fig. 2C and supplementary material Fig. S4). We did not observe gross denervation across all muscles, even in end-stage fish. This may reflect differences in zebrafish muscle innervation, which is polyneuronally innervated throughout life and may allow more compensatory innervation to occur. However, these data do indicate that there are changes at the NMJ in zebrafish overexpressing mutant Sod1.
G93R adult fish develop progressive motor abnormalities
To examine motor output, fish were subjected to whole-body exercise using a swimming tunnel. To measure endurance in fish, critical swimming speed (Ucrit) was determined by monitoring the ability to swim against an increasing current over time (Pagala et al., 1998; Plaut, 2000). Fish were placed in a swimming tunnel and a water current was applied. Zebrafish naturally swim against this current. Every five minutes the current was increased until the fish could no longer swim against the current. Adult (10 months old) WT os4 and nTg fish swam for similar lengths of time (Fig. 3A). When 12-month G93R os10 adults were tested, we found that they were compromised in their ability to swim against increasing current (Fig. 3B). Furthermore, a subset of G93R os6 fish (eight of 22, 36% of those tested) also showed significant endurance defects at 16 months of age (Fig. 3C). Fish that showed endurance defects in the swim tunnel test did not recover and continued to perform poorly over time. WT os3 fish were unaffected compared with nTg controls at 24 months of age (Fig. 3D). Thus, both G93R lines were unable to maintain swimming when subjected to increasing current. This is in contrast to WT and nTg fish that could swim against stronger currents. Therefore, the presence of mutant Sod1 in zebrafish results in a reduced ability to perform in this swimming test.
Muscle physiology in G93R adults
To determine whether the motor abnormalities observed in G93R os10 fish at 12 months of age were caused by changes in skeletal muscle contractile properties, we examined the optimal force and fatigability of trunk skeletal muscle. Adult G93R os10 fish and sibling controls were anesthetized and optimal force measurements were obtained by directly stimulating across the body trunk muscle from behind the gills to the tail. No difference in maximum twitch force was detected between nTg and G93R os10 fish indicating that the muscle contractile properties were intact (Fig. 4A). To determine whether the muscle fatigued in a similar manner, repeated stimulations of 4 milliseconds at 0.2 Hz was applied to the muscle. The response of skeletal muscle to repeated stimulation did produce significantly different responses between nTg and G93R os10 fish, as indicated by the difference in slope upon repeated stimulation (Fig. 4B,C). We found that G93R os10 fish exhibited better fatigue resistance than nTg body muscle. This is similar to what was found in mice overexpressing mutant SOD (Derave et al., 2003) and suggests a common muscle phenotype in these two species. These data also support the hypothesis that the majority of the movement impairments (i.e. decreased swim tunnel performance and partial paralysis) in G93R os10 fish are the result of defects in the neural input to the muscle, as opposed to defects in intrinsic muscle properties.
Examination of pathology in G93R fish reveals motoneuron loss and muscle atrophy
G93R os10 fish exhibited a progressive deficiency in locomotion with some fish showing intermittent paralysis over time. Some fish exhibited signs of more progressive paralysis, including severely decreased movement for several consecutive days (supplementary material Movie 1). These fish were considered end-stage and were sacrificed for pathological analysis. Cellular pathology in ALS includes spinal cord motoneuron loss and degeneration. To analyze this in zebrafish overexpressing mutant Sod1, spinal cords from end-stage G93R os10 fish were processed for choline acetyl transferase (ChAT) labeling to examine motoneurons (Fig. 5A). For this analysis, only ChAT-positive cells that were ≥10 μm in diameter were counted to ensure that only motoneurons were included in these counts. We observed a significant reduction in motor neuron number with end-stage fish having ∼38% loss compared with nTg siblings (Fig. 5B).
SOD1 mouse models of ALS have revealed mitochondrial vacuolization in spinal cord motor neurons (Dal Canto and Gurney, 1995; Wong et al., 1995). Consistent with these findings, electron microscopy (EM) on spinal cords from end-stage G93R os10 fish revealed that spinal cord motoneurons were compromised. Shrunken motoneurons containing vacuolated mitochondria with disintegrating cristae were consistently observed (Fig. 6A). Furthermore, EM examination of fast trunk muscle also revealed defects. G93R os10 fish had myofibrillar and mitochondrial degeneration, and extensive collagen deposition, suggesting muscle atrophy and degeneration at end-stage (Fig. 6B). Therefore, fish overexpressing mutant Sod1 exhibited motoneuron loss, and both motoneuron and muscle degeneration, consistent with ALS phenotypes.
Reduced survival in G93R lines
To address whether fish overexpressing mutant Sod1 had a shorter life span, we analyzed survival over time (Fig. 7). Both G93R os10 and G93R os6 fish exhibited significantly shorter survival in comparison to WT os3 and nTg controls. G93R os10 and G93R os6 fish did not exhibit reduced survival at 7±2 or 14±2 months of age in comparison to nTg or WT os3 fish. However, by 18±1 months of age, significant differences in survival between the G93R lines and WT os3 or nTg fish were observed. A significant reduction in survival for G93R os6 and G93R os10 fish was observed at 22±2 and 27±2 months of age in comparison to WT os3 or nTg fish. WT os3 and nTg survival did not differ significantly at any age. Prior to death, many of the G93R os6 and G93R os10 fish exhibited severely decreased movement or partial paralysis. However, no nTg or WT os3 fish exhibited behavioral abnormalities or paralysis. These data indicate that fish overexpressing mutant Sod1 exhibited reduced rates of survival.
Here we describe the development and characterization of transgenic zebrafish overexpressing mutant Sod1 as an approach to model FALS. Fish overexpressing mutant Sod1 have NMJ defects, spinal cord motoneuron loss, muscle degeneration, decreased endurance, partial paralysis and premature death. Many of the cellular pathologies are consistent with those seen in mice and rats overexpressing SOD1, supporting the idea that overexpression of SOD1 in vertebrate animals results in the development of common pathologies. However, we would not expect all aspects of disease phenotypes to be identical across species owing to differences in the genetics of laboratory strains and animal physiology. For example, zebrafish laboratory strains are not highly inbred owing to the negative impact of inbreeding on reproduction (Guryev et al., 2006; Nechiporuk et al., 1999). Thus, disease onset and the time between onset and death, which is very stereotyped in some inbred mouse lines, is much more variable in the G93R zebrafish lines, and may be because of genetic heterogeneity within strains. This is not unlike FALS patients, who show phenotypic variability even when carrying the same SOD1 mutation (Aggarwal and Nicholson, 2005; Hand et al., 2002). NMJ denervation in SOD1 mutant mice is variable and muscle dependent, with fast muscle regions affected first and more dramatically and slow regions spared (Frey et al., 2000). Only around 8% of synapses at the sternomastoid muscle are denervated in symptomatic G93A mice and around 70% are denervated at end-stage (Schaefer et al., 2005), suggesting a progressive loss of NMJ synapses. Fish that overexpress mutant Sod1 show pre-symptomatic changes at NMJs; this was more severe in adults, supporting progressive NMJ defects. However, zebrafish expressing mutant Sod1 do not exhibit the same degree of NMJ abnormality or motoneuron loss as rodent models. Adult zebrafish, however, have the capacity to regenerate motoneurons (Reimer et al., 2008), and zebrafish muscle is polyneuronally innervated throughout life unlike in mice and humans (Westerfield et al., 1986). Thus, it is possible that compensatory mechanisms such as collateral innervation and motoneuron regeneration occur in these animals and may affect the phenotype.
Transient overexpression of human mutant SOD1 RNA in zebrafish was reported to cause motor axon outgrowth defects such as shorter axons or branched axons during embryonic development (Lemmens et al., 2007). The transgenic lines expressing mutant Sod1 reported here did not show defects in motor axon outgrowth. Lemmens et al. (Lemmens et al., 2007) reported a 6.1% decrease in axonal length [approximately 101 μm in SOD1(wt) versus approximately 95 μm in SOD1(G93A) embryos] in the SOD1(G93A)-injected embryos compared with SOD1(wt)-injected embryos. Because we did not measure axons directly, we may not have detected such a subtle result. It is more likely, however, that this difference may simply reflect the high levels of RNA used in the transient study compared with the steady, lower levels of mutant Sod1 present in the transgenics.
Loss of muscle strength is a hallmark phenotype of ALS. Both G93R os6 and os10 fish displayed a significant decrease in endurance, as shown by their performance in the swim tunnel test. Endurance tests measure the ability of an animal to sustain a force over time and, although it is difficult to relate aquatic to non-aquatic tests, the swim tunnel is a close approximation of a treadmill. Both muscle integrity and cardiovascular health contribute to endurance. In SOD mutant mice, both neural and vascular defects are early phenotypes (Zhong et al., 2008). Thus, we contend that the neuromuscular defects observed in G93R fish certainly contribute to the decreased endurance, however we cannot rule out a role for the cardiovascular system.
Analysis of muscle force properties supports the hypothesis that alterations in the neural component of the neuromuscular system contribute to movement defects because we found that the intrinsic force generated by body muscle in nTg and G93R os10 fish was indistinguishable. We did find areas of muscle fiber degeneration by EM suggesting that there are regions of compromised muscle fibers. However, when the whole fish is analyzed, the muscle functions normally. This supports the idea that the defects in swimming endurance are not the result of changes in the ability of muscle to generate force and indicate that the problem is neural as opposed to muscular. One area where there was a difference was in how the muscle responded to repeated stimulation as a way to test fatigue. In these tests, the G93R os10 line was actually more resistant to fatigue. This result is consistent with mouse ALS SOD models (Derave et al., 2003; Sharma and Miller, 1996). One possible explanation for why these fish are more resistant to fatigue may be to do with the fact that repetitive stimulation exhausts muscle, which can cause oxygen radical damage. Higher levels of Sod may help curb this damage.
Here, we have generated a vertebrate model of FALS and show that zebrafish overexpressing mutant Sod1 exhibit the hallmark phenotypes of ALS. We suggest that this model will be useful for further investigation into early changes in the neuromuscular system and motoneuron physiology during disease progression. Moreover, zebrafish are highly amenable to chimeric analysis and thus will offer a tool to address the continued debate concerning the cell autonomy of FALS.
MATERIALS AND METHODS
Adult and larval zebrafish (Danio rerio) were maintained at The Ohio State University fish facility at 28.5°C and bred according to established procedures (Westerfield, 2000). Animal protocols were approved by the Ohio State University Committee on Use and Care of Animals.
The transgenic lines reported here are available by contacting http://zebrafish.org/zirc/home).. If there is sufficient interest, they will be deposited in the zebrafish international resource center (ZIRC:
The zebrafish sod1 gene was isolated by BLASTing the zebrafish genome with the mouse cDNA, and a BAC containing the zebrafish sod1 gene was identified (DanioKey BAC library, BAC accession number AL837524). Since this BAC contained other genes, we isolated a fragment of the BAC containing an 11.7-kb upstream region, the sod1 gene with its five exons spanning 5.06 kb, and 4.5 kb of downstream sequences. We recombineered this into pBAC5, into which we had added ISceI meganuclease sites to increase the rate of transgenesis (Grabher et al., 2004; Liu et al., 2003). The zebrafish heat shock protein 70 (hsp70) promoter and the DsRed (Discosoma sp. red fluorescent protein) marker gene were also introduced into the pBAC5 by recombineering. A 3-kb fragment containing the last two exons of sod1 was then excised from the engineered pBAC5 and mutagenesis was performed using the Stratagene site-directed mutagenesis kit. The primers for the G93R mutation were: forward, 5′-GACCGCTGATGCCAGTCGTGTTGCAAAAATTG-3′ and reverse, 5′-CAATTTTTGCAACACGACTGGCATCAGCGGTC-3′. The mutagenized fragment was then recombineered back into the engineered pBAC5. The final construct [Tg(sod1:sod1G93R or WT; hsp70:DsRed)] was then cut with ISceI meganuclease (New England Biolabs) and injected into zebrafish embryos at the single cell stage.
Generation of transgenics and quantitative PCR
Adult wild-type hybrids of AB* and LF strains, designated AB*LF, were used to generate transgenics. Each ISceI-digested BAC was diluted in buffer containing 1× enzyme buffer, 1 U/ml of ISceI, Phenol Red and 1× injection buffer (10 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 100 mM NaCl, 30 μM spermine and 70 μM spermidine) to a final concentration of DNA ranging from 80–100 ng/μl, and 1 nl of DNA solution was injected into early one-cell stage embryos. At 48 hpf, the embryos were heat shocked for 1 hour at 37°C in a standard thermocycler (Bio-rad Peltier Thermal Cycler PTC-100) and examined at 72 hpf under a fluorescence microscope. Only those embryos with the strongest DsRed expression levels, and with a high number of cells expressing DsRed, were grown to adulthood and subsequently screened as potential F0s. The potential F0s were bred to AB*LF fish, and offspring were subjected to heat shock and screened for DsRed expression. DsRed-positive embryos were isolated, raised to adulthood, and bred to generate founder transgenic fish (F1s). Now that the DNA had integrated into the germ line, the DsRed was in all cells. Transgenic F1s were bred to generate the individual transgenic lines. In total, approximately 500 embryos per line were injected and 50 highly chimeric fish were grown to adulthood. Among these, 10–30% gave rise to transgenic F1s.
Transgene copy number was determined by quantitative PCR (qPCR). Genomic DNA was isolated from the tail fin using the DNeasy blood and tissue kit (Qiagen). Primers were designed to unique genomic sequences for sod1: forward, 5′-TAAAGGGCCACAAATCACAC-3′ and reverse, 5′-TTTCAGTGACAATCCGTACAG-3′; and β-actin: forward, 5′-ATGAGACCACCTTCAACTCC-3′ and reverse, 5′-TGAAATCACTGCAAGCAAACTG-3′. qPCRs using sod1 and β-actin primers were run at various primer concentrations with a standard dilution of genomic DNA from AB*LF fish in order to identify a concentration that yielded the optimal cT spread. Each reaction contained 0.3 μM of forward and reverse primer, 5 ng genomic DNA, and iQ SYBR Green Supermix (Bio-rad). qPCR was performed under the following conditions: one cycle at 94°C for 5 minutes; 45 cycles of 94°C for 30 seconds, 60°C for 30 seconds and 72°C for 30 seconds; and then a step down from 94°C to 4°C, by 0.5°C every 10 seconds, using an iCycler (Bio-rad). To determine copy number, the difference in cT of sod1 and cT of β-actin was given the value X (ΔcT). 2X was then calculated for each sample. The fold change between the non-transgenic control and transgenic line was determined [ΔΔcT=2X(AB*LF)–X(Tg)]. Following the methods of Alexander et al. (Alexander et al., 2004) and assuming two copies in the non-transgenic control fish, sod1 gene copy number of the transgenic was calculated [2(ΔΔcT)] (Alexander et al., 2004).
Swimming endurance test
Critical swimming speed, Ucrit, which is the maximum velocity a fish can sustain for a set period, was measured using a simplified water tunnel (AccuScan Instruments Inc.) (Plaut, 2000). Adult zebrafish were individually introduced into the water flow chamber of a swim tunnel. The water current flow rate was gradually increased to 4.1 cm/sec, and the fish were subjected to this current for 5 minutes. Subsequently, every 5 minutes, the flow rate was increased by 4.1 cm/sec increments until the fish could no longer maintain swimming and fell into a mesh net at the end of the water flow chamber. The fish was allowed two opportunities to re-enter the highest achieved current by pausing the time, reducing the currently flow, and slowly increasing the current back to the fatigue velocity. If the fish could not sustain swimming in this current, the time was recorded when the fish stopped swimming. The critical swimming speed was calculated based on the formula: Ucrit=Ui + (UiiTi/Tii), where Ui=the highest velocity maintained for a whole interval (cm/sec), Uii=the velocity increment (4.1 cm/sec), Ti=the time elapsed at fatigue velocity (minutes) and Tii=the time interval (5 minutes) (Brett, 1964; Plaut, 2000).
Force measurements in zebrafish
The contractile strength of intact zebrafish was assessed following anesthesia by tricaine. The whole fish was mounted at slack length in a circulating bath containing oxygenated water (26°C) from the zebrafish housing tank. One end of the fish was held in place by inserting a hook (connected to a force transducer) through the body, just caudal to the gills. A second hook, connected to a linear micro-manipulator, was placed through the body at the tail. Electrical stimulation was delivered using two parallel platinum-iridium electrodes positioned along the length of the bath on either side of the fish. The fish was stretched from slack length and subjected to a 4-ms stimulus at 1-minute intervals until optimum length was reached, defined as the length at which maximum twitch force was achieved, and forces were recorded and stored digitally. After a five-minute rest period, the fish was subjected to a series of 4-ms stimuli at 0.2 Hz until the force had declined by at least 50% of the initial value. All force measurements are expressed per unit of cross-sectional area (mN/mm2), as done previously for similar skeletal muscle experiments (Martin et al., 2009). Cross-sectional area (CSA) was calculated using the following equation: CSA=(muscle mass in g)/[(optimal fiber length in cm) × (muscle density in g/cm3)], where muscle density is 1.06 g/cm3.
Adult zebrafish were deeply anesthetized in fish water, with 2% tricaine, and decapitated. Tissues (brain, spinal cord, muscle and liver) were removed and homogenized with a TissueMiser homogenizer (Fisher Scientific) in homogenizing buffer (50 mM Tris, 10 mM EDTA, pH 7.2, 10% SDS) with protease inhibitors (Roche). Tissue was boiled and centrifuged at 16,000 × g for 5 minutes at 4°C. The total soluble protein concentrations were determined by the Dc protein assay according to manufacturer’s directions (Bio-Rad). Western blots were performed as described (Schagger, 2006). Briefly, protein lysates were separated on 7.5% tricine SDS-polyacrylamide gels and transferred to poly-vinylidene difluoride (PVDF) membranes (Immobilon P, Millipore). Blots were blocked for 1 hour with 5% milk in tris-buffered saline with 0.1% Tween20 (TBST) and, subsequently, probed with rabbit anti-SOD1 antibody (1:5000 in blocking buffer; Santa Cruz) or mouse anti-actin (1:1000; Abcam) overnight (ON) at 4°C. Following four washes with TBST for 5 minutes each, blots were incubated with the appropriate secondary antibody (goat anti-mouse IgG-HRP or goat anti-rabbit IgG-HRP, 1:5000 in 5% milk/TBST, Jackson ImmunoResearch Laboratories, Inc.). After five washes with TBST for 5 minutes each, signal was detected by chemiluminescence (Amersham Western Blotting Detection Reagents, GE Healthcare) by exposures with an Omega 12iC Molecular Imagine System (UltraLum, Inc.) and by film. Densitometry was performed on the digital images using Ultraquant Image acquisition and analysis software V6.0 (UltraLum, Inc.). For subsequent blotting, membranes were stripped with Restore Plus western blot stripping buffer (Thermo Scientific) for 15 minutes at room temperature (RT) and washed five times with TBST for 5 minutes per wash prior to blocking, as described above. Quantitation was performed by normalizing the signal from each lane on the Sod1 blot to the β-actin signal in each lane.
Whole-mount staining for synaptic vesicle 2 (SV2) and α-bungarotoxin (BTX) was performed on 11-dpf zebrafish larvae as follows. Larvae were fixed in 4% paraformaldehyde (PFA) ON at 4°C and permeabilized by successive incubations in dH2O for 5 minutes, ice-cold acetone for 7 minutes, and dH2O for 5 minutes. Larvae were incubated in PBDT buffer (1% DMSO, 1% BSA, 0.5% TritonX-100, 1× PBS) with 5% normal goat serum (NGS) for 60 minutes, followed by incubation with Alexa 488-conjugated α-BTX (1:100 in 5% NGS/PBDT, Molecular Probes) for 30 minutes at RT. Larvae were washed three times in PBDT for 15 minutes per wash and incubated in primary mouse monoclonal anti-SV2 antibody (1:50 in 5% NGS/PBDT, Developmental Studies Hybridoma Bank) at 4°C ON. Larvae were then washed six times with PBST (0.5% TritonX-100, 1× PBS) for 15 minutes per wash and incubated in Alexa 633-conjugated goat anti-mouse antibody (1:400 in 2% NGS/PBDT, Chemicon International) for 4 hours at RT followed by six 15-minute washes in PBST and clearing in PBS:glycerol. The larvae were mounted onto slides with Vectashield (Vector Labs) and imaged using confocal microscopy (TCS SL, Leica Microsystems).
To examine synapses in adult fish, adult zebrafish (12 months) were anesthetized in tricaine, decapitated, and fixed in 4% PFA ON at 4°C. Fish were embedded in OCT (Tissue-Tek) and snap frozen in isopentane. Cryostat serial cross sections (16 μm) of the caudal region of the fish were collected on superfrost plus slides (ColePalmer) and stored at −80°C until processed further. Sections were rehydrated for 5 minutes in PBDT buffer with 5% NGS, incubated with Alexa 488-conjugated α-BTX (1:100 in 5% NGS/PBDT) for 30 minutes at RT, and washed three times for 15 minutes each in PBDT. This was followed by incubation in primary mouse monoclonal anti-SV2 antibody (1:50 in 5% NGS/PBDT) at 4°C ON. Sections were washed four times with PBST for 15 minutes each and incubated in Alexa 633-conjugated goat anti-mouse antibody (1:400, Chemicon International) for 2 hours at RT. Sections were washed six times with PBST, mounted using Vectashield (Vector Labs) and imaged using a Leica TCS SL confocal microscope (Leica Microsystems). Image stacks of 16 μm thickness (0.6 μm/section) were obtained, and then processed using ImageJ Software (National Institutes of Health) to quantitate fluorescence intensity and identify co-localization of BTX and SV2 signals. Additionally, the overall NMJ density (Ncoloc) and pre- and post-synaptic volume were calculated using Costes’ colocalization quantification and quantitative immunocolocalization program (Intensity colocalization analysis) (Costes et al., 2004; Li et al., 2004).
ChAT immunohistochemistry was performed as described (Clemente et al., 2004). Briefly, fresh frozen sections (20 μm) of spinal cord were collected from end-stage fish that had been fixed in 4% PFA in 0.1 M phosphate buffer (PB) (pH 7.4) for 10 minutes. The sections were washed twice in 0.1 M PB (pH 7.4) for 5 minutes per wash. To decrease background staining, the sections were incubated in 2% H2O2/PBS for 20 minutes. Sections were washed three times in PBDT with 0.2% Tween20 for 5 minutes, incubated in PBDT with 0.2% Tween and 10% normal donkey serum (NDS) for 2 hours at RT, followed by incubation in goat polyclonal anti-ChAT antibody (1:125 in PBDT with 0.2% Tween and 10% NDS, Chemicon International) for 2 days at 4°C. Sections were washed four times with PBST with 0.2% Tween20 for 15 minutes each and incubated in biotinylated donkey anti-goat secondary antibody (1:250 in PBDT with 5% NDS, Santa Cruz Biotechnology) for 2 hours at RT. Sections were washed four times with PBST with 0.2% Tween20 for 15 minutes each and washed twice with 0.01 M PB (pH 7.4) for 15 minutes. Sections were incubated in Vectastain RTU reagent (Vector Labs) for 90 minutes and were washed four times with 0.01M PB (pH 7.4) for 15 minutes each. The reaction product was visualized with 0.025% 3,3-diaminobenzidine (DAB) and 0.0033% H2O2 in 0.01 M PB (pH 7.4) or 0.2 M Tris (pH 7.6). The course of the reaction was monitored and stopped by washing in PBST. Sections were serially dehydrated for 10 minutes each in 70% and 100% EtOH, cleared in xylene for 15 minutes, and mounted with Permount (Fisher Scientific). Sections from the dorsal fin region were blinded to genotype and each section was counted for ChAT-positive motor neurons. A minimum of 40 serial sections were counted from three fish for each genotype.
Numerous tanks of WT os3, G93R os6, G93R os10 and nTg fish were followed from the age of 3 months, which is when the zebrafish reached adulthood, until 27.5 months of age. To track the survival of more than 1000 fish, the numbers of fish were counted periodically. Each time fish were counted, they were placed into an age bracket: 7±2, 14±2, 18±1, 22±2 and 27±2 months of age. At any given age, several independent clutches of fish were counted for each line. At the respective ages listed above, for nTg fish n=55, 147, 516, 628 and 520; for WT os3 fish n=34, 34, 95, 155 and 89; for G93R os6 n=44, 101, 169, 152 and 154; and for G93R os10 n=49, 147, 333, 370 and 230. Fish were also monitored for abnormal behavior, such as severely reduced movement or paralysis. The percentage of surviving fish was calculated for each age bracket. A Z-test for two proportions was used to determine significance with 99% confidence. All data are reported as the percentage of surviving fish at the 95% confidence intervals.
Statistics were performed using SPSS 10. Motor axons were analyzed using a non-parametric analysis using the Mann-Whitney test. Image analysis for NMJ analysis was performed using NIH ImageJ software and quantitative analysis of images was performed using colocalization analysis and an intensity colocalization analysis plugin (Costes et al., 2004; Li et al., 2004). The values from the individual images were compared between control and transgenics using an unpaired t-test. Critical swimming speed was analyzed using an unpaired t-test. Survival data was reported as the percentage of fish that died within a given age bracket with 95% confidence intervals. A Z-test for two proportions was performed to determine the significance of survival data at 99% confidence. All P values reported are two-tailed and the significance was set at P<0.05.
We thank Tera Lindquist for the motor axon analysis and the fish facility staff for excellent fish care. This work was supported by the ALS Association (grant number 822 to C.E.B. and A.H.M.B.) with additional support from The National Institutes of Health (grant number RO1NS050414 to C.E.B. and P30NS045758), Dave Eltschlager ALS fund, ACM business solutions and Accuscan Instruments, Inc. Deposited in PMC for release after 12 months.
↵* These authors contributed equally to this work
The authors declare no competing financial interests.
T.R. initiated the project, designed and made the transgenic DNA constructs, and generated the lines. He also developed the protocols for analysis, performed the initial characterization, and designed and developed the behavioral analysis system for measuring the swim strength. A.N.L. performed all of the western blots and qPCR. She also helped analyze the antibody labeling, performed the swim tunnel tests, and was involved in the muscle physiology. R.H.P. performed antibody labeling for synapses and the following analysis. C.W. helped to identify and establish the transgenic lines. P.M.L.J. and B.D.C. performed the muscle physiology. A.H.M.B. contributed to the design of the transgenic constructs. C.E.B. oversaw the project, analyzed all data and wrote the manuscript.
Supplementary material for this article is available at http://dmm.biologists.org/lookup/suppl/doi:10.1242/dmm.005538/-/DC1
- Received February 8, 2010.
- Accepted March 20, 2010.