accordion, a zebrafish behavioral mutant, has a muscle relaxation defect due to a mutation in the ATPase Ca²⁺ pump SERCA1

Several factors have made the zebrafish an attractive vertebrate model for the study of motor development. First, forward genetics can be applied to zebrafish to identify genes that are essential for proper behaviors(Granato et al., 1996). Second, embryos are transparent, allowing the visualization of dynamic events,such as Ca²⁺ transients and axonal outgrowth in live fish(Fetcho and O'Malley, 1997; Lewis and Eisen, 2003). Third,electrophysiological techniques can be used to analyze the physiology of embryonic neurons and muscles to dissect embryonic behaviors(Drapeau et al., 2002). Fourth, zebrafish embryos exhibit readily assayable and well-characterized behaviors (Eaton and Farley,1973; Saint-Amant and Drapeau,1998).

Zebrafish embryos display three stereotyped behaviors by 36 hours post-fertilization (hpf) (Saint-Amant and Drapeau, 1998). The earliest behavior consists of repetitive spontaneous alternating coiling of the tail, much like a metronome. This simple slow coiling behavior is independent of stimulation and starts at 17 hpf, reaches a peak frequency of 1 Hz at 19 hpf, and declines to less than 0.1 Hz by 26 hpf. After 21 hpf, embryos start to respond to mechanosensory stimulation with coils that are stronger than spontaneous coils. Typically embryos respond with two alternating coils. By 26 hpf mechanosensory stimulation initiates swimming episodes. The frequency of muscle contractions during swimming increases from 7 Hz at 26 hpf to 30 Hz at 36 hpf, the latter being similar to the frequency of swimming of adult zebrafish(Buss and Drapeau, 2001).

The production of any touch response can be divided into several steps from sensory perception to muscle activation (reviewed by Roberts, 2000; Berchtold et al., 2000). In zebrafish, two types of mechanosensory neurons sense touch stimuli. Head and yolk stimulation acts through trigeminal neurons, whereas tail stimulation activates Rohon-Beard neurons (Drapeau et al., 2002). Once triggered by sensory input, interneuronal networks located in the hindbrain and spinal cord produce the appropriate motor rhythm (Fetcho, 1992). Motoneurons are activated by these central networks, release acetylcholine at neuromuscular junctions (NMJs) and depolarize the muscle membrane(Buss and Drapeau, 2002; Li et al., 2003). In muscle,depolarization is sensed by voltage-dependent L-type Ca²⁺ channels,such as dihydropyridine receptors, and activates ryanodine receptors in the sarcoplasmic reticulum (SR) membrane to release Ca²⁺ from the SR to the cytosol (Fill and Copello,2002). Increased cytosolic Ca²⁺ activates troponin C that initiates actin/myosin sliding, thus contracting the muscles(Gordon et al., 2000). Cytosolic Ca²⁺ levels are rapidly decreased by sarco(endo)plasmic reticulum Ca²⁺-ATPase 1 (SERCA1), a Ca²⁺ pump expressed in skeletal muscle, which enables relaxation (McLennan et al., 1997).

Any defect in the above pathway induces motor disorder in human. For example, impairment of Ca²⁺ regulation by SERCA1 mutation in human causes Brody disease, a rare inherited disorder of skeletal muscle function (Odermatt et al.,1996). The clinical signs and symptoms in individuals involve exercise-induced impairment of muscle relaxation, stiffening and cramps(Brody, 1969; MacLennan, 2000). In fact, in vitro study with cultured muscle cells from individuals with this disease revealed slow Ca²⁺ reuptake and reduced Ca²⁺-ATPase activity that can explain the relaxation defect (Benders, 1994).

The large-scale Tübingen mutagenesis screen isolated 63 behavioral mutants that displayed abnormal touch responses between 48 and 60 hpf(Granato et al., 1996). Some of the locomotion mutants, raised in this screen as well as the others, were further characterized. Reduced touch sensitivity mutants such as macho,alligator and steifftier had defects of Na⁺ channel function in mechanosensory Rohon-Beard neurons(Ribera and Nüsslein-Volhard,1998). shocked mutants, which showed abnormal swimming,were defective in motor processing in the central nervous system (CNS)(Cui et al., 2004). Reduced or abnormal motility mutants diwanka, unplugged and space cadethad defects in axon projections (Zeller and Granato, 1999; Zeller et al., 2002; Zhang and Granato,2000; Zhang et al.,2001; Lorent et al.,2001). Immotile or reduced motility mutants, such as nic/twister, sofa potato, ache/zieharmonika and twitch once,were defective at the NMJ (Westerfield et al., 1990; Lefebvre et al.,2004; Ono et al.,2001; Behra et al.,2002; Downes and Granato,2004; Ono et al.,2002). Another immotile mutant, relaxed, had a coupling defect between muscle excitation and contraction(Ono et al., 2001). Mutants carrying reduced striations in somatic muscles, such as fibrils unbundeled and sapje, were shown to have abnormal myofiber organization (Felsenfeld et al.,1990; Bassett et al.,2003). The genes mutated in the mutants nic/twister, sofa potato, ache/zieharmonika, twitch once and sapje were,respectively, identified as the α-subunit of the nicotinic acetylcholine receptor, the δ-subunit of the nicotinic acetylcholine receptor, acetylcholinesterase, rapsyn and the Duchenne muscular dystrophy gene (Sepich et al., 1998; Lefebvre et al.,2004; Ono et al.,2004; Behra et al.,2002; Downes and Granato,2004; Ono et al.,2002; Bassett et al.,2003).

Another zebrafish mutation, accordion (acc), exhibited apparent simultaneous contractions of trunk muscles on both sides of the embryo resulting in the shortening of the trunk in response to touch and was named in analogy to the action of the musical instrument(Granato et al., 1996). As application of a glycine receptor blocker, strychnine, to wild-type embryos phenocopied the abnormal behavior, it was proposed that the accphenotype was due to defects in glycine-mediated transmission within the CNS(Granato et al., 1996). Agreeing with this anticipation, inhibitory glycine-mediated transmission is essential to prevent simultaneous, bilateral muscle contractions in a variety of vertebrates (Fetcho, 1990; Grillner, 2003; Kudo et al., 2004; Roberts, 2000). However, the acc phenotype can, in principle, be generated in a variety of other ways. As muscle defects can result in abnormal behavior(Bassett et al., 2003; Granato et al., 1996; Lefebvre et al, 2004; Ono et al., 2002), one of these ways could be via a muscle defect that increases the duration of muscle contractions without any defects in the alternation or timing of output from motoneurons in the two sides of the spinal cord. Such a defect would result in overlap of contractions on the two sides and shortening of the trunk.

In this paper, we have used electrophysiology to show that the output of the CNS and the function of the NMJ are normal in acc. However, after active contraction, acc truck muscles stiffened for a while and relaxation was five times slower than that in wild-type siblings. There was a correlated slowing of the decay of cytosolic Ca²⁺ transients in mutant muscles. These results suggested impaired Ca²⁺ re-uptake into the SR from the cytosol in mutant muscles. Indeed, application of thapsigargin, an inhibitor of SERCA, to wild-type embryos phenocopied acc defects. We identified mutations in the atp2a1 gene,which encodes SERCA1, of acc mutants. The molecular identification of the atp2a1/acc gene was confirmed by rescue of the mutant phenotype by injection of wild-type atp2a1 mRNA into mutant embryos and by phenocopy of the mutation by injection of antisense atp2a1 morpholino oligonucleotides (MO) into wild-type embryos. Mutations in ATP2A1 in humans are known to cause Brody disease, an exercise-induced muscle relaxation disorder (Brody, 1969; Odermatt et al., 1996), making the acc mutant an attractive animal model for Brody disease.

Zebrafish were bred and raised according to established procedures(Westerfield, 1995; Nüsslein-Volhard and Dahm,2002), which meet the guidelines set forth by the University of Michigan animal care and use protocols. acc^tq206 was generated from the large-scale Tübingen screen(Granato et al, 1996). acc^mi25i and acc^mi289a were isolated in our behavioral screen. Zebrafish were mutagenized by following published procedures (Mullins et al.,1994).

Embryonic behaviors were observed and video recorded using a dissection microscope. Mechanosensory stimuli were delivered to the yolk with forceps. For closer examinations of muscle contractions, 24 hpf embryos were anaesthetized (0.02% tricaine) and pinned with tungsten wires (25 μm diameter) through the notochord onto a silicon elastomer (Sylgard)-lined dish in Evans solution (134 mM NaCl, 2.9 mM KCl, 2.1 mM CaCl₂, 1.2 mM MgCl₂, 10 mM glucose, and 10 mM HEPES at 290 mOsm and pH 7.8). Tricaine was removed by replacing the tricaine/Evans solution with Evans with no tricaine. Mechanosensory stimuli were delivered to the yolk by ejecting bath solution (20 psi, 20 mseconds pulse) from a pipette with a 15-30 μm tip using a Picospritzer II (General Valve Corporation). Several somites were observed in high resolution with Hoffman modulation optics (40× water immersion objective). Videos were captured with a CCD camera (WVBP330,Panasonic) and a frame grabber (LG-3, Scion Corporation) and analyzed with Scion Image on a G4 Macintosh. The contraction and relaxation durations were measured by counting frames from the videos (1 frame=33 mseconds).

Embryos at 24 hpf were anaesthetized and pinned to a Sylgard dish in Evans solution. Tungsten wires were put through the forebrain, because the intact forebrain and midbrain are not necessary for the touch response and spontaneous contractions at 24 hpf(Saint-Amant and Drapeau,1998). Tricaine/Evans was replaced with Evans containing no tricaine and mechanosensory stimuli delivered to the tail for the touch response. Videos were captured and analyzed with Scion Image. The durations for active contraction, constant contraction and relaxation were measured by counting frames (1 frame=33 mseconds).

The dissection protocols for in vivo patch recordings have been described previously (Buss and Drapeau,2000). Briefly, 48 hpf zebrafish embryos were anaesthetized in tricaine and pinned on a Sylgard dish. The skin was peeled off to allow access to the underlying muscle cells. For electrophysiological recordings, embryos were partially curarized in an Evans solution containing 6 μM d-tubocurarine but not tricaine. Patch electrodes were pulled from borosilicate silicate glass to yield electrodes with resistances of 6-10 MΩ. The patch electrode was visually guided to patch muscle cells using Hoffman modulation optics (40× water immersion objective). The electrode solution was consisted of 105 mM potassium gluconate, 16 mM KCl, 2 mM MgCl₂, 10 mM HEPES, 10 mM EGTA and 4 mM Na₃ATP at 273 mOsm and pH 7.2. Recordings were performed with an Axopatch 200 amplifier(Axon Instruments), low-pass filtered at 5 kHz, and sampled at 10 kHz. Data was collected with Clampex 8.2 software (Axon Instruments) and analyzed with Clampfit 9.0 software (Axon Instruments). Mechanosensory stimulation to induce fictive swimming was delivered by ejecting bath solution (20-50 psi, 10-30 mseconds pulse) from a pipette with a 15-30 μm tip using a Picospritzer II to the tail of the pinned embryo.

As described previously, it is possible to inject current through a patch to evoke muscle contractions in live zebrafish and still maintain the whole cell configuration (Buss and Drapeau,2000). Embryos were completely curarized with an Evans solution containing 15 μM d-tubocurarine. The current intensity and the current pulse duration were set to 2-7 nA and 15-20 mseconds, respectively, which were sufficient to evoke a contraction. This condition mimics the cycle duration of motoneurons output that must be occurring for the embryo to swim at the observed tail beat frequencies of 30 Hz(Buss and Drapeau, 2001). The contractions of the muscle were observed on a video monitor and captured with a CCD camera to analyze with a computer.

Thapsigargin (10 mM stock in DMSO, Sigma) was diluted to 1 μM in Evans solution and applied to embryos 1 hour before assay.

We injected Calcium Green-1 dextran (10,000 M_r,Molecular Probes) into 1- to 4-cell stage embryos from acc carrier in-crosses. At later stages, random subsets of cells including muscles were seen to contain Calcium Green-1 dextran. The genotype of the injected embryos(24 and 48 hpf) was determined by responses to touch and the embryos were anaesthetized with 0.02% tricaine and pinned to a Sylgard dish. The tricaine was washed out and embryos were bathed in Evans solution containing a muscle myosin inhibitor, N-benzyl-p-toluene sulphonamide at 5 mM (Sigma), which does not affect to the Ca²⁺ regulation(Cheung et al., 2001). This drug suppressed the muscle contractions and enabled us to observe the Ca²⁺ transients with a confocal microscope (LSM510, Carl Zeiss). As plane-scanning requires too much time to adequately measure Ca²⁺transients, we performed line-scanning at 132 Hz (7.6 mseconds/scan) of a fluorescent muscle cell. The time course of the average strength of fluorescence in the line at each scan was calculated with the maximum fluorescence of each record equalized. The observed changes in fluorescence indicate Ca²⁺ transients during spontaneous coiling at 24 hpf. N-methyl-D-aspartate (NMDA, 50 μM) was bath applied to 48 hpf embryos to activate swimming (Cui et al.,2004), as spontaneous movement was hardly observed at 48 hpf(Saint-Amant et al., 1998).

acc carrier fish (Tübingen strain) were crossed with wild-type WIK fish to generate mapping carriers that were crossed to identify mutants for meiotic mapping to microsatellites and genes using PCR as described previously(Nüsslein-Volhard and Dahm,2002; Gates et al.,1999; Shimoda et al.,1999). For mapping acc against netrin 1, a polymorphism of the DraI restriction site in a 136 bp PCR product(forward primer, 5′-GCATATGAAATATGATAGAGGAAGATT-3′; reverse primer, 5′-TGAAACAAGTCTGCTTGAACCAG-3′) found in the map crosses was used. The LN54 radiation hybrid panel was used for the physical mapping(Hukriede et al., 1999).

The following primers were used for cloning of ATP2A1 cDNA:forward primer, 5′-GGGGCCATCTGTTTTGTCCCTATCCTTCAC-3′; reverse primer, 5′-GGGTGATTTTTCACTGGTTCCCGGTTGG-3′. To knock down the protein synthesis (Summerton and Weller,1997), SERCA1 antisense MO(5′-CGTTCTCCATCCTGTCTGCTCAAAG-3′) was designed against 16 mer of 5′-UTR sequence and adjacent 9 mer coding region on mRNA. The CAT sequence in bold corresponds to the start ATG. Standard control MO(randomized sequence available from Gene Tools) was used for control MO. Capped atp2a1 mRNA synthesis was performed with mMESSAGE mMACHINE T7 Ultra (Ambion). The 5′-UTR sequence of the synthesized mRNA was amended(5′-CCTCGAGCCGCCACCATGGAGAACGCA-3′) in order to avoid knockdown by the antisense MO. The ATG sequence in bold is the start ATG. Injections (0.25 mM for MO and 1 μg/μl for mRNA) were performed as described previously (Xiao et al.,2003; Nüsslein-Volhard and Dahm, 2002). The Pro789Leu mutation (CCA to CTG), previously identified as a mutation underlying Brody disease(Odermatt et al., 2000), was introduced into the zebrafish atp2a1 construct by the Quick Change method (Stratagene) with the following primers: forward primer,5′-CCTGAGGCTCTGATCCTGGTTCAGCTGCTGTGG-3′; reverse primer,5′-CCACAGCAGCTGAACCAGGATCAGAGCCTCAGG-3′. The CTG and CAG sequence in bold indicates the sense and antisense codon,respectively, of the introduced Leu residue.

Immunostaining was carried out as described previously(Devoto et al., 1996). Briefly, embryos at 24 hpf were anesthetized (0.02% tricaine), fixed in 4%paraformaldehyde and then washed several times in 0.1 M phosphate-buffered saline (PBS), pH 7.4. Embryos were blocked in PBS containing 2% BSA and 2%sheep serum, incubated in F59 antibody (anti-slow twitch myosin, 1:20,Developmental Studies Hybridoma Bank) diluted in PBS, and washed in PBS. Alexa 488-conjugated anti-mouse IgG was used as a secondary antibody (1/1000,Molecular Probes). In situ hybridization was performed as described previously(Li et al., 2004).

Mixture of two atp2a1 probes covering nucleotides 1∼997 (N terminus∼M4 domain in SERCA1 protein) and 2598∼3220 (M7∼C terminus) were used for in situ hybridization. For sectioning after color development, embryos were equilibrated in 15% sucrose/7.5% gelatin in PBS at 37°C and then embedded in it at –80°C. Sections (20 μm) were cut with a cryostat (CM3050S, Leica).

As part of an ongoing ENU mutagenesis screen, we isolated two new alleles(mi25i and mi289a) of acc that were autosomal recessive and lead to death by 10 days (not shown). We performed phenotypic analysis on the original acc allele (tq206) from Tübingen(Granato et al., 1996),because it displayed the strongest phenotype as described below. When wild-type embryos were touched with forceps at 24 hpf, they reacted with two or three fast, alternating coils of the tail that immediately followed each other (Fig. 1A, see Movie 1 in the supplementary material). By contrast, acc embryos appeared to simultaneously contract the trunk axial muscles on both sides and this resulted in a dorsal flexure of the trunk(Fig. 1B, see Movie 2 in the supplementary material). The mutants stayed in a contracted state long after wild-type embryos have relaxed. To examine a muscle movement at higher resolution, we observed a contraction of several somites in embryos loosely pinned on a Sylgard dish and stimulated to evoke a contraction. Although wild-type embryos displayed quick muscle contractions (0.61±0.03 seconds, n=5, see Movie 3 in the supplementary material), acc embryos showed significantly slower contractions(4.80±0.28 seconds, n=5, see Movie 4 in the supplementary material). At 48 hpf, wild-type siblings swim away when touched, whereas acc mutants failed to swim (data not shown). The acc larvae died at 7-10 days post-fertilization, possibly from the inability to swim and feed effectively.

To compare the three alleles of acc mutants with wild type, we measured the time course of active contraction when muscles are shortening,constant contraction when muscles are contracted but not changing in length,and relaxation when muscles are lengthening in responses to mechanosensory stimulation at 24 hpf (Table 1). The duration of the active contraction phase was not significantly different between wild-type siblings and acc mutants,whereas the relaxation phase took five times longer in all the alleles of acc compared with wild-type siblings. Furthermore, the constant contraction period was significantly longer (1.7∼2 seconds) in acc embryos, whereas this stiffness was not observed (less than 0.1 second) in wild-type siblings. Among three alleles of acc mutants,tq206 showed the strongest phenotype in terms of the relaxation kinetics.

Mutant embryos spontaneously coiled at 24 hpf with normal frequency(0.27±0.02 Hz for wild type, n=10; 0.25±0.02 Hz for acc, n=10), but the kinetics of coiling was aberrant. The duration of the active contraction phase was not significantly different between wild-type siblings and acc, but the relaxation phase and constant contraction period were significantly longer in acc than in wild-type siblings(Table 1). Thus, the mutation appears not to perturb the ability to spontaneous coil, but rather results in prolonged coils. These observations suggest that rhythmic output from the CNS underlying spontaneous coiling is not perturbed in acc embryos but muscle relaxation is aberrant in mutants.

A significant increase in the duration of relaxation in accembryos should decrease the stimulation frequency at which tetanus occurs. To test this, we stimulated muscles by injecting current with a patch electrode and video recorded the resultant contractions in mutant and wild-type siblings at 48 hpf. Wild-type muscles followed trains of depolarizations at frequencies from 1 to 35 Hz (n=5), whereas acc muscles followed at lower frequencies but failed to follow trains of depolarizations at 15±1.4 Hz(n=5), showing that acc muscles do reach tetanus at lower stimulation frequencies compared with wild-type muscles. This result, along with the relaxation data, strongly suggests that one site of defect in the mutants is muscle and that the overlap in muscle contractions on the two sides due to the slow relaxation causes the shortening of the trunk seen in acc embryos during an escape response following a mechanosensory stimulus.

In addition to behavioral defects, acc embryos exhibited morphological defects (Granato et al.,1996). After 48 hpf, the trunks of acc embryos were frequently bent (Fig. 2A,B) and this phenotype worsened with age (not shown). Disruption of the notochord was also observed after 36 hpf (Fig. 2D,E). Labeling of slow twitch muscle fibers in acc with mAb F59 (Devoto et al, 1996)showed that they were disturbed (Fig. 2G,H), but no increased cell death assayed with TUNEL and Acridine Orange labeling was observed (data not shown). These morphological defects were probably secondary ones resulting from the mechanical stress caused by the aberrant contractions. Supporting this idea, prolonged treatment with tricaine, a weak Na⁺ channel inhibitor that prevented motor behavior, abolished the morphological defects in acc mutants(Fig. 2C,F,I). Treatment with N-benzyl-p-toluene sulphonamide, a specific inhibitor of muscle myosin, also prevented the morphological defects (data not shown).

The production of any motor behavior can be divided into several steps,such as sensory perception, production of rhythmic activity in the CNS,synaptic transmission at the NMJ, Ca²⁺ regulation in muscle cell,and contraction of muscle (reviewed by Roberts, 2000; Fetcho, 1992; Berchtold et al., 2000). The fact that muscle relaxation is much slower in acc embryos suggests that one site of defect is muscle. However, it remains possible that processes preceding muscle relaxation may also be defective. As application of strychnine, a blocker of glycine-mediated inhibitory synaptic transmission,phenocopies acc-like bilateral contractions following mechanosensory stimulation (Granato et al,1996), one possibility is that inhibitory interactions within the CNS are also defective in acc. To examine this, we demonstrated both CNS morphology and physiology of mutants at the NMJ.

The overall morphology of the CNS, including the hindbrain and spinal cord,which are necessary and sufficient for the early behaviors in zebrafish, was not obviously perturbed in acc embryos (data not shown). The pattern of the hindbrain reticulospinal interneurons, spinal Rohon-Beard sensory neurons, and spinal motoneurons and their axons labeled with anti-acetylatedα-tubulin (which labels all axons)(Chitnis and Kuwada, 1990; Bernhardt et al., 1990) and anti-Islet (which labels Rohon-Beard and motoneurons nuclei)(Korzh et al., 1993) were normal (data not shown). Furthermore, the pattern ofα-bungarotoxin-labeled acetylcholine receptor on muscles(Ono et al., 2001) was normal(data not shown).

The physiological output from the CNS at the NMJ was assayed by in vivo patch recordings from trunk axial muscle of 48 hpf embryos(Fig. 3A). Muscle voltage recordings from wild-type sibling embryos showed sustained episodes of rhythmic depolarizations following mechanosensory stimulation(Fig. 3B). This rhythmic activity had a frequency (25 to 30 Hz) that was comparable with the swimming frequency observed during free swimming and presumably underlies swimming contractions (Buss and Drapeau,2001). The average duration of the muscle response in wild-type siblings was 1.28±0.22 seconds (n=10) and maximum amplitude was –54.1±1.0 mV. When strychnine (2 μM) was added to the bath solution, the pattern of activity in the muscle following mechanosensory stimulation dramatically changed so that the responses were shorter(0.29±0.03 seconds, n=10, Student's t-test, P<0.01) and lost their fast rhythmic activity, although the maximum amplitude (–54.9±1.0 mV) was comparable with wild-type siblings (Fig. 3C). Muscle recordings from acc embryos following mechanosensory stimulation showed a rhythmic voltage response in which the duration (1.38±0.22 seconds, n=10), maximum amplitude (–54.8±0.8 mV) and rhythmicity (25-30 Hz) were indistinguishable from that seen in wild-type siblings (Fig. 3D). Thus, the muscle recordings in acc embryos are similar to that in wild-type siblings and do not resemble the output of the CNS when glycine-mediated inhibitory transmission is blocked by strychnine. These results are consistent with the hypothesis that the output from the CNS and the response of the NMJ following mechanosensory stimulation is unperturbed in mutants and that the defect resides in muscle.

Depolarization of the muscle membrane leads to a transient increase of cytosolic Ca²⁺ that cause contraction and relaxation of muscle by actin/myosin sliding (Berchtold et al.,2000). As there is a muscle relaxation defect in accmutants, we examined whether the regulation of intracellular Ca²⁺is defective in mutant muscles. To visualize Ca²⁺ levels in live zebrafish, embryos were injected with Calcium Green-1 dextran, pinned on a Sylgard dish and imaged during a spontaneous coiling at 24 hpf by using a confocal microscope (Fig. 4A). As plane-scanning takes too long (>0.3 seconds/scan) to capture the Ca²⁺ transients, we performed line-scanning of muscles at 132 Hz(7.6 mseconds/scan, see Materials and methods). The time course of Ca²⁺ increase in acc muscle was indistinguishable from that of wild-type muscles, but the decay was significantly slower in the mutants (Fig. 4B,C). The time to decrease to half amplitude of Ca²⁺ (half decay) was seven times longer in acc (2.47±0.37 seconds, n=28) than in wild type (0.35±0.03 seconds, n=28; Student's t-test, P<0.01). We also examined Ca²⁺ transients in muscles of 48 hpf embryos during swimming (data not shown). Similar to Ca²⁺transients during spontaneous coiling, the half decay of Ca²⁺during swimming was significantly slower in acc muscles compared with that of wild-type siblings (acc, 0.94±0.06 seconds, n=10; wild-type siblings, 0.08±0.01 seconds, n=10;Student's t-test, P<0.01). Thus, the clearance of Ca²⁺ from the cytosol is much slower in acc muscles compared with that in wild-type muscles, as predicted by the differences in relaxation kinetics between mutant and wild-type muscles. As Ca²⁺removal from the cytosol is accomplished mainly by SERCA1, a Ca²⁺pump found in the SR that enables relaxation(MacLennan et al., 1997), our results are consistent with a defect in SERCA1 in acc embryos.

If SERCA1 is mutated in acc embryos, thapsigargin, a specific blocker of SERCA (Treiman et al.,1998), should phenocopy the mutation. Treatment of 24 hpf wild-type embryos with 1 μM thapsigargin resulted in touch-induced muscle contractions on the two sides of the trunk to overlap just as in accembryos (data not shown). In addition, stimulation of muscle contraction by direct current injection revealed that thapsigargin-treated wild-type muscle cells failed to follow the stimulation at frequencies above 15 Hz(n=2), much like acc mutants. The Ca²⁺ imaging results and phenocopy by thapsigargin are consistent with the notion that a defect in SERCA1 causes the acc phenotype.

To identify the gene responsible for the acc mutation, we meiotically mapped the mutation to two flanking markers, z4161 (6.8 cM, 246 recombinants in 3616 meioses) and netrin 1 (0.2 cM, 5 recombinants in 2448 meioses), in linkage group 3. As the acc phenotype suggested a defect in SERCA1, we physically mapped atp2a1 gene that encodes SERCA1 using the LN54 radiation hybrid panel and found it to be close (4.3 cR)to netrin 1 in linkage group 3(Fig. 5A).

To test whether atp2a1 contains mutations, we cloned and sequenced SERCA1 cDNA, which encodes 997 amino acid residues (GenBank Accession Number AB178533). SERCA1 contained 10 transmembrane domains (M1-M10) as well as a nucleotide-binding and a phosphorylation domain(Fig. 5B). A point mutation in SERCA1 was found in all three alleles of acc. Ile97 in M2 was mutated to Asn in acc^mi25i; Ser766 in M5 to Phe in acc^tq206; and Thr848 in M7 to Ile in acc^mi289a. These missense mutations, which change the hydrophilicity/hydrophobicity properties of the amino acids, were found in transmembrane domains that were completely conserved among a variety of vertebrates (Fig. 5C). Previous structural and mutational analysis has suggested that these membrane domains turn or move during function and that most of the reported point mutations in these membrane domains disrupt or diminish the pump activity (reviewed by Toyoshima and Inesi, 2004; Wuytack et al., 2002). Thus,the acc mutations are likely to disrupt or reduce SERCA1 function.

As SERCA1 is found in the SR of muscle, atp2a1 should be expressed in muscle. Indeed, in situ hybridization showed that atp2a1 is expressed by axial muscles as early as 15 hpf, which precedes the appearance of motor behaviors (Fig. 6A,B). Cross-sections of the trunk showed that expression appears to be exclusively in muscles and not in the CNS (Fig. 6C). No expression of atp2a1 in the CNS was observed at any level of the CNS (data not shown). Although expression of atp2a1in the CNS below the resolution of our in situ hybridizations cannot be excluded, muscle specific expression of ATP2A1is consistent with the primary acc defect found in muscles.

The molecular identification of acc as atp2a1 encoding for SERCA1 was confirmed by mutant rescue and antisense phenocopy. We injected wild-type SERCA1 mRNA, SERCA1 mRNA carrying the acc^tq206mutation, or mRNA carrying a mutation at Pro789 to Leu [which is a cognate mutation reported in human Brody disease (see Discussion)] into recently fertilized embryos of acc heterozygous carriers. Nearly all the progeny injected with wild-type mRNA displayed normal coiling following touch at 24 hpf (Fig. 7B, 97.5%,157/161), whereas only a few responded with the mutant phenotype (2.5%,4/161). Approximately 25% of uninjected progeny(Fig. 7A, 28.6%, 16/56), acc^tq206 mRNA-injected progeny(Fig. 7C, 23.2%, 29/125), and the Brody disease mRNA-injected progeny(Fig. 7D, 26.5%, 43/162) showed the mutant response, while the rest responded normally. As injection of mutant forms of the SERCA1 mRNA did not induce a higher incidence of embryos to respond aberrantly, they appear not to have a dominant effect. Because one would expect 25% of progeny to be mutant from a cross of heterozygous carriers, these results are consistent with rescue of normal touch responses by injection of wild-type mRNA. Furthermore, 35/157 (22.5%) of wild-type mRNA-injected embryos that responded normally at 24 hpf displayed the mutant response after 48 hpf, suggesting that the rescue by wild-type mRNA was transient (data not shown). This is probably due to degradation of the injected mRNA. As ∼25% of the progeny recovered mutant responses, they were likely to be homozygous embryos that were rescued transiently by the wild-type mRNA.

To knockdown synthesis of SERCA1, antisense MO against SERCA1 mRNA was injected into recently fertilized wild-type embryos. Antisense MO-injected wild-type embryos displayed the acc behavior after touch(Fig. 7F, 92%, 33/36), whereas control MO-injected wild-type embryos did not(Fig. 7E, 0%, 0/17). The antisense MO was designed to block translation of endogenous SERCA1 mRNA but not the injected SERCA1 mRNA, owing to engineered differences in the 5′-untranslated sequence of the injected mRNA (see Materials and methods). When the antisense MO was co-injected with the engineered wild-type SERCA1 mRNA, the antisense MO phenotype was blocked and embryos responded normally to touch (Fig. 7G,95%, 37/39). The inhibition of the SERCA1 knockdown phenotype by wild-type mRNA was specific, as embryos injected with both antisense MO and acc^tq206 mRNA showed acc behavior(Fig. 7H, 88%, n=28/32). Thus, wild-type SERCA1 mRNA rescue of acc mutants and antisense knockdown of SERCA1 confirm that atp2a1, encoding the SR Ca²⁺ pump SERCA1, is the responsible gene for accmutation.

Several findings demonstrate that acc embryos have a muscle relaxation defect. First, the kinetic analysis of the touch response and spontaneous coiling showed that the constant contraction and relaxation phases were significantly longer in acc embryos. Second, direct electrical stimulation of muscle demonstrated that acc muscles failed to follow stimulation at much lower frequencies compared with wild-type muscles. Third,in vivo imaging of Ca²⁺ transients revealed that the decay in cytosolic Ca²⁺ during relaxation was significantly slower in acc muscle. Fourth, the acc gene encodes for the muscle-specific SERCA1 Ca²⁺ pump. Therefore, slower reuptake of Ca²⁺ causes muscles to relax much more slowly. This results in slower spontaneous coiling and overlap of muscle contraction on the two sides of the trunk during touch-induced escape responses.

acc was originally proposed to have a neuronal defect because strychnine-treated embryos displayed acc-like bilateral muscle contractions (Granato et al.,1996). Although the primary defect in acc mutants is a muscle relaxation defect caused by mutations in muscle-specific SERCA1, it is possible that there might be a secondary neuronal defect that is due to deficient motility of mutants. So far, we have not detected a neuronal phenotype. We found that the voltage response of wild-type muscles to touch,when glycine-mediated inhibitory synaptic transmission is blocked by strychnine, is highly aberrant. By contrast, acc muscles exhibited normal rhythmic activity following mechanosensory stimulation. This along with the fact that the morphology of the nervous system is not obviously aberrant suggests that many features of the nervous system are normal. However, we cannot rule out a neuronal phenotype secondary to the primary muscle defect in acc embryos.

acc embryos also displayed morphological defects, such as a bent trunk, disrupted notochord and disturbed slow-twitch muscle fibers. These are secondary to the muscle relaxation defect, as paralysis by tricaine or N-benzyl-p-toluene sulphonamide prevented these morphological phenotypes. In fact, similar notochord and slow twitch defects are seen in another behavior mutant, nic^{twister dbn12}, that has a gain-of-function mutation in the α-subunit of the muscle-specific nicotinic acetylcholine receptor (Lefebvre et al.,2004). In nic^{twister dbn12} homozygous mutant,prolonged neuromuscular transmission causes persistent activation of axial muscles bilaterally that results in acc-like hypercontractions and as a consequence cause notochord and slow twitch muscle abnormalities.

We found three missense mutations that disrupt or diminish the SERCA1 function of acc mutants. Each mutation was located in a highly conserved α-helix, which forms a membrane domain. Most mutations introduced in M5 disrupt or severely reduce SERCA1 function in vitro,suggesting an essential role of the M5 domain(Sørensen, et al.,1997; Strock et al.,1998; Asahi et al.,1999; Guerini et al.,2000; Zhang et al.,2000; Ma et al.,2003; Dode et al.,2003). In fact, acc^tq206 contains a mutation in the M5 domain that is consistent with the functional importance of the M5 for SERCA1 function. The structure and function of SERCA1 has been characterized by crystal structures in different states and by numerous point mutations. SERCA1 has two Ca²⁺-binding sites: one consists of M5,M6 and M8; the other consists of M4 and M6(Toyoshima et al., 2000). In M5, both Asn768 and Glu771 form hydrogen bond with Ca²⁺ by their side-chain oxygen atoms. During the conformational change from the Ca²⁺-binding state to the dissociation state, the middle part of the M5 domain called the flexible hinge (Ile765-Asn768) tilts about 30°,whereas the adjacent lower region of M5 (Gly 770-Leu 777) hardly moves at all. This tilt results in turning the upper part of M5 and disrupts the hydrogen bond for Ca²⁺ binding(Toyoshima and Nomura, 2002; Nielsen et al., 2003). The mutation Ser766Phe found in acc^tq206 is located in the hinge region and could account for the strongest phenotype.

During the conformational change of SERCA1, M4 inclines together with the M5 as a rigid body to induce movement of the M2 helix(Toyoshima and Nomura, 2002). This movement by M2 probably regulates the binding of sarcolipin, a regulator of SERCA1 in skeletal muscle (Odermatt et al., 1998; Asahi et al.,2002; MacLennan et al.,2003). Although this binding has not been studied structurally, a binding model between SERCA1 and phospholamban, a sarcolipin-like protein expressed in cardiac muscles (MacLennan and Kranias, 2003), suggests that the hydrophobic residues in the M2, including Ile97, are important for this interaction(Toyoshima et al., 2003; Asahi et al., 2003). The acc^mi25i allele has an Ile97Asn mutation in M2, suggesting that regulation of SERCA1 by sarcolipin may be aberrant in this allele. In contrast to the other membrane domains, M7 does not move much, because the M7 helix associates with the lower region of M5(Toyoshima and Nomura, 2002),and M7 has not been well characterized by mutations. The fact that the acc^mi289a allele has a Thr848Ile mutation in M7 is the first functional evidence for a role of M7 for SERCA1 function.

Brody disease is a rare inherited disorder of skeletal muscle function in humans. It is characterized by painless cramping induced by exercise and linked to an impairment of skeletal muscle relaxation(Brody, 1969). Most families harboring Brody disease show an autosomal recessive inheritance pattern and carry a mutation in the ATP2A1 gene on chromosome 16(Odermatt et al., 1996; Odermatt et al., 1997; Odermatt et al., 2000). In a minority of cases, Brody disease is inherited in an autosomal recessive pattern but there are no mutations in ATP2A1(MacLennan, 2000; Zhang et al., 1995) or in an autosomal dominant pattern with a (2;7) chromosomal translocation(Novelli et al., 2004). Thus,there are at least two genes other than ATP2A1 that can give rise to Brody disease (MacLennan,2000).

Among seven different SERCA1 mutations so far identified in Brody disease,five induce a stop codon 5′ to the Ca²⁺-binding sites and are considered to be null mutations (Odermatt et al., 1996; Odermatt et al.,1997; Odermatt et al.,2000). The other missense mutations are found in M6 and in a loop between the M6 and M7 (Odermatt et al.,2000). Surprisingly, the individuals that carry apparent null alleles of ATP2A1 are able to relax their skeletal muscle albeit at a much slower rate without any obvious respiratory distress(MacLennan, 2000). Interestingly, SERCA1-null mice were defective in diaphragm function and died soon after birth, making them less attractive as an animal model for Brody disease (Pan et al., 2003). Compensatory or redundant mechanism for Ca²⁺ removal from the muscle cytosol in humans but not in mice, such as Na⁺-Ca²⁺ exchangers, Ca²⁺-ATPase in the plasma membrane, Ca²⁺ uptake by mitochondria and/or ectopic expression of other members of the SERCA family might account for the difference in viability between mouse and human(Wuytack et al., 2002).

In zebrafish, seven acc mutations were isolated in the Tübingen screen (Granato et al.,1996) and two in our ongoing screen. All of these mutations are autosomal recessive, suggesting the same inheritance as ATP2A1 Brody disease. Zebrafish acc mutants and human Brody disease share the most crucial feature of aberrantly slow relaxation of muscles because of much slower reuptake of Ca²⁺ from the muscle cytosol to the SR caused by mutations in the atp2a1 gene. Although zebrafish acc mutants die by day 10, their fast development and accessibility up to the time of death allows for detailed physiological analysis of the consequences of atp2a1 mutations. As zebrafish embryos are readily accessible to molecular, genetic, pharmacological and physiological interventions, acc mutants could serve as an attractive animal model for Brody disease. For example, acc mutants could be used to screen for drugs that relieve the muscle defect in vivo by enhancing Ca²⁺ re-uptake or reducing Ca²⁺ release from the SR during a contraction. Interestingly, mutations in at least six genes in addition to acc can give rise to an acc-like behavioral phenotype in zebrafish(Granato et al., 1996). It would be interesting to see if any of these genes might also be genes that are associated with Brody disease.

We thank S. Sprague and L. Crowe for fish care, J. Manuel for generating F2 families, and R. Hume for helpful discussion and generous use of electrophysiology equipment. Supported by grants from the NINDS to J.Y.K. and Michigan Life Science Corridor to D.G. L.S.-A. was partly supported by a Fellowship from FRSQ. H.H. was supported by a Long Term Fellowship of the Human Frontier Science Program (HFSP).