Zebrafish model for congenital myasthenic syndrome reveals mechanisms causal to developmental recovery

Abstract

Mutations in muscle ACh receptors cause slow-channel syndrome (SCS) and Escobar syndrome, two forms of congenital myasthenia. SCS is a dominant disorder with mutations reported for all receptor subunits except γ. Escobar syndrome is distinct, with mutations located exclusively in γ, and characterized by developmental improvement of muscle function. The zebrafish mutant line, twister, models SCS in terms of a dominant mutation in the α subunit (α_twi) but shows the behavioral improvement associated with Escobar syndrome. Here, we present a unique electrophysiological study into developmental improvement for a myasthenic syndrome. The embryonic α_twiβδγ receptor isoform produces slowly decaying synaptic currents typical of SCS that transit to a much faster decay upon the appearance of adult ε, despite the α_twi mutation. Thus, the continued expression of α_twi into adulthood is tolerated because of the ε expression and associated recovery, raising the likelihood of unappreciated myasthenic cases that benefit from the γ−ε switch.

The developmental switch from the embryonic αβδγ to the adult αβδε muscle ACh receptor isoform occurs in every vertebrate species studied to date (1–3). Certain mutations in ε result in the lifelong myasthenic disorder slow-channel syndrome (SCS) (4–6), whereas mutations in the γ subunit are restricted to Escobar syndrome, characterized by improvement in neuromuscular performance during development (7, 8). All documented cases of Escobar syndrome are thought to represent ACh receptor nulls that are corrected by the developmental appearance of the wild-type ε subunit, thereby rescuing receptor expression (9). However, for the most part, there have been no direct analyses of the consequences of the γ-subunit mutations identified in Escobar patients. This stands in sharp contrast to SCS, in which mutations in all subunits other than γ have been studied in depth at an electrophysiological level (6, 10). The consequences for mutations in the α, β, δ, and ε subunits causal to SCS include prolonged channel openings, spontaneous openings, increased desensitization, and increased sensitivity to ACh (6, 10, 11). The absence of equivalent studies for γ-subunit mutations likely is attributable to the recessive inheritance pattern of Escobar syndrome. Thus, the idea that all γ-subunit mutations represent functional nulls remains mostly conjecture. If any gain-of-function mutations in γ were identified, it would point to a common intersection of the two syndromes, with potential therapeutic implications.

Support for the idea of an intersection between SCS and Escobar syndrome was provided by studies of twister zebrafish. These mutant fish exhibit the gain-of-function phenotype characteristic of SCS but undergo developmental improvement characteristic of Escobar syndrome (12). Electrophysiological analysis of twister revealed features common to human SCS, including synaptic prolongation, increased sensitivity to ACh, activation by choline, and spontaneous receptor openings (13). However, within a week of development, the fish showed the ability to swim and eventually could no longer be distinguished from wild type. The assignment of the mutation to the α subunit rather than the γ subunit rendered the γ−ε switch an unlikely candidate for behavioral recovery. Nevertheless, we elected to test this idea directly using electrophysiology.

Materials and Methods

Zebrafish (Danio rerio) were maintained in accordance with the standards set forth in the International Animal Care and Use Committee. The motility mutant twister line carries the formal nomenclature of nic1^{twister dbn12} (12); however, for the purposes of the present study, it is referred to as twi^+/−, which represents heterozygous carriers. The sex of the larva was unknown.

To quantitate behavioral improvement, the swimming was recorded at 1,000 frames per second using a Fastcam 512-PCI camera (Photron Instruments). A tap of the dish was used to elicit the powerful initial contraction (C-bend) and subsequent rhythmic swimming. The timing and strength of both behaviors were quantitated using Flote zebrafish motion analysis software (14).

The methods used to record spontaneous and evoked synaptic currents under whole-cell voltage clamp were identical to those published previously (3, 13, 15). Synaptic currents were recorded using a HEKA Instruments EPC-10/2 amplifier (Instrutech) and 1 μM tetrodotoxin was added for recording spontaneous synaptic currents. Synaptic currents were sampled at 10-μs intervals and filtered at 5 kHz using PatchMaster software (Instrutech). Synaptic current decays were fit offline to the sum of three exponential functions by Igor Pro (WaveMetrics) using the Levenberg-Marquardt algorithm to search for the minimum χ² value. The fit required the entire peak to end range for proper estimation of all three exponential components. All data are presented as mean ± SD, and statistical comparisons were made using paired t tests. All chemicals were all obtained from Sigma Chemical and made fresh in recording solution before each experiment. The ACh receptor antagonist α-bungarotoxin was obtained from Molecular Probes.

The expression levels of ε and γ cDNAs were made using quantitative PCR (qPCR). cDNA was transcribed from RNA extracts of whole embryos between the ages of 24 and 192 h post fertilization (hpf). The transcript level (raw copy numbers) of each subunit was normalized to the endogenous control, elongation factor 1-alpha (elf1a). Elf1a represents a validated endogenous reference gene for qPCR analysis of zebrafish tissue (16).

The γ-subunit antisense morpholino oligonucleotide used in this study was designed by Gene Tools LLC to interfere with proper splicing. The sequence of the γ-subunit morpholino used in this study corresponded to TAATTAAGGCACATACTCACTTCCC. The effectiveness of the morpholino was validated on the basis of PCR (3). For these experiments, 0.5–1.5 nL of γ-subunit morpholino was injected at stock concentrations (1 μM) into fertilized zebrafish eggs at the one- or two-cell stage, and muscle recordings were performed within 72–86 hpf.

Results

To assess swimming in wild-type and twi^+/− zebrafish, we performed a kinematic analysis of the escape response using high-speed video and motion-tracking software. Body curvature, also called total head–tail bend angle, was defined as the sum of the two angles formed between the head and tail segments compared with a middle body segment (14) and was measured in 1-ms intervals. In response to a mechanical tap on the recording chamber, 48 hpf wild-type animals consistently responded with an initial C-bend of the axial musculature followed by a bout of rhythmic swimming (Fig. 1A). By contrast, twi^+/− fish generated a single prolonged C-bend at 48 hpf, impeding the onset of rhythmic swimming (Fig. 1B). At 72 hpf, twi^+/− fish exhibited faster recovery from the C-bend that allowed a small contralateral bend (Fig. 1B). However, at 120 hpf, twi^+/− fish were able to recover from the C-bend rapidly enough to generate rhythmic swimming (Fig. 1B). The ability to generate alternating contractions continued to improve gradually, and at 192 hpf the swimming was indistinguishable between mutant and wild-type fish (Fig. 1 A and B).

Fig. 1.

Stimulus-induced escape response in wild-type and twi^+/− animals at developmental time points. (A) Escape responses in wild-type animals at 48, 72, 120, and 192 hpf (upper to lower panels, respectively). Each response was characterized by an initial C-bend that exhibited the highest degree of curvature (167.3 ± 28.7°, n = 9) followed by a bout of rhythmic swimming. The tail-beat frequency and amplitude were lower at 48 hpf (Upper) because the animal had not yet developed a swim bladder and was resting on the bottom of the recording dish. (B) Escape response in twi^+/− animals at 48, 72, 120, and 192 hpf (upper to lower panels, respectively). At 48 hpf, the response consisted of a single contraction that lasted an average of 703 ± 291 ms (n = 25). By 72 hpf, the initial contraction was shorter in duration and the animal was able to perform a counterbend by contracting the musculature on the alternate side of the body. At 120 hpf, the animal performed a C-bend and counterbend similar to that of wild-type animals, along with slow rhythmic swimming. At 192 hpf, the mutant and wild-type fish were indistinguishable.

To investigate the mechanism surrounding the observed behavioral recovery, we assessed the active and passive properties of twi^+/− and wild-type muscle membranes between 48 hpf and 120 hpf. The input resistance and membrane time constants for muscle were determined using subthreshold depolarizations under a whole-cell current clamp. The input resistance of wild-type muscle was not significantly different from that of twi^+/− muscle at either time point, and both showed significant, but equivalent, drops in input resistance (47% for twi^+/− and 38% for wild-type fish) between 48 hpf and 120 hpf. Likewise, the membrane time constants were not significantly different between twi^+/− and wild-type fish and neither changed between 48 hpf and 120 hpf. Next, the excitable properties of the muscle were determined by delivering current pulses that were just above threshold for generation of an action potential (AP). Measurements of the major derivatives (change in voltage over change in time [dV/dt]) associated with the AP provided comparisons of the associated active currents underlying depolarization and repolarization. The average values of dV/dt associated with the spike generation were similar for wild-type and twi^+/− fish, and neither showed developmental change over this period. However, the current associated with AP repolarization increased significantly for both wild-type and twi^+/− fish (160% for twi^+/− and 183% for wild-type). The developmental change in repolarization likely contributes little to recovery, as it does not enable the twi^+/− fish to generate multiple spikes. Thus, we turned to developmental alterations in synaptic currents, and associated prolonged depolarization, as causal agents for recovery.

Whole-cell voltage clamp recordings from twi^+/− fast muscle showed greatly prolonged spontaneous synaptic currents compared with wild-type fish. In twi^+/− fish, synaptic current decay gradually accelerated over the developmental period associated with behavioral recovery (Fig. 2). Synaptic currents from wild-type fish, however, exhibited no developmental differences in decay. Moreover, at all stages tested, the synaptic currents in wild-type muscle decayed along a single exponential time course (Table 1) owing to the fact that, unlike mammals and amphibians, the open burst durations for αβδε and αβδγ receptors are similar in zebrafish (3, 13). By contrast, the synaptic currents in 48-hpf twi^+/− fish, corresponding to the peak of the behavioral deficit, required the sum of three exponentials for fit (Table 1 and Fig. 3 A–C). Published single-channel studies on twister have shown that the fast component (τ_f) corresponds to an early developmental mixture of wild-type αβδε and αβδγ receptors, the intermediate component (τ_i) to α_twiβδε receptors, and the slow component (τ_s) to α_twiβδγ receptors (13). Whether receptors containing a single mutant α subunit exhibit kinetics different from those of dual mutant α subunits remains an open question. In any case, the open burst duration for α_twiβδε differs from that of α_twiβδγ receptors, pointing to a unique interaction with γ not seen with ε. Thus, at 120 hpf, when most mutant receptors are expected to contain ε and not γ, the decay of synaptic currents in twi^+/− fish might be well fit by two exponentials, corresponding to τ_f and τ_i components seen at earlier developmental stages (Fig. 3C). The τ_s component was significantly reduced or absent in all fish tested. However, to directly compare all developmental stages, as well as to obtain unbiased time constants and amplitudes for each component, we adhered to fitting the twi^+/− synaptic currents at all stages with three exponential components. Fit in this manner, the mean time constants for each component were similar at all stages, but the relative contributions differed (Fig. 3H and Table 1). The fractional contribution by τ_f and τ_i to overall current amplitude increased slightly, but the fractional contribution by the τ_s component exhibited a highly significant downward shift (Fig. 3G and Table 1).

Fig. 2.

Developmental acceleration of synaptic current decay in twi^+/− fish. Three normalized representative mEPCs from twi^+/− fish are shown for each of the indicated developmental time points. A normalized mEPC from a 48-hpf wild-type fish is shown for comparison. The synaptic currents were aligned at the rising phase.

Table 1.

Time constants and fractional contribution by individual components of synaptic decay for wild-type and twi^+/− animals during development

Age, hpf	Wild-type τ decay, ms (n)	twi+/− τf decay, ms (n)	twi+/− τi decay, ms	twi+/− τs decay, ms	twi+/− τf fraction	twi+/− τI fraction	twi+/− τs fraction
48	0.64 ± 0.23 (9)	1.06 ± 0.29 (14)	8.61 ± 2.53	70.21 ± 22.03	0.45 ± 0.09	0.40 ± 0.09	0.15 ± 0.08
72	0.58 ± 0.10 (21)	0.78 ± 0.36 (22)	7.18 ± 1.56	69.81 ± 29.67	0.40 ± 0.08	0.51 ± 0.10	0.09 ± 0.04
120	0.51 ± 0.03 (5)	0.50 ± 0.09 (11)	5.54 ± 0.93	67.62 ± 11.39	0.49 ± 0.09	0.50 ± 0.09	0.02 ± 0.08

Fig. 3.

Developmental acceleration in synaptic current decay in twi^+/− fish results from a loss of the slow component of decay. Representative normalized mEPCs from 48-hpf (A), 72-hpf (B), and 120-hpf (C) twi^+/− fish fit by the sum of 2 (left trace) and 3 (right trace) exponential components. The mEPC decay is shown in black, and the sum of the fitted components is shown in red. The mean time constants and fractional amplitude contribution for the fast (red), intermediate (blue), and slow (green) components are shown for individual recordings at 48 (D), 72 (E), and 120 hpf (F). The overall means ± SD for each distribution are shown in black. (G) Graph of overall mean ± SD values of fractional amplitudes showing significant reductions in the slow component (green) at 72 hpf (*P = 0.02; paired two-tailed t test) and at 120 hpf (**P < 0.0001). (H) The time constants for τ_f (red), τ_i (blue), and τ_s (green) at 48, 72, and 120 hpf.

To establish the causality of altered synaptic current time course on membrane depolarization, we performed paired motor neuron–muscle recording in twi^+/− fish (15, 17). This approach revealed the naturally occurring evoked synaptic depolarization driving twi^+/− muscle. Muscle cells were held in current clamp at an initial resting potential of −90 mV, and a single muscle AP was triggered by stimulating the motor neuron at a frequency of 1 Hz. Between 12 and 20 muscle APs were averaged to generate an average waveform for each condition (Fig. 4A). APs from 48-hpf twi^+/− muscle, representing the height of the motility defect, were compared with 120-hpf twi^+/− muscle, which was taken as the recovered time point. These recordings revealed that repolarization proceeded along a more rapid time course during the later developmental period (Fig. 4A). To assess the differences, the average time required for the membrane potential to return to resting level was calculated. At 48 hpf, 11.5 ± 0.4 ms was required for wild-type fish to return to the 20% level and 17.3 ± 0.5 ms for them to return to the 10% level (Fig. 4B; n = 5 pairs). In 120-hpf wild-type animals, the respective values were 7.0 ± 0.5 ms and 11.8 ± 0.7 ms. By contrast, repolarization occurred slowly in twi^+/− fish. At 48 hpf, return to 20% level required 80.3 ± 19.9 ms and return to 10% level required 137 ± 41.4 ms (Fig. 4B; n = 6 pairs). At 120 hpf, these values corresponded to 41.9 ± 5.0 ms and 66.4 ± 6.1 ms (Fig. 4B; n = 3 pairs). The accelerated repolarization at 120 hpf reflects the loss of the τ_s component of synaptic current.

Fig. 4.

The effects of accelerated synaptic current decay on membrane repolarization. (A) Muscle cell voltage responses to the single firing of a motor neuron. Between 12 and 20 individual muscle responses were averaged to generate the average waveforms corresponding to 48-hpf wild-type (WT; gray), 48-hpf twi^+/− (black), and 120-hpf twi^+/− (red) fish. The scale bar represents 10 mV and 10 ms. (B) Comparison of the repolarization time course among 48-hpf wild-type (gray), 48-hpf twi^+/− (black), and 120-hpf (red) twi^+/− animals. (C) Sample traces for the first three sequential muscle responses to motor neuron firing at 20 Hz. Repolarization was incomplete for both 48-hpf (black traces) and 120-hpf (red traces) twi^+/− fish. However, the take-off level for responses 2 and 3 was much more positive at 48 hpf.

Swimming requires muscle to follow high-frequency trains of APs that are generated by spinal motor neurons. As a test of the muscle’s ability to fire repetitively, the motor neuron was stimulated at 20 Hz, a frequency observed for spontaneous slow swims, whereas the muscle cell was held in current clamp at an initial holding potential of −90 mV (Fig. 4C). At 48 hpf, repolarization was incomplete and slowed with each successive stimulation (Fig. 4C). By 120 hpf, the current decay accelerated and repolarization was markedly faster, resulting in a more rapid repolarization and return to baseline (Fig. 4C).

The accelerated repolarization of the muscle was associated with the selective loss of the slow component of synaptic current decay (τ_s), supporting a specific role of the α_twiβδγ receptors in conferring the motility defect. Muscle recordings at the recovered time point (120 hpf) lacked this slow component, pointing to the developmental γ−ε switch as potentially causal to behavioral recovery. To determine the expression levels of ε and γ, measurements of the respective cDNAs were made to examine the abundance of each subunit transcript in the RNA extracts of whole wild-type embryos between the ages of 24 and 192 hpf (Fig. 5). The transcript levels were indicated as the relative abundance (in copy number) of the target transcript compared with the copy number of the control transcript elongation factor 1 (elf1a). Elf1a represents a validated reference gene for qPCR analysis of zebrafish tissue (16). Resultant normalized transcript levels indicated that the levels of ε transcript increased dramatically between 24 and 48 hpf, during which the γ transcript levels were constant (Fig. 5). This corresponded to ε/γ ratios of 1.6 at 24 hpf, 4.5 at 48 hpf, and 4.0 at 72 hpf. The nearly threefold increase in relative abundance of ε transcripts during the 24–48 hpf developmental period preceded the onset of behavioral and synaptic improvement. This might be expected as the degradation time constant measured for αβδγ receptors in newly developing synapses of both amphibians (18) and mammals (19) is ∼1–2 d. Taking into account the expected time course of receptor turnover, there is reasonable agreement with the observed timing of transcript change and the functional alterations at the synapse. At later time points, γ transcript levels continued to decline, increasing the ε/γ ratio to 8.3 at 144 hpf (Fig. 5). By 192 hpf, both γ and ε transcript levels decreased, but the ε/γ ratio remained elevated at 10.3. To further validate the findings of this qPCR analysis, a second set of primer-probes was used for both ε and γ transcripts. Results from the second primer-probe set followed a similar pattern at each time point, along with corresponding ε/γ ratios.

Fig. 5.

Developmental pattern of zebrafish γ- and ε-subunit expression. qPCR measurements from cDNA pools of ∼20 zebrafish embryos. Target transcript levels (copy numbers) have been normalized to endogenous reference, elf1a, a transcript expressed uniformly throughout development. Gray boxes denote γ-subunit expression and black boxes denote ε-subunit expression. Samples were run in triplicate, and mean and SD calculations are shown for each indicated time point.

The idea that decreases in the γ-subunit levels were causal to behavioral recovery was tested directly by reducing the early expression of the γ subunit using an antisense morpholino. This would favor specific assembly of receptors containing the ε subunit. Four nanograms of γ-subunit morpholino was injected into the fertilized twister eggs at the one- or two-cell stage, and the effects on both motility and synaptic currents were tested at 48 hpf. Visual inspection revealed that the miniature end-plate current (mEPC) decay was much faster in morpholino-injected fish than in noninjected twi^+/− fish (Fig. 6A). mEPC decays were fit with three exponential components, with average time constants of 0.52 ± 0.13 ms for τ_f, 7.93 ± 0.92 ms for τ_I, and 58.49 ± 15.33 ms for τ_s (n = 8 cells). The fractional contribution of τ_f corresponded to 0.56 ± 0.13, τ_i corresponded to 0.41 ± 0.15, and τ_s corresponded to 0.04 ± 0.03 (Fig. 6B). This represented a significant reduction in the fractional contribution of τ_s between the injected and noninjected 48-hpf twi^+/− fish (P = 0.02; paired t test), similar to the reduction observed in 120-hpf animals. Additionally, morpholino-injected twi^+/− fish were able to mount a weak rhythmic swimming.

Fig. 6.

Injection of a γ-subunit antisense morpholino in skeletal muscle reduces the slow component of synaptic current decay. (A) Representative normalized mEPCs recorded at 48 hpf twi^+/− from noninjected (gray trace) and γ morpholino–injected (black trace) animals. (B) The overall fractional contributions by τ_f (red), τ_i (blue), and τ_s (green) of the exponential fit for noninjected and morpholino-injected twi^+/− fish. The fractional contribution of τ_s was significantly reduced by injection of the γ morpholino (*P = 0.02 paired t test).

Discussion

Our results point to the γ−ε subunit switch as the principal regulator of behavioral improvement in our zebrafish model for SCS. The accelerated repolarization of the muscle following increased ε expression resulted in muscle APs that were able to follow higher-frequency stimulation. To date, there have been no human SCS mutations shown to undergo recovery as a result of the γ−ε subunit switch. Escobar syndrome represents the only myasthenic syndrome that recovers as result of the switch, but this is a direct consequence of mutations exclusively in the γ subunit (7, 8). The mutations in γ are thought to inhibit receptor expression rather than alter receptor kinetics, as seen in SCS. Indeed, the three Escobar-associated mutations in γ that have been tested by means of heterologous expression in human embryonic kidney cells indicated greatly reduced expression of ACh receptors (9). No electrophysiological analysis has been performed on any of these γ-subunit mutations, so it remains formally possible that gain-of-function mutations exist for this subunit as well. The existence of so many mutations of other ACh receptor subunits resulting in gain of function supports the notion that γ-subunit counterparts may have gone undetected. As with a total loss of receptors, a γ gain-of-function mutation affecting receptor kinetics has the potential to cause severe secondary damage, as reflected in SCS. That being the case, early in utero treatment with quinidine, the drug used to treat SCS, also may be effective in treating Escobar syndrome (18–21).

Most importantly, our findings from twister zebrafish point to the possibility that mutations in subunits other than γ might result in undisclosed forms of SCS, which undergo improvement during development. In our case, mutations in the α subunit result in differential functional interactions with the γ versus the ε subunit. Although there is no precedent for human forms of SCS benefiting from the γ–ε switch, our findings from zebrafish raise the likelihood that cases may have gone undetected as a result of recovery. There is much precedence of zebrafish serving as a model for different human forms of myasthenic syndrome. In the case of rapsyn deficiency, zebrafish pointed the way to the original identification of the mutated target protein (22–24). At this point, no predictions can be made about specific mutations that would benefit from the γ−ε switch in humans. From our studies, the α subunit is a prime candidate, but no reported cases of recovery for SCS patients exist. Although we observed a full behavioral recovery in twister mutant fish, there may be long-term secondary consequences in human forms, once again calling for therapeutic intervention. Finally, if twister equivalents exist in the human population, sequencing of α, β, and δ receptor subunits is needed to identify the culprit mutation.