Muscle dysfunction in a zebrafish model of Duchenne muscular dystrophy

Abstract

Sapje zebrafish lack the protein dystrophin and are the smallest vertebrate model of Duchenne muscular dystrophy (DMD). Their small size makes them ideal for large-scale drug discovery screens. However, the extent that sapje mimic the muscle dysfunction of higher vertebrate models of DMD is unclear. We used an optical birefringence assay to differentiate affected dystrophic sapje larvae from their unaffected siblings and then studied trunk muscle contractility at 4–7 days postfertilization. Preparation cross-sectional area (CSA) was similar for affected and unaffected larvae, yet tetanic forces of affected preparations were only 30–60% of normal. ANCOVA indicated that the linear relationship observed between tetanic force and CSA for unaffected preparations was absent in the affected population. Consequently, the average force/CSA of affected larvae was depressed 30–70%. Disproportionate reductions in twitch vs. tetanic force, and a slowing of twitch tension development and relaxation, indicated that the myofibrillar disorganization evident in the birefringence assay could not explain the entire force loss. Single eccentric contractions, in which activated preparations were lengthened 5–10%, resulted in tetanic force deficits in both groups of larvae. However, deficits of affected preparations were three- to fivefold greater at all strains and ages, even after accounting for any recovery. Based on these functional assessments, we conclude that the sapje mutant zebrafish is a phenotypically severe model of DMD. The severe contractile deficits of sapje larvae represent novel physiological endpoints for therapeutic drug screening.

duchenne muscular dystrophy (DMD) is the most common muscular dystrophy, occurring once in every 5,000 male births (36). DMD is a progressive muscle dysfunction disease characterized by delayed motor development in infancy, progressing to loss of ambulation by the end of the first decade of life, and culminating in respiratory or cardiac failure in young adulthood. Striated muscles of DMD patients lack the rod-like protein dystrophin as a result of inherited or spontaneous mutations in the X-linked dystrophin gene (25). In skeletal muscles, dystrophin localizes to the inner surface of the sarcolemma where its NH₂ terminus binds to actin and its COOH terminus to the transmembrane protein β-dystroglycan. Dystrophin is critical for stabilizing the cell membrane against the mechanical forces produced during contraction (40).

There is currently no cure for DMD, but much progress has been made in the development of gene delivery, gene editing, exon skipping, and stem cell-based approaches for replacement of full-length or truncated, but functional, dystrophin protein constructs (18). Nevertheless, these promising therapies confront a number of hurdles related to tissue delivery, low efficiency, or an inability to target all DMD mutations. A therapeutic alternative to dystrophin replacement is to target signaling pathways disrupted in DMD using already approved drugs, such as statins and phosphodiesterase inhibitors (39, 50).

One approach for discovering novel DMD therapeutics is to conduct large-scale screens of approved drugs. These screens can be cell based (38) or conducted on small model organisms, such as Drosophila melanogaster, Caenorhabditis elegans, or Danio rerio. Of these model organisms, only zebrafish share the basic morphology, physiology, and genomics common to all vertebrate species. For instance, > 70% of human genes, and > 80% of human disease-related genes, have an orthologous gene in the zebrafish (27). In terms of DMD, the zebrafish dystrophin ortholog encodes a protein that has a high degree of NH₂ and COOH terminus homology with human dystrophin (21). These features, combined with their small size, rapid ex utero development, body transparency, permeability to drugs, and relative ease of genetic manipulation, make zebrafish larvae attractive models for DMD drug discovery (15). In fact, dystrophic zebrafish strains, such as sapje (4, 17) and sapje-like (21), have been used to screen > 1,000 Food and Drug Administration-approved drugs and bioactive compounds, resulting in the identification of several novel DMD therapeutic candidates (28, 49).

In higher vertebrates, dystrophin deficiency has variable effects on muscle function. Canine models are characterized by severe muscle weakness and a greatly heightened susceptibility of muscles to injury by high-strain eccentric contractile activity (9, 30, 48, 53). The mdx mouse presents a relatively mild muscle weakness coupled with a moderate injury response (10, 34, 37, 40, 52). Where dystrophin-deficient zebrafish fall along this spectrum is not well established, making it difficult to put the results of studies using sapje and sapje-like into the proper physiological perspective. In particular, the sensitivity of sapje to mechanical strain, a hallmark of dystrophin deficiency, has not been established. Furthermore, as most drug screens occur as sapje larvae are undergoing rapid development (29), and because twitch force of wild-type zebrafish larvae has been reported to vary during this time period (44), the stability of contractile measurements conducted on sapje larvae remains an open question. Therefore, the goal of this project was to rigorously evaluate the validity of using 4–7 days postfertilization (dpf) sapje larvae as a model for the muscle dysfunction that characterizes DMD.

METHODS

Zebrafish.

Zebrafish (D. rerio) were housed, bred, and used under the guidelines and approval of the Institutional Animal Care and Use Committee at Boston Children's Hospital. Pairs of heterozygous sapje (dmd^ta222a) fish were mated, and fertilized eggs were collected, cleaned, and cultured at 28.5°C. The resulting larvae were incubated at 28.5°C through 7 dpf. Unused 7 dpf larvae were euthanized with tricaine.

Birefringence assay.

The dystrophic phenotype was evaluated using a birefringence assay at 4 dpf as previously described (28, 45). Briefly, larvae were anesthetized with 0.02% tricaine, carefully aligned between two glass-polarizing filters, and viewed with a stereomicroscope as one filter was rotated to maximize birefringence. Larvae showing consistently bright, well-organized myotomes were classified as unaffected, and those displaying patchy areas of disrupted and disorganized somites were classified as affected. Figure 1A shows that unaffected and affected larvae, which appear indistinguishable under bright-field illumination, can easily be differentiated when viewed under polarized light. Unaffected and affected larvae were placed in separate dishes containing fresh fish water and returned to incubation until use.

Fig. 1.

Details of the zebrafish preparation. A: micrographs of 4 days postfertilization (dpf) sapje larvae illuminated with bright-field or polarized light. B: schematic diagram of the experimental apparatus. p., preparation; t., force transducer; m., position motor; b., chamber containing fish bicarbonate buffer; i.m., inverted microscope. For clarity, the electrodes that flanked the preparation are not shown. The diagram is not drawn to scale. C: arrowhead indicates the gastrointestinal opening, the anatomical location where the preparation was attached to the force transducer connector. The attachment point to the motor connector was ≈1 mm distal to the gastrointestinal opening.

Physiological preparation.

An individual larvae was selected at random from the unaffected or affected pools for study. Because larvae develop very rapidly, experiments on a given day were conducted by alternately selecting unaffected and affected fish for study. The selected fish was anesthetized in 0.02% tricaine and decapitated.

The experimental apparatus for assaying muscle function is shown schematically in Fig. 1B. The larval body was transferred to a small experimental chamber (≈4 ml total volume) containing fish bicarbonate buffer of the following composition (in mmol): 117.2 NaCl, 4.7 KCl, 1.2 MgCl₂, 1.2 KH₂PO₄, 2.5 CaCl₂, 25.2 NaHCO₃, 11.1 glucose (13). The buffer was maintained at 25°C and equilibrated with a mixture of 95% O₂, 5% CO₂ throughout the experiment. The proximal end of the fish body was attached to a small length of titanium wire extending from the output tube of an isometric force transducer (Aurora Scientific, model 403A) and the distal end to a 3 mm segment of titanium wire that had been epoxied to the lever arm of a high-speed motor (Aurora Scientific, model 308B). The experimental chamber was attached to the stage of an inverted microscope (Olympus IX-70). Platinum electrodes, connected to a biphasic muscle stimulator (Aurora Scientific, model 701A) flanked the preparation. A personal computer, multipurpose data acquisition board (National Instruments, model PCI-6229), and a custom program written in LabVIEW (National Instruments, version 2010), were used to control preparation stimulation and motor lever arm position while recording length and force data to disk.

Preparation attachment.

We took particular care that preparations were attached to instrumentation at a consistent anatomical location so that comparisons between different preparations would be valid. Simple overhand loops of 10-0 monofilament suture were used to securely hold the preparation to the titanium wires. For the proximal (transducer) connection, the loop was positioned at the level of the gastrointestinal opening (Fig. 1C). This represented the widest part of the preparation while providing a recognizable anatomical landmark to ensure consistent attachment between preparations. Reproducibility was facilitated by the notch between the anal and caudal fins, which helped guide the suture to the correct anatomical position for fastening. A similar suture loop was used to attach the distal portion of the preparation to the wire extending from the motor. The distance between the proximal and distal sutures was ∼1 mm. Care was taken to avoid attaching the preparation too close to the tip of the tail as these preparations were found to be too compliant.

Preliminary set-up of the preparation.

Images of the attached preparation were captured using a charge-coupled device (CCD) camera. Sarcomere length (SL) was determined from these images by two-dimensional Fourier spectral analysis (43). Sarcomere length was initially set to 2.05–2.10 μm. The preparation was then stimulated to contract with 200 μs biphasic square-wave pulses. In preliminary experiments, we found that force was slightly greater at 300 Hz stimulation than at 200 Hz. Thus, 300 Hz was used in all tetanic contractions. Supramaximal current and optimal preparation length for tension (PL₀) were empirically established using brief (30 ms) tetani separated by rest periods of at least 60 s. After PL₀ was established, SL were reassessed and recorded. All subsequent contractile measurements were initiated at PL₀.

Force measurements.

A twitch was elicited by a single supramaximal pulse. Two or three twitches, separated by 60 s, were recorded, and the trial with the greatest twitch force was used in statistical analysis. Tetanic contractions were elicited by nine supramaximal pulses at 300 Hz. The greatest tetanic tension attained in the experiment was taken as peak tetanic force. All force values reported represent active force, defined as the peak force attained during stimulation minus the resting force immediately prior to stimulation.

Lengthening contraction protocol.

A single lengthening, or eccentric, contraction protocol was developed to assess the susceptibility of larval preparations to high-strain contractile activity. Preliminary experiments were conducted to identify strain magnitudes that induced a force deficit but did not cause preparation breakage. The final protocol consisted of a preisometric trial, a single eccentric contraction, and a postisometric trial, all separated by 60 s rest. For some preparations, postisometric contractions were repeated once every 60 s over a 10 min period to assess long-term recovery. During the eccentric trial, the preparation was stimulated isometrically (200 μs pulses at 300 Hz) for 20 ms and then lengthened by a distance equivalent to either 5 or 10% of PL₀ at a velocity of 1.0 PL₀/s. Stimulation ceased at the immediate conclusion of the stretch. The unstimulated preparation was held at its extended length for 400 ms and then returned to its original PL₀ at −1.0 PL₀/s. The preisometric and all postisometric trials were identical to the respective eccentric trial except that the length of the preparation was not altered.

Measurement of preparation cross-sectional area.

Immediately after the conclusion of the functional measurements, the experimental apparatus, with the preparation still attached at PL₀, was moved to a stereomicroscope (×40). There, the sutures were loosened slightly so that the preparation could be carefully rotated about the titanium wire connectors. Each titanium wire had a small 90° bend at the tip that served as a stop to prevent the suture and preparation from sliding off the wire. A CCD camera was used to capture images of the preparations maximum width. The preparation was rotated 90°C, and images captured of its depth at the same anatomical position. All images included an internal length calibration so that pixels could be converted into μm using ImageJ (http://imagej.nih.gov/ij/). The maximum cross-sectional area (CSA) of the preparation was calculated by modeling the cross section of the zebrafish preparation as an ellipse.

Statistical analysis.

Our primary hypothesis was that force would be less, and susceptibility to injury by the lengthening protocol greater, in the affected preparations. Our secondary hypothesis was that these differences would be stable with larval age, i.e., across 4–7 dpf. Dependent variables were analyzed by a two-way analysis of variance, with main effects of phenotype and age. Our primary interest was in a phenotype main effect and/or a phenotype by age interaction. To evaluate significant interactions, independent two-tailed t-tests were used to identify differences within a phenotype across the four ages studied and between affected and unaffected phenotypes at each age point. The family-wise type I error rate for these comparisons was controlled at P < 0.05 using a Bonferroni adjustment (a critical P < 0.0031 for each comparison). Linear and nonlinear regression were used to model the relationship between force and preparation CSA and to describe the recovery of force following the lengthening contraction protocol. Statistical calculations and curve fitting were conducted using the aov, multi, and nlm packages of R (42). All values are reported as means ± SE.

RESULTS

Zebrafish larvae were classified as unaffected (normal muscle structure) or affected (abnormal structure) using a birefringence assay (Fig. 1A). Larval tail sections were attached to wire connectors extending from an isometric force transducer and a position motor (Fig. 1B), with the proximal end of the larvae fastened at a consistent anatomical location (Fig. 1C). Images of a representative larvae prepared in this manner are shown in Fig. 2A.

Fig. 2.

Cross-sectional area (CSA) and optimal sarcomere length. A: micrograph showing a larval preparation attached to titanium wires with 10-0 monofilament suture loops. In the top micrograph, the preparation was aligned to measure its maximal width (distance between the 2 arrowheads). In the bottom micrograph, the preparation was rotated 90° to measure its depth (the distance between the black and white arrowheads). Preparation CSA was assumed to be elliptical. Calibration bars = 500 μm. B: mean preparation CSA. CSA changed with age, but this response did not differ between phenotypes (no phenotype effect; age effect, P = 0.0470; no interaction). Number of preparations: 6, 19, 23, and 28 unaffected and 4, 15, 26, and 33 affected at 4, 5, 6, and 7 dpf, respectively. C: high-magnification micrograph of an unaffected preparation. Individual myofibers extend laterally between adjacent myotomes (indicated by arrowheads). Sarcomere length was obtained from images similar to this as detailed in methods. D: mean sarcomere length (SL). Overall, SL was significantly greater in unaffected larvae (phenotype effect, P = 0.0001; no age effect; no interaction). Number of preparations: 6, 19, 23, and 28 unaffected and 2, 13, 26, and 32 affected at 4, 5, 6, and 7 dpf, respectively. ●, Unaffected preparations; ○, affected preparations. Values represent means ± SE. In many instances the error bars are obscured by the mean symbol. dpf, Days postfertilization.

Preparation CSA.

Preparation CSA was modeled as an ellipse using images obtained of the preparation's maximal width (Fig. 2A, top) and, after rotating the preparation 90° (Fig. 2A, bottom), its depth. No difference in CSA was observed between unaffected and affected preparations (Fig. 2B). CSA declined slightly with age, and although this appeared to be most pronounced for the fish with abnormal birefringence, there was no statistical difference in this response for affected and unaffected populations.

Optimal sarcomere and preparation length.

The length of the preparation was adjusted to maximize force during a brief tetanus. Once this optimal preparation length (PL₀) had been determined, optimal SL was obtained from high-magnification images of the preparation (Fig. 2C). The optimal SL for tetanic force was 2.11 ± 0.01 μm (Fig. 2D) for unaffected larvae. This SL falls squarely on the plateau region of the SL-tension relationship reported for carp, another member of the teleost family (46). Our data are also in good agreement with the SL of 2.15 μm reported to optimize twitch tension of 5–7 dpf Tübingen zebrafish (13), especially considering that twitch tension is optimized at a slightly longer SL than tetanic tension (3).

Optimal SL was significantly shorter for affected vs. unaffected preparations, although the absolute difference between the two phenotypes was relatively small in magnitude (Fig. 2D). Optimal SL was stable over days 4–7 for both phenotypes. PL₀ was similar for unaffected and affected fish, with mean values of 922 ± 43, 1,037 ± 27, 1,061 ± 21, and 1,026 ± 15 μm at 4, 5, 6, and 7 dpf, respectively. Note that these PL₀ values simply describe the preparation under study. They do not represent the actual larvae tail length as the position of the distal attachment site was at the discretion of the investigator.

Peak force.

When stimulated with a single stimulus or a stimulus train, unaffected larval preparations responded with a robust force transient (Fig. 3A). In contrast, twitch and tetanic force were considerably depressed in affected preparations. This relationship held for the entire dataset across all developmental ages, where both twitch (Fig. 3B) and tetanic force (Fig. 3C) were significantly less for the affected preparations.

Fig. 3.

Twitch and tetanic force. A: representative twitch (left) and tetanic (right) force records from a 6 dpf unaffected and a 6 dpf affected zebrafish preparation. B: twitch force. Mean twitch force was significantly less for affected preparations, with both groups showing similar responses across age (phenotype effect, P < 0.0001; age effect P < 0.0001; no interaction). Number of observations: 6, 19, 23, and 28 unaffected and 4, 13, 25, and 32 affected preparations at 4, 5, 6, and 7 dpf, respectively. C: tetanic force. Tetanic force responded differently for unaffected and affected preparations as larvae matured, but independent of age, tetanic force was always significantly less for affected preparations (phenotype effect, P < 0.0001; age effect, P = 0.0002; interaction P = 0.0185). Post hoc analysis revealed that force of affected preparations was less on the initial day of study (4 dpf) compared with the last 2 days of study (6 and 7 dpf). In contrast, the force of unaffected preparation did not vary across age. At each time point, force was always greater for the unaffected larvae, consistent with the main effect of genotype. Number of observations: 6, 19, 23, and 28 unaffected and 4, 15, 26, and 33 affected preparations at 4, 5, 6, and 7 dpf, respectively. D: twitch force normalized to CSA. Mean twitch force/CSA was significantly less for affected preparations, with both groups showing similar changes with age (phenotype effect, P < 0.0001, age effect, P < 0.0001, no interaction). Number of observations same as in B. E: tetanic force normalized to CSA. Tetanic force/CSA responded differently for unaffected and affected preparations as larvae matured, but on average, tetanic force/CSA was always significantly less for the affected preparations (phenotype effect, P < 0.0001; age effect, P = 0.0002; interaction, P = 0.0071). Post hoc analysis revealed that affected preparations produced significantly less force than age-matched unaffected preparations and that force of affected preparations (but not unaffected preparations) rose as larvae matured (affected 4 dpf significantly different from affected 6 and 7 dpf and affected 5 dpf significantly different from affected 7 dpf). Number of observations same as in C. F: twitch force-to-tetanic force ratio. The mean ratio of twitch to tetanic force was significantly less for affected preparations, with both groups showing similar changes with age (phenotype effect, P < 0.0001; age effect, P = 0.0009; no interaction). ●, Unaffected preparations; ○, affected preparations. Values represent means ± SE. In many instances the error bars are obscured by the mean symbol. dpf, Days postfertilization; CSA, preparation cross-sectional area.

These impairments in force could not be explained by differences in the CSA of the unaffected and affected preparations because an average 50% deficit in twitch force (Fig. 3D) and an average 40% deficit in tetanic force (Fig. 3E) remained after normalization of force to preparation CSA. The twitch-to-tetanus force ratio for affected preparations was significantly less than for unaffected larvae (Fig. 3F), indicating that twitch force was impaired to a greater relative extent than tetanic force.

There was a significant developmental effect on tetanic force that tended to reduce differences between unaffected and affected groups as the larvae matured. As tetanic force of the unaffected preparations was stable across the study, this narrowing of the differences between unaffected and affected preparations was the result of increases in force production of the affected larvae, especially between days 4 and 5. Nevertheless, even at day 7, affected preparations still produced 30% less tetanic force per CSA than unaffected preparations.

To examine force differences between unaffected and affected larvae more closely, we assessed the relationships between tetanic force, preparation CSA, and phenotype by an analysis of covariance (Fig. 4). As detailed in Table 1, the slopes of the relationship between CSA and absolute force differed significantly between phenotypes. As expected, tetanic force rose as the CSA of the unaffected preparations increased. In contrast, the force produced by affected larvae was unrelated to their CSA. This indicates that the normal relationship between force and preparation CSA was absent, or uncoupled, in affected larvae.

Fig. 4.

Relationship between cross-sectional area (CSA) and tetanic force. The relationship between preparation CSA and tetanic force for each phenotype. Regression coefficients are presented in Table 1. Number of observations same as in Fig. 3E. ●, Unaffected preparations; ○, affected preparations.

Table 1.

Relationship between tetanic force and preparation cross-sectional area

Group	y-Intercept, mN	Slope, mN/mm2
Unaffected	0.777 ± 0.018	28.4 ± 6.7
Affected	0.952 ± 0.012	−3.3 ± 4.7

Data are means ± SE. Number of preparations same as reported in Fig. 3. Analysis of covariance indicated that slopes were significantly different between phenotypes (P = 0.001), but that y-intercepts were not significantly different (P = 0.42). Furthermore, the slope of the unaffected preparations was significantly different from zero (P < 0.0001), while the slope of the affected larvae was not significantly different from zero (P = 0.49). The model's coefficient of determination was 0.69.

Kinetics of contraction.

Representative twitches from an unaffected and an affected preparation are superimposed in Fig. 5A to highlight differences in the kinetics of force development and relaxation. These differences are more evident when force responses are scaled relative to peak twitch force and include a slight right-ward shift in the twitch force record and a noticeable slowing of force relaxation. In fact, the mean time from the onset of stimulation to peak twitch force, or contraction time, showed a slight slowing in the affected preparations across all developmental stages (Fig. 5B). The time required for twitch force to decline to half-maximal force, or half-relaxation time, was significantly slower in 4 dpf affected preparations, but not at later time points (Fig. 5D).

Fig. 5.

Twitch kinetics. A: superimposed twitch force records of a 4 dpf affected and a 4 dpf unaffected preparation. At the left, the records have been superimposed in the same absolute force scale. At the right, superimposed records have been normalized to their peak twitch force. B: contraction time (CT). Mean CT was significantly slower for affected preparations (phenotype effect, P = 0.0047; no age effect; no interaction). Number of observations: 6, 19, 23, and 28 unaffected and 4, 13, 25, and 32 affected preparations at 4, 5, 6, and 7 dpf, respectively. C: maximum rate of twitch tension development (+dP/dt). Mean +dP/dt was significantly slower for affected preparations, with both groups showing similar changes with age (phenotype effect, P < 0.0001; age effect, P < 0.0001; no interaction). Number of observations same as in B. D: half-relaxation (hRT) time. Affected and unaffected preparations showed different age-related responses in hRT (no phenotype effect; age effect, P < 0.0001; interaction, P < 0.0001). Post hoc analysis revealed that hRT of 4 dpf affected preparations was significantly slower compared with all other affected time points and the 4 dpf unaffected value. Number of observations same as in B. E: maximum rate of twitch tension relaxation (−dP/dt). −dP/dt responded differently for unaffected and affected preparations as larvae matured, but on average, −dP/dt was always significantly slower for the affected preparations (phenotype effect, P < 0.0001; age effect, P = 0.0094; interaction, P = 0.0058). Post hoc analysis revealed that −dP/dt of affected (but not unaffected preparations) preparations became faster as larvae matured (affected 4 dpf significantly different from all other affected time points). However, affected preparations always relaxed significantly slower than age-matched unaffected preparations. Number of observations same as in B. ●, Unaffected preparations; ○, affected preparations. Values represent means ± SE. In many instances the error bars are obscured by the mean symbol. dpf, Days postfertilization.

Differences in peak twitch force between affected and unaffected groups could confound interpretation of contraction time and half-relaxation times. To address this, we calculated the first derivative of force (per CSA) as a function of time. To arrive at the maximal rate of twitch tension development, or +dP/dt, differentiation was conducted over the portion of the twitch force record as tension was rising. To obtain the maximal rate of twitch tension decline, or −dP/dt, a similar procedure was carried out beginning at the peak twitch force and continuing until force was within a few percent of baseline. Mean +dP/dt and −dP/dt were both significantly slower in affected preparations at all developmental ages (Fig. 5, C and E). Taken together, these data indicate a substantial slowing of both tension development and relaxation in affected preparations.

Lengthening contractions.

To test the susceptibility of sapje larvae to high mechanical strain, preparations were activated and subjected to a single eccentric contraction. The protocol, which is illustrated in Fig. 6A, consisted of a preisometric contraction, a single eccentric contraction, and a postisometric contraction. The difference between post- and preisometric force was used as an index of the susceptibility of the preparation to injury. Because preisometric force varied considerably between the two larval populations, this force difference was expressed as a percent change, F_%Δ, as follows:

$$F_{\%\Delta }=\left[\left(F_{\text{post}}-F_{\text{pre}}\right)/F_{\text{pre}}\right]\times 100\%$$ (1)

Fig. 6.

Force deficits after a single eccentric contraction. A: representative records illustrating force of an unaffected preparation and an affected preparation obtained 60 s before (“pre”), during (“ecc”), and 60 s after (“post”) a single eccentric contraction. B: the change in force 60 s after the 5 and 10% strain protocols. Note that force is expressed relative to the force attained during the pre-eccentric trial. For the 5% protocol, the force deficit responded differently for unaffected and affected preparations as larvae matured, but on average, the deficit was always significantly greater for the affected preparations (phenotype effect, P < 0.0001; age effect, P = 0.0014; interaction, P = 0.0087). Post hoc analysis revealed that the affected force deficit was less at 7 dpf than at 5 and 6 dpf. Nevertheless, the force deficit of the affected preparations was always greater than that of the age-matched unaffected preparation. Number of observations: 4, 11, 13, and 11 unaffected and 4, 5, 15, and 13 affected preparations at 4, 5, 6, and 7 dpf, respectively. For the 10% protocol, the mean relative force deficit was significantly greater for affected preparations (phenotype effect, P < 0.0001; age effect, P = 0.0005; no interaction). Number of observations: 6, 7, and 10 unaffected and 6, 9, and 12 affected preparations at 5, 6, and 7 dpf, respectively. ●, Unaffected preparations; ○, affected preparations. Values represent means ± SE. In many instances the error bars are obscured by the mean symbol. dpf, Days postfertilization.

Force deficits were significantly greater in affected larvae, compared with unaffected preparations, 1 min after a lengthening contraction of 5% strain (Fig. 6B). Increasing strain magnitude to 10% resulted in greater force deficits, especially for the affected preparations (Fig. 6B). Thus, affected 4–6 dpf preparations lost 50% or more of their initial force after 5% strain and 80–95% after 10% strain. In contrast, force deficits of unaffected larvae at similar time points and strains never exceeded 20% and were often much less. Both unaffected and affected larvae appeared to became somewhat more resistant to strain as they matured. However, even at 7 dpf, 5 and 10% strain was associated with force deficits of ≈30 and 50%, respectively, which was three- to fivefold greater than observed for the unaffected preparations. In summary, affected preparations were more susceptible to immediate force losses after high strain contractile activity at all time points studied.

Long-term recovery from lengthening contractions.

The extreme force loss experienced by the affected preparations 60 s after a single eccentric contraction motivated us to investigate whether these force deficits were a short- or long-term phenomena. Figure 7 shows F_%Δ values attained every 60 s across the 10 min that followed an eccentric contraction of 5 or 10% strain.

Fig. 7.

Long-term recovery of force following the eccentric contraction protocol. Top row: recovery of force following the 5% stain protocol for 5, 6, and 7 dpf unaffected and affected preparations. Bottom row: recovery of force following the 10% stain protocol at similar time points. Note that all forces expressed as a percentage of the pre-eccentric force. Unaffected points were fit with Eq. 2 and affected points with Eq. 3. Regression parameters are compiled in Table 2 and predicted forces after full recovery are presented in Table 3. Number of observations at 5% strain: 3, 6, and 6 unaffected and 4, 6, and 5 affected preparations at 5, 6, and 7 dpf, respectively. Number of observations at 10% strain: 2, 5, and 5 unaffected and 2, 6, and 6 affected preparations at 5, 6, and 7 dpf, respectively. ●, Unaffected preparations; ○, affected preparations. Values represent means ± SE. In many instances the error bars are obscured by the mean symbol. dpf, Days postfertilization.

For the unaffected preparations, the long-term recovery of average relative force, uF_%Δ, across time, t, in seconds, was fit by the following linear relationship,

$$u{F}_{\%\Delta }=\left(F_{\text{slope}}\right)\left(t\right)+F_0$$ (2)

where uF_slope represents the relative force loss per second and F₀ represents the relative isometric force at recovery time 0, i.e., the instantaneous force deficit at the immediate conclusion of the eccentric contraction. For the affected preparations, the average recovery of relative force of the affected preparations, aF_%Δ, was fit by the following exponential function

$$a{F}_{\%\Delta }=F_{\text{rec}}\left[1-e^{-\left(1/\tau \right)\cdot t}\right]+F_0$$ (3)

where the parameter F₀ represents the relative isometric force at time 0 of recovery, F_rec quantifies the force recovered from F₀ as a percent of preisometric force, t is time in seconds, and τ is the recovery time constant in seconds. The estimated model parameters and their standard errors are reported in Table 2, and the resultant fits are plotted as solid lines in Fig. 7.

Table 2.

Regression coefficients describing force recovery in the initial 10 min following a single eccentric contraction

		Unaffected		Affected
Strain, %	Age, dpf	F0, % preforce	Fslope, % preforce/s	F0, % preforce	Frec, % preforce	τ, s
5	5	−3.2 ± 0.4	−0.008 ± 0.001	−91.3 ± 2.4	70.7 ± 2.0	156 ± 10
	6	−1.9 ± 0.4	−0.005 ± 0.001	−87.8 ± 9.8	66.5 ± 9.6	63 ± 8
	7	−2.9 ± 0.5	−0.008 ± 0.001	−49.4 ± 3.6	38.8 ± 3.6	64 ± 5
10	5	−16.1 ± 1.3	0.003 ± 0.003	−94.3 ± 1.1	47.7 ± 4.0	509 ± 92
	6	−7.6 ± 0.4	−0.006 ± 0.001	−78.9 ± 2.5	32.8 ± 2.0	199 ± 35
	7	−6.9 ± 1.2	0.001 ± 0.003	−77.8 ± 1.3	55.6 ± 1.0	236 ± 15

Values are means ± SE. Number of preparations same as Figure 7. dpf, Days postfertilization. Recovery force of unaffected preparations was fit by Eq. 2, and recovery force of affected preparations was fit by Eq. 3. F₀ is force at recovery time 0; F_slope is the change in force per unit recovery time; F_rec specifies the magnitude of force recovered; τ is the time constant. Note that force is expressed as a percentage of the isometric force prior to the eccentric contraction (preforce).

The instantaneous force deficit was much greater for affected preparations, as F₀ ranged from −2 to −16% of preforce for unaffected preparations and from −50 to −94% of preforce for the affected preparations. The rate of recovery force at each strain, and across all developmental stages, changed only thousandths of a percent every second for unaffected preparations. In the majority of cases, the slopes were negative, which likely represents a generalized loss of preparation viability during the recovery period. In contrast, affected preparations showed substantial recovery of force following the eccentric contraction. The magnitude of the force recovered from F₀, or F_rec, was equivalent to 33–71% of preisometric force.

System theory predicts that after a period of three time constants, a function has attained 95% of its asymptotic value. We therefore adopted three time constants as our criteria for final recovery.

Table 3 reports the predicted relative force values of unaffected and affected preparations at 3τ using the model parameters reported in Table 2. At 5% strain, the model predicts that 5 and 6 dpf affected larvae will recover to within 75% of their preisometric force. In comparison, unaffected larvae are predicted to recovery to within 93–97% of preforce after similar strain. At 10% strain, affected preparations are predicted to recovery to ∼50% of their preisometric force compared with a predicted 90% recovery for unaffected larvae.

Table 3.

Predicted relative force after complete recovery from an eccentric contraction

Strain, %	Stage, dpf	3τ, s	Unaffected Recovery, % preforce	Affected Recovery, % preforce
5	5	468	93.1	75.9
	6	189	97.2	75.4
	7	192	95.6	88.5
10	5	1,527	88.5	51.0
	6	597	88.8	52.3
	7	708	93.8	75.0

Complete recovery was defined as relative force at 3 time constants (3τ). Relative force at 3τ was determined using the regression coefficients presented in Table 2.

Affected preparations seem to be more resistant to strain at 7 dpf, recovering to 89% of preisometric force after 5% strain and to 75% of preforce after 10% strain. However, at these same conditions, force of unaffected preparations recover to ∼95% of their original force. This model reveals that when recovery is taken into account, affected preparations at all time points remain considerably more susceptible to eccentric contractions than unaffected larvae, although by 7 dpf, unaffected preparations appear to be somewhat less susceptible than at earlier time points.

DISCUSSION

There are over 60 laboratory animal models of dystrophin deficiency, ranging from small invertebrates such as flies and nematodes, to mammals such as mice and canines (35). Zebrafish occupy a unique niche along this spectrum, sharing characteristic morphological, physiological, and genetic features with mammals while still possessing many of the attributes of invertebrate model organisms, such as small size, high reproductive capacity, rapid development, and accessible genomics. These features make dystrophin-deficient zebrafish ideal for large-scale DMD drug discovery screens (28, 49). Despite these advantages, our understanding of how dystrophin deficiency affects zebrafish physiology, and particularly as it relates to muscle function, is incomplete.

Force depression.

Dystrophin-null larvae rarely survive beyond 14 dpf so mutant lines were maintained by breeding heterozygotes, which have a normal phenotype and lifespan. As is standard practice in DMD drug screens (28, 49), we used a rapid, nonlethal optical birefringence assay (5, 45) to differentiate dystrophin-null larvae from their dystrophin-positive siblings.

Peak tetanic force produced by unaffected larvae was stable across the 4 days of study. As expected, we observed a linear relationship between tetanic force and preparation CSA in this population of larvae with their average tetanic force per unit preparation CSA averaging just under 60 kPa. In contrast, affected larvae showed severe reductions in absolute tetanic force, attaining only 30, 50, and 60% of the absolute tetanic force produced by their unaffected siblings at 4, 5, and 6–7 dpf, respectively. These deficits persisted even after force was normalized to the CSA of the preparation, suggesting a substantial reduction in the quality of the contractile tissue. In fact, the normal linear relationship between force and CSA was absent in the affected population, indicating an uncoupling of these two variables. These data all provide evidence that the quality of the trunk musculature was adversely impacted in the affected larvae.

It is important to consider these sapje force deficits in relation to other vertebrate models of DMD. Fast limb muscles of the mdx mouse often undergo hypertrophy so that they produce near normal absolute force and 70–80% of normal force per physiological CSA (1, 10, 12, 33, 34, 52). Force deficits are much more severe in utrophin-dystrophin double knockout mice (mdx:utrn dKO) and canine models of DMD where absolute force falls between 10 and 60% of normal and force/CSA attains ∼50% of wild-type values (11, 16, 30, 33, 48, 53). The force deficits observed for the affected sapje larvae in the present study are relatively greater than those generally reported for mdx limb muscles but within the range of those observed for the more severely affected mdx:utrn dKO mouse or canine models of DMD.

One factor underlying these force deficits is the sarcomeric lesions that can be observed when sapje larvae are illuminated by polarized light. These gaps and lesions are initiated by a stochastic process of myofiber detachment from the normally dystrophin-rich myosepta (4). This interface appears to be especially vulnerable as the absence of the transmembrane dystrophin-associated protein dystroglycan, or proteins that assist in anchoring the fiber to the extracellular matrix, such as laminin-α2 and collagen XXII, are also associated with myofibrillar detachment, impaired performance, and contractile defects in zebrafish larvae (19, 20, 22). However, if the sapje force deficit observed here was due entirely to physical disruption of sarcomeres one would expect twitch and tetanic force to be depressed in direct proportion. Instead, we observed that twitch force was depressed relatively more than tetanic force, as has been previously observed for canine DMD models (30). Because twitch force is more sensitive than tetanic force to reductions in intracellular Ca²⁺ transients, these results suggest that excitation-contraction coupling is impaired in affected larvae. If true, this would be consistent with reported reductions in the depolarization-induced Ca²⁺ transient observed in mdx and mdx:utrn dKO mouse muscle fibers (8, 23, 26).

Zebrafish larval tail-beat frequencies approach 60 Hz during burst swimming (6). To contract and relax at this frequency most likely necessitates the use of twitches to power swimming (2). The reduction in twitch force, the slowing of twitch force development, and the delayed relaxation of force following a twitch stimulus would therefore be expected to greatly impair the swimming performance of affected sapje larvae. Impaired relaxation could also be a contributor to the sarcomere disruption that characterizes the affected larvae as slowly relaxing myofibers may be actively lengthened by newly recruited contralateral fibers as the larvae swims.

Sensitivity to mechanical strain.

A heightened sensitivity to high-strain contractile activity is a defining characteristic of fast muscles of vertebrate DMD models, including mdx mice (10, 37, 40, 52), mdx:utrn dKO mice (11), and dystrophin-deficient canines (48, 53). By modeling recovery of affected larvae following a single eccentric contraction, we predicted long-term force deficits of up to 25% of initial force after a strain amplitude of 5% and up to 50% after 10% stain. Corresponding deficits for the unaffected larvae were predicted as only 7 and 12%. Thus, affected sapje larvae share a heightened susceptibility to high mechanical strain similar to mammalian models of DMD.

Single eccentric contraction protocols do not observe differences between wild-type and mdx mouse muscles until strain amplitude reaches 30% (10), which is considerably greater than the strain amplitude that induced force deficits in the affected sapje larvae. Canine muscles show an exaggerated sensitivity to eccentric contractions (48, 53) but require multiple contractions to reduce force to the levels observed in the present study. While care must be taken when comparing the single eccentric protocol used in the present study with multiple contraction protocols used by others, sapje appear more sensitive to mechanical strain compared with the mdx mouse and may even be more sensitive than dystrophin-deficient dogs.

In zebrafish larvae, dystrophin is concentrated at the ends of the myofibers where they interface with the vertical myosepta (4, 21). This myotendineous junction is the site of mechanical failure in dystrophin-deficient sapje larvae (4). The arrangement of fish muscle into myotomes, where short myofibers are arranged serially and separated by vertical myosepta, increases the relative number of potential failure points per unit muscle length compared with mammalian limb muscles. This may predispose dystrophin-deficient fish muscles to exaggerated injury during high-strain contractions. In addition to mechanical disruption of myofibers, several other factors contribute to excessive contraction-induced force losses in dystrophic mdx muscle, including elevated levels of reactive oxygen species (51) and contraction-induced disruptions in action potential propagation (7, 41). It is not known if these mechanisms contribute to the exaggerated posteccentric loss in force observed here for sapje larvae.

In a previous study, desmin levels of zebrafish were reduced 50% by morpholino knock-down (31). Muscles of morphant larvae showed abnormal morphology via confocal microscopy, yet they were more resistant to eccentric-induced injury. Thus, the muscle tissues resistance to mechanical strain depends upon the specific protein deficit rather than the existence of abnormal muscle structure. This implies that functional, rather than morphological, approaches will be required in future studies evaluating the susceptibility of zebrafish muscle to mechanical strain.

Using a criteria of three time constants, our data predicts that 3–25 min, depending on developmental age and strain magnitude, is required for the force of affected sapje preparations to stabilize following a single eccentric contraction. This prolonged partial recovery was not observed in unaffected preparations subjected to an identical protocol so it is not a generalized response of zebrafish muscle to a single eccentric contraction, nor can it be explained by any muscle fatigue that may have developed during the trial (as unaffected and affected protocols were identical). We interpret our data as an indication that high strain contractile activity entails an additional, slowly recovering mechanism(s) of force loss. Because partial force recovery occurs too rapidly to be attributed to protein synthesis, it must be due to a more transient mechanism. Ion channel activity, which is sensitive to stretch and known to be disrupted in the mdx mouse model (54), is one possibility although the milder mdx model shows much less force recovery following an eccentric injury (12) than observed here for sapje.

Developmental aspects of contractility.

All of the contractile properties reported here were stable for unaffected larvae across 4–7 dpf. In contrast, affected larvae showed some functional changes with increasing age. Most of these differences were between the 4 dpf larvae and later time points. In future studies, it may be advantageous to initiate functional assays at 5 dpf because any functional changes that occur between then and 7 dpf appear relatively minor. The exception to this was the susceptibility of larvae to mechanical strain. Here, affected larvae appeared to develop partial resistance to injury as they become older. Whether this represents a biological change, such as maturation of the extracellular matrix or other connective tissues, or the possibility that larvae with low resistance die earlier and are therefore not selected for analysis as the study progresses remains to be determined.

Comparison to previous work.

A previous study by Li et al. (32) examined the force produced by sapje larvae at a single 5 dpf time point. Our finding of a 60% reduction in twitch force at 5 dpf is in good agreement with their report of a 50% reduction. Li et al. also reported that the optimal sarcomere length for twitch force was shorter in sapje larvae, something that we have confirmed for tetanic force in the present study. We are unaware of any previous studies that have examined developmental effects on sapje larvae contractility. However, Sloboda et al. (44) reported developmental effects for wild-type larvae, with twitch force significantly lower at 3 dpf but similar between 4 and 7 dpf. In agreement, we found that twitch force of unaffected larvae was stable at 4–7 dpf.

Implications for drug discovery.

Other zebrafish models of neuromuscular disease, such as the dag1 model of dystroglycanopathy (19), candyfloss and other models of laminin-α2 deficiency (20, 22), the runzel model of titinopathy (47), and the relatively relaxed model of ryanodine receptor-1 related myopathies (14, 24), should all be amenable to the procedures outlined here. The present methodology is more labor and time intensive compared with the rapid birefringence assay used to identify candidate therapeutics in high-throughput screens and is therefore not a practical substitute for the birefringence assay. However, the present approach may be valuable for the validation of lead compounds after their initial identification via birefringence. Although this adds an extra step in the drug discovery process, it may ultimately be time and cost effective as it helps to identify the most physiologically promising candidates for advancement to preclinical testing in resource-intensive mammalian neuromuscular disease models.

Summary

Zebrafish larvae have many advantages for modeling DMD. The present work contributes to the expanding use of this species in two ways. It validates sapje larvae as the smallest vertebrate species for modeling the muscle dysfunction that characterizes DMD and it provides guidelines for assessing muscle force and sensitivity to mechanical strain in 4–7 dpf larvae.

GRANTS

Funding for this project was provided by the National Institutes of Health (NIH) Grants R01 AR-064300 (L. M. Kunkel), R01 AR-044345 (A. H. Beggs), and R01 HD-075802 (A. H. Beggs), the Wellstone Center (NIH U54 HD-060848, L. M. Kunkel), the Muscular Dystrophy Association, USA (MDA 383249, A. H. Beggs), the AUism Charitable Foundation (A. H. Beggs), RYR1 Foundation (A. H. Beggs), and the Boston Children's Hospital Intellectual and Developmental Disabilities Research Center (NIH P30 HD-18655).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

J.J.W., A.B., and L.K. conception and design of research; J.J.W., M.A., D.G., and G.K. performed experiments; J.J.W. and B.S. analyzed data; J.J.W., M.A., and B.S. interpreted results of experiments; J.J.W. prepared figures; J.J.W., M.A., and B.S. drafted manuscript; J.J.W., M.A., B.S., D.G., G.K., A.B., and L.K. edited and revised manuscript; J.J.W., M.A., B.S., D.G., G.K., A.B., and L.K. approved final version of manuscript.