Overexpression of ACTA1D286G-eGFP in zebrafish recapitulates the nemaline myopathy phenotype
We created a conditional transgenic zebrafish model for ACTA1 nemaline myopathy, Tg(ACTA1
D286G-eGFP) that expresses human ACTA1D286G in skeletal muscles. The Tg(ACTA1
D286G-eGFP)
high fish recapitulate hallmark pathological features of the disease with the presence of nemaline bodies in skeletal muscles at 48 and 72 hpf (Supplementary Figure S1). In addition to the trunk musculature, other striated muscles such as the facial, heart, and ocular muscles also contained numerous nemaline bodies (Fig. 1). At 72 hpf, we observed aggregates at the myosepta. These aggregates differ in appearance to the earlier-forming nemaline bodies, having a globular appearance rather than a clear and defined rod shape (Supplementary Figure S1).
Gomori trichrome staining of skeletal muscle demonstrated the presence of patches of intense purple staining, indicative of nemaline bodies, correlating with eGFP expression (Fig. 2a). Immuno-CLEM analysis revealed densely stained nemaline bodies that correlated with the eGFP-positive structures observed by confocal microscopy (Fig. 2b). Using conventional electron microscopy on Tg(ACTA1
D286G-eGFP) zebrafish, we also observed filamentous actin accumulations adjacent to the sarcomere and, occasionally, the sarcomeres appeared disorganized compared to those in Tg(ACTA1-eGFP) zebrafish (Fig. 2c).
Nemaline bodies are highly dynamic and transitory in Tg(ACTA1
D286G-eGFP) fish
To investigate the origins of the nemaline bodies within the cell, we carried out in vivo time-lapse analysis of the Tg(ACTA1
D286G-eGFP)
high fish. This showed, for the first time, that nemaline bodies form at the sites of muscle attachment at approximately 30 hpf and then extended into the muscle cell (Fig. 3a, Supplementary Movie S1). As development progressed, the nemaline bodies were seen to move throughout the cytoplasm (Fig. 3a, Supplementary Movie S1). The nemaline bodies were highly dynamic, rather than being tethered to the Z-disk or extracellular matrix as previously proposed [46, 57]. By 60 hpf, the characteristic rod-like bodies fragmented and disappeared from the cell, coincident with the formation of the globular aggregates at the myosepta (Fig. 3a, Supplementary Movie S2). Indeed time-lapse analysis of Tg(ACTA1
D286G-eGFP)
high fish showed that as nemaline bodies’ fragment, the fluorescent fusion protein previously incorporated within them can be encapsulated into globular aggregates, at the myosepta (Fig. 3a, Supplementary Movie S2).
In addition to the Tg(ACTA1
D286G-eGFP)
high zebrafish strain that showed high ACTA1D286G-eGFP expression, we also constructed a Tg(ACTA1
D286G-eGFP) low-expressing strain (Tg(ACTA1
D286G-eGFP)
low; Fig. 4b). We analyzed the frequency of nemaline bodies during early larval stages in both transgenic strains. We observed almost no nemaline bodies in the Tg(ACTA1
D286G-eGFP)
low strain correlating with the much lower level of ACTA1D286G-eGFP expression (Figs. 3b, 4b). Contrastingly, the Tg(ACTA1
D286G-eGFP)
high strain showed the presence of nemaline bodies at 2 dpf, followed by a dramatic reduction at 4 dpf (Fig. 3b). Quantification of the aggregate phenotype identified an increase in the percentage of fish developing globular aggregates in skeletal muscle at 4 dpf, coincident with the disappearance of nemaline bodies (Fig. 3c). This supports the suggestion that early forming nemaline bodies may be one source of the ACTA1 that forms the globular aggregates.
To determine if the different aggregates could be distinguished immunologically, we performed antibody and phalloidin (labeling F-actin) staining on zebrafish muscle expressing ACTA1D286G-eGFP. Although nemaline bodies could be detected at 2 dpf using an antibody against eGFP, they were not detected by Actinin2 or Actinin3 antibodies, nor by phalloidin (Fig. 5a, Supplementary Figure S2). In contrast, the globular aggregates at 4 dpf were positive for Actinin2, Actinin3 and phalloidin (Fig. 5a, Supplementary Figure S2). We also co-expressed the cardiac α-actin cDNA fused with mCherry (actc1a-mCherry) together with ACTA1D286G-eGFP in the skeletal muscle and showed that both types of aggregates incorporated cardiac actin (Supplementary Figure S2). To determine if similar α-actinin-negative nemaline bodies were present in patients with mutations in ACTA1, we tested patient skeletal muscle biopsies. As observed in Tg(ACTA1
D286G-eGFP)
high zebrafish, we found that nemaline bodies can exhibit different protein signatures. In a biopsy from a patient with a T66I mutation in ACTA1 [18], nemaline bodies darkly stained by Gomori trichome are both phalloidin and actinin2 positive (Fig. 6). Whereas for a patient with an I136M mutation in ACTA1 [18], nemaline bodies darkly stained by Gomori trichome are phalloidin positive and either actinin2 positive or negative (Fig. 6).
Cytoplasmic actin aggregates correlate with reduced skeletal muscle function
Using time-lapse analysis, we observed the formation of globular aggregates in the skeletal muscle and hypothesize that these may contribute to muscle weakness. To test this, we overexpressed a human wild-type ACTA1 eGFP-tagged transgene in the zebrafish skeletal muscle (Supplementary Figure S1). We constructed transgenic strains with high and low levels of ACTA1wildtype-eGFP expression (Tg(ACTA1
wildtype-eGFP)
high and Tg(ACTA1
wildtype-eGFP)
low, respectively) (Fig. 4b). In contrast to the rod-shaped nemaline bodies observed in Tg(ACTA1
D286G-eGFP)
high fish, expression of ACTA1wildtype-eGFP never induced the formation of rod-like structures (Supplementary Movie S3). Rather, Tg(ACTA1
wildtype-eGFP)
high fish only produced globular aggregates that formed at the myosepta and showed an identical immunological signature to those observed in Tg(ACTA1
D286G-eGFP)
high fish (Fig. 5, Supplementary Figure S2). This demonstrates that globular aggregates are a result of increased ACTA1 and not specifically caused by the ACTA1
D286G mutation. These globular aggregates were formed at earlier developmental stages in Tg(ACTA1
wildtype-eGFP)
high fish (Supplementary Movies S3 and S4) compared to Tg(ACTA1
D286G-eGFP)
high fish (Supplementary Movie S2).
To assess the pathogenicity of globular aggregates in the skeletal muscle, we examined muscle function in both Tg(ACTA1
wildtype-eGFP) and Tg(ACTA1
D286G-eGFP) transgenic strains at 2 and 6 dpf. Using a touch-evoked escape response assay, we were able to determine the maximum acceleration of 2 dpf larvae as a direct measure of muscle force. We found a significant decrease in the maximum acceleration of both Tg(ACTA1
wildtype-eGFP)
high and Tg(ACTA1
D286G-eGFP)
high compared to control zebrafish, demonstrating that cytoplasmic actin is pathogenic (Fig. 4a). We also recorded the swimming performance at 6 dpf and observed a significant decrease in the number of movements and in the distance traveled by both Tg(ACTA1
wildtype-eGFP)
high and Tg(ACTA1
D286G-eGFP)
high strains compared to controls (Fig. 4d).
In addition, we assayed the activity of the low-expressing zebrafish strains. While the swimming abilities of Tg(ACTA1
wildtype-eGFP)
low fish were comparable to control fish, Tg(ACTA1
D286G-eGFP)
low fish displayed significantly reduced swimming at 6 dpf (Fig. 4c, d). This reduction in swimming activity correlated with the level of transgene expression (Fig. 4b). These findings demonstrate that cytoplasmic actin aggregates correlate with reduced skeletal muscle performance.
We next examined a model of nemaline myopathy resulting from a loss of Neb to determine whether actin aggregates were a common feature associated with nemaline myopathy. We injected two neb splice-site-targeting morpholinos (both singularly and in combination) into wild-type zebrafish. The efficacy of the morpholinos was confirmed by qRT-PCR showing that neb mRNA levels are reduced by approximately 80 % by each morpholino and approximately 90 % when both morpholinos are used in combination (Fig. 7b, c). The effect on Neb was further confirmed by examining sarcomere length, with Neb morphants (Fig. 8a) displaying significantly shorter sarcomere lengths compared to controls (Fig. 7d, e).
When neb morpholinos were injected singularly and in combination into both the high- and low-expressing Tg(ACTA1
wildtype-eGFP) strains we observed an increase in eGFP-positive accumulations at the myosepta (Fig. 8a). These accumulations were also phalloidin positive (Fig. 8b), akin to those observed in Tg(ACTA1
D286G-eGFP)
high and Tg(ACTA1
wildtype-eGFP)
high zebrafish, and were not observed in control embryos.
Loss of α-actin results in Actinin2 accumulation and reduced skeletal muscle function
Our analysis has shown that overexpression of ACTA1D286G produces nemaline bodies and causes detrimental effects on muscle function. However, nemaline bodies have been reported in patients carrying recessive ACTA1 mutations leading to an absence of skeletal muscle α-actin protein. Skeletal muscle biopsies from these patients contain disorganized myofibrils and nemaline bodies, presumably caused by an imbalanced stoichiometry between sarcomeric components [32].
Using our zebrafish system, we also created a model for recessive nemaline myopathy. Zebrafish possess two skeletal muscle α-actin genes (acta1a and acta1b) and two cardiac α-actin genes (actc1a and actc1b), which are all expressed during early muscle development. qRT-PCR analyses showed that actc1b is the predominant isoform expressed in the skeletal muscle at 2 dpf (Fig. 9a). Thus, we chose to knockdown Actc1b to reduce the amount of α-actin in skeletal muscle.
We injected two morpholinos (one targeting the exon 2 splice donor site and one targeting the 5′UTR). We observed a reduction in α-actin protein by approximately 43 % (Fig. 9b, c) along with decreased phalloidin staining in Actc1b morphants compared to control zebrafish (Fig. 10). Morphologically, Actc1b morphants (Fig. 11a) display impaired skeletal muscle function, showing a reduction in both maximum acceleration at 2 dpf and in swimming distance at 6 dpf (Fig. 11b, c).
EM analyses revealed the presence of electron-dense projections extending from the myosepta (Fig. 12). Thickened Z-disks were clearly evident and in some cases the electron-dense region appeared to span the full width of a sarcomere (Fig. 12). We also observed disorganized and broken myofibrils surrounded by large numbers of mitochondria in the Actc1b morphant skeletal muscle compared to control skeletal muscle (Fig. 12). We performed antibody and phalloidin staining on the skeletal muscle and identified phalloidin- and Actinin2-positive nemaline bodies, concentrated near the myosepta, as well as Actinin2-positive aggregates and thickened Z-disks throughout the myofibrils in Actc1b morphants that are not present in the control skeletal muscle (Fig. 10).
We injected the actc1b morpholino into the Tg(ACTA1
wildtype-eGFP)
low strain and analyzed the results using in vivo time-lapse imaging. We showed that unlike early forming nemaline bodies in Tg(ACTA1
D286G-eGFP) muscle, these nemaline bodies were stable in the skeletal muscle, forming from 32 hpf. They were located both at the myosepta and scattered throughout the muscle fibers (Supplementary Movie S5) and were not observed in control embryos (Supplementary Movie S6).
ACTA1D286G is mislocalized and more rapidly exchanged in the sarcomere
We showed that an imbalance of actin levels in the skeletal muscle results in aggregates and muscle weakness. However, we also wanted to determine whether overexpression of ACTA1-eGFPD286G has an additional, mutation-specific, effect in the skeletal muscle. Previous studies have suggested that ACTA1D286G may have a reduced ability to polymerize and be incorporated into the sarcomere [7, 49]. To examine the dynamics of ACTA1D286G in Tg(ACTA1
D286G-eGFP)
high zebrafish, we performed FRAP analysis on both the thin filament and the Z-disk at 2 dpf. We found a significant reduction in the average time taken for 50 % recovery at both locations following photobleaching for ACTA1D286G-eGFP (filament: 22.74 ± 10.22 s, Z-disk: 52.38 ± 9.19 s) compared to Tg(ACTA1
wildtype-eGFP)
high fish (filament: 106.7 ± 34.6 s,: Z-disk: 85.79 ± 16.71 s) (Fig. 13c). This showed that the ACTA1D286G protein is readily incorporated into the sarcomere, but is more rapidly exchanged at both the Z-disk and thin filament than ACTA1wildtype, suggesting it may be less stably associated.
During the FRAP analysis, we observed that the localization of the two proteins within the sarcomere differed. While the ACTA1wildtype-GFP protein demonstrated a clear striated pattern, with well-defined Z-disk structures, ACTA1D286G-eGFP showed more generalized distribution throughout the length of the thin filament (Fig. 13a). We quantified this difference by comparing the fluorescence intensity of the Z-disk to the rest of the filament in both Tg(ACTA1
D286G-eGFP)
high and Tg(ACTA1
wildtype-eGFP)
high fish. We found a significant decrease in the ratio of Z-disk/filament eGFP intensity in Tg(ACTA1
D286G-eGFP)
high (1.3 ± 0.07) compared to Tg(ACTA1
wildtype-eGFP)high (1.6 ± 0.1) fish. This finding suggested that ACTA1D286G is preferentially localized along the thin filament (Fig. 13b).