We studied the behaviour of 7 myopathy-related actin mutants in C2C12 myotubes, resulting in a wide range of different phenotypes that in many cases changed during differentiation of the cells. Previous studies have mainly used undifferentiated myoblast or fibroblast cell lines [7
], although Ilkovski et al. [8
] have studied V163L and four other mutants in myotubes. In general, the phenotypes described in these studies matched our findings in the undifferentiated myoblasts, with some differences in how pronounced the effect appeared in the cells (see below).
Notably, none of the mutants affected differentiation, the total proportion of differentiated cells was always at least 60% (the value for wt-actin), for some mutants even as high as 80% (G268R, D286G, Table ). However, the appearance of most of the mutants investigated differed before and after differentiation (Fig. ).
Generally, most of the mutant actin isoforms showed better incorporation into the actin cytoskeleton in differentiated myotubes than in undifferentiated myoblasts (Fig. ). This interesting result raises the question, of whether some of the previously described effects are only observed after short-term expression of actin mutants in an in vitro cell culture model, but are not relevant for the real situation in skeletal muscles.
The mutants I64N and N115S are not readily integrated into stress fibres, since they produced a diffuse cytoplasmic staining in the majority of undifferentiated myoblasts. For the I64N-mutant this differs slightly from the result by Costa et al., who found some integration into fibroblast stress fibres as well as long cytoplasmic rods [7
]. Cytoplasmic rods were also observed in our study in a smaller subset of myoblasts, so we suspect these small differences in appearance may be due to the use of different epitope tags or vectors with somewhat different expression levels (N-terminal myc-tag in Costa et al. and C-terminal EGFP-tag in our study). In case of I64N, poor integraton into stress fibres fits nicely with the predicted polymerization defect [5
], also confirmed biochemically in co-polymerization experiments with wt-actin [7
]. The N115S mutation is likely to affect the closure of the nucleotide binding cleft, a defect that could also result in a decreased polymerization propensity [5
]. However, both isoforms are capable to contribute to sarcomeric structures, since they showed good integration after several days of differentiation (Fig. ). Only in 5 % of the myotubes an abnormal phenotype could be observed.
Generally, there is poor correlation between the number of myotubes with aberrant actin structures and the severity of the disease caused by a specific mutant. The G268R mutant behaved similarly to the wildtype in undifferentiated myoblasts (Fig. ), confirming the results described in the other expression studies [7
]. After differentiation, even 72 % of the cells showed good incorporation into actin structures (versus 57% for wt-actin), despite being linked to a severe form of NEM. Because of their pathologically unremarkable phenotype in differentiated myotubes, the NEM-linked mutants I64N, N115S and G268R are likely to be incorporated into sarcomeric structures in patient muscles. Therefore, we suggest that they exert a dominant effect via a functional defect (e.g. contraction defect), rather than mislocalisation. Alternatively, differentiation for 4–6 days might still be not sufficient to show aberrant structures produced by these mutants.
The D286G mutant, which is linked to severe NEM, appeared delocalized in most of the myoblasts (Fig. ), similar to I64N and N115S. However, unlike the other mutants it did not integrate in the majority of myotubes during differentiation, but instead formed big, rod-like aggregates in the cytoplasm (Fig. ; long arrows). These rods did not stain with phalloidin, and appeared similar to the aggregates described by Costa et al. in fibroblasts. D286 is located near the hydrophobic pocket, so a profound polymerization defect is likely, which could result in the aggregation of unpolymerized actin. This prediction fits well to the behaviour of the mutant within differentiated cells and to the severity of the disease. It would be interesting to investigate whether patients with this mutation had a higher level of aggregates within muscle biopsies and whether these contained filamentous actin or aggregated actin.
One characterizing feature of NM and also used for diagnostic means is the presence of nemaline rods and bodies in the sarcoplasma of patient skeletal muscles. These giant accumulations are thought to emerge from Z-lines and apart from actin thin filaments are largely composed of alpha-actinin and other Z-line proteins [1
]. Since the transfection of several NEM-linked mutants resulted in the formation of aggregates within the cytoplasm of the cells (Fig. , ), it has been hypothesized that these correspond to the cytoplasmic nemaline bodies found in patient skeletal muscles [7
]. However, in our expression studies there are clearly different kinds of accumulations: aggregates that co-stain with phalloidin (V163L, H40Y, G15R), big rods that do not co-stain (I64N, D286G), and small, punctuate aggregates as sometimes found with the wild-type or the phenotypically unremarkable mutants (N115S, G268R), also not co-staining with phalloidin. Aggregates that stain with phalloidin contain actin filaments, as it would be expected in nemaline bodies. The lack of co-staining could either indicate accumulations that are composed of unpolymerized actin or that are not accessible for phalloidin. To further investigate the correlation between the aggregates found in in vitro
studies and in patient muscles we performed co-staining with alpha-actinin as the main component of nemaline bodies. Generally, there was no co-localization of alpha-actinin with any kind of aggregates formed by the actin mutants, especially the giant rods found with the D286G mutant (Fig. ). Very rarely, some co-staining with aggregates produced by the V163L-isoform was observed (Fig. ), however, since that occurred only in some cases and was always very faint, the significance of this finding is not clear. The fact that no co-staining was found in the differentiated myotubes is especially interesting, since here early sarcomeric structures including Z-lines have already been constituted, thus providing the original environment for the formation of nemaline bodies. One possible explanation for the lack of co-staining is that the accumulations are not accessible for the antibody. However, we suggest that the propensity of some mutants to form different kinds of rods and aggregates in vitro
might reflect a specific molecular defect rather than having any connection with the nemaline bodies found in patient muscles. This view is supported by the fact, that the appearance of nemaline rods can vary from 1 % to virtually all fibres and does not correlate with the degree of muscle weakness [14
]. Moreover, they are also found in other myopathy diseases as well as in small numbers of normal muscles [1
A slightly different situation applies for intranuclear rods. In our study, the only mutants that produced intranuclear aggregates are the H40Y and V163L-isoforms, which also cause IRM in patients [15
]. This agrees with the study from Ilkovski [8
] who found that high levels of insoluble V163L appeared very early after transfection of myoblasts and persisted in myotubes. This indicates that the ability to form accumulations within the nucleus could be an inherent characteristic of some mutants and directly correlated to the appearance of IRM in patients. However, Ilkovski et al. reported nuclear aggregates in myoblasts transfected with the R183G-mutant, which were not found in the muscle biopsy of a patient suffering from severe NEM caused by this isoform [8
]. An issue here could be that only limited biopsy material is available from patients, and biopsies tend to be highly variable. Interestingly, nuclear aggregates produced by H40Y-actin in myoblasts and fibroblasts were never observed in differentiated cells, and also did not represent the dominant phenotype in V163L-transfected myotubes (Fig. ), arguing against a direct relation between those aggregates and the rods observed in patient muscles. The molecular reason for the formation of intranuclear rods is still elusive, but a connection has been suggested to the previously reported stress-induced translocation of actin into the nucleus [16
]. Similar to nemaline bodies, intranuclear rods might be the result of a general response of skeletal muscle fibres to certain pathological situations. Thus, the appearance of intranuclear aggregates in vitro
might be an indicator for a link to IRM, but cannot be used as an established diagnostic tool.
In addition to rod formation there was another interesting feature predominantly seen in the differentiated myotubes, that is the appearance of aberrant stress fibres. In no case did we observe wavy or otherwise abnormal stress fibres in the myoblasts, but especially for the H40Y-, V163L- and the AM-linked G15R-mutant they were present in a large subset of myotubes (Fig. ). At least the V163L actin also appeared striated, as if incorporated into sarcomere-like structures (Figure , inset). The exact phenotype ranges from fragmented and shortened to wavy and thickened stress fibres. The lack of those aberrant structures in undifferentiated cells indicates that they develop after an initial integration of mutant molecules into endogenous actin structures, which eventually induce the formation of abnormal capacities. Fig. supports this view, since the middle part of the myotube shows normal incorporation of the H40Y-mutant into stress fibres (long arrow), whereas at the far end of the cell, some fragmentation might have occurred (short arrow). Aberrant stress fibres are not seen in undifferentiated cells, but appear for all of the mutants investigated in varying proportions within differentiated myotubes (Fig. ). The percentage of cells containing wavy stress fibres seem to be loosely linked to the severity of the disease, with the highest numbers coming up for the mutants H40Y, G15R and V163L, all causing severe forms of myopathy. A "poisonous" effect of mutant actin isoforms has been previously suggested because of the fact that isoforms were found to be expressed and present within insoluble actin filaments from patient muscles [8
] and the recessive nature of "loss-of function" actin mutants with severely defective folding [7
Unfortunately, we cannot rule out the possibility that differential expression levels among the mutants may contribute to the phenotypes seen. The best system would use the endogenous actin promoter to express mutant actins in cells, but we have not yet succeeded in this aim.