Drosophila small heat shock protein CryAB ensures structural integrity of developing muscles, and proper muscle and heart performance

Most heat shock proteins operate as molecular chaperones in stress conditions and play a central role in the maintenance of normal cellular function by preventing the aggregation of misfolded proteins into large deleterious complexes (Morimoto, 1998). However, some heat shock proteins are also activated in a stress-independent way to control the transport, folding, assembly and degradation of various proteins (Colinet et al., 2010; Lindquist and Craig, 1988; Mymrikov et al., 2010; Sorensen et al., 2003). The cell-specific expression of small heat shock proteins (sHsps) during development, cell differentiation or after activation by growth factors and oncogenes has been reported in a wide range of organisms, including Drosophila and mammals (Michaud and Tanguay, 2003; Morimoto, 1998). For example, the Drosophila sHsp Hsp22, in spite of its heat-induced expression and nonspecific chaperone activity, is regulated during normal development and plays a role in protection against apoptosis (Colinet et al., 2010; Morrow and Tanguay, 2003).

In human, seven sHsps are expressed in cardiac and skeletal muscles, including αB-crystallin (CRYAB), HSPB2 and HSPB3; they are induced during the initial phase of skeletal muscle differentiation, controlled by MYOD1, suggesting a muscle-specific function in development. Also, Hspb2 and CryAB proteins display sarcomeric localizations; they are localized on Z-bands or I-bands of the skeletal and cardiac muscle myofibrils, maintaining and protecting contractile cytoskeletal structures (Mymrikov et al., 2010; Sugiyama, 2000).

CryAB displays particularly intriguing cell-specific functions, acting as a chaperone for intermediate filament (IF) proteins such as desmin in muscle cells and vimentin in eye lenses. IFs are known to overlie Z-discs and M-lines, stabilizing myofibrils and protecting them from mechanical insult (Goldfarb and Dalakas, 2009). IFs also link membranous organelles, such as nuclei and mitochondria, and play an important role in maintaining their correct morphology and positioning.

CryAB stabilizes IFs and prevents them from aggregating under stress conditions, and also assists IFs during developmental reorganizations (Nicholl and Quinlan, 1994). Importantly, mutations in CryAB, such as replacement of arginine by glycine at conserved position 120 (R120G), lead to altered CryAB-IF interactions responsible for desmin-related myofibrillar myopathy (DRM) (Dalakas et al., 2000; Goldfarb and Dalakas, 2009; Goldfarb et al., 2004). The R120G missense mutation alters the spatial organization of CryAB, leading to the loss of its chaperone activity and the formation of desmin aggregates in muscle fibers (Vicart et al., 1998; Perng et al., 2004). As a result, the desmin IF network is severely affected, leading to altered myofibril alignment, irregular sarcomere architecture and abnormal mitochondrial organization (Goldfarb and Dalakas, 2009). In DRM patients this leads to progressive muscle weakness, including the limb, neck and respiratory muscles, for which no appropriate treatments have yet been developed.

To further characterize muscle-specific roles of CryAB and better understand DRM-associated muscle defects we took advantage of its evolutionary conservation to analyze the function of the Drosophila CryAB counterpart. The fruit fly Cryab ortholog, CryAB, which is also known as lethal (2) essential for life [l(2)efl], has been shown to be involved in the modulation of insulin signaling and the regulation of lifespan (Flatt et al., 2008) but its expression and function in differentiated muscle have not been analyzed.

Here we show that, like its vertebrate counterpart, the Drosophila CryAB protein displays stress-independent muscle-specific expression. In larval muscles it accumulates around the myonuclei and at the level of Z-bands and is required for correct nuclei localization, myofibril integrity and muscle performance. CryAB features a conserved actin-binding domain and is required for the correct sarcomeric pattern of actin filaments and also of the Z-band-associated actin crosslinker Cheerio. CryAB binds to Cheerio in muscles, indicating that this interaction might help to maintain myofibrillar integrity in Drosophila muscle. We also report that muscle-targeted expression of CryAB carrying the DRM-associated R120G mutation leads to severe alterations in the striated actin pattern, reduced muscle performance and marked cardiac arrhythmia. These observations point to a conserved stress-independent role of CryAB in the maintenance of muscle cell cytoarchitecture. Thus, our findings provide evidence that the Drosophila sHsp CryAB shares not only sequence but also functional similarities with its human counterpart, suggesting that Drosophila can serve as a model system for dissecting molecular determinants of myofibrillar myopathies.

The Drosophila sHsp genes are clustered within the 67B region on the left arm of the third chromosome (Corces et al., 1980), except for CryAB [l(2)efl], which is located on the right arm of the second chromosome (Kurzik-Dumke and Lohmann, 1995). Drosophila CryAB shows significant homology to human CRYAB (Fig. 1A) and to other sHsp members, in particular the Hspb1 and Hspb8 subfamilies (Fink et al., 2009; McWilliam et al., 2013) (Fig. 1C). CryAB encodes a protein of ∼22 kDa (Fig. 1B) that has a highly conserved α-crystallin domain located in the C-terminal portion of the protein (Fig. 1A), similar to other sHsps (Ingolia and Craig, 1982). Within this domain, CryAB features several conserved motifs, including an actin-binding domain, a capping box and the residues R120 and D109 involved in desminopathies (Fig. 1A).

In addition to coordinately induced expression in response to a heat stress, several members of the sHsp family show a specific pattern of expression in diverse tissues and cells (Marin and Tanguay, 1996; Michaud and Tanguay, 2003). To determine whether CryAB also displays stress-independent expression during development we first analyzed its expression pattern in third instar larvae. Using a polyclonal antibody raised against the whole CryAB protein, we found that CryAB is abundantly expressed in the larval body wall muscles (Fig. 2). Within an individual larval muscle, CryAB accumulated in the perinuclear area (Fig. 2A) and displayed a striated sarcomeric pattern at the level of Z-lines and M-lines (Fig. 2A-D). We confirmed this observation by the immunogold detection of CryAB on the ultrathin sections visualized in electron microscopy (EM) (supplementary material Fig. S1). When analyzing CryAB distribution on longitudinal optical sections through the muscle fibers, we also found that it accumulated around the nuclei and enveloped the external surface of myonuclei (Fig. 2E). We observed a granular pattern of CryAB both around the nuclei (Fig. 2E) and at the Z-discs (Fig. 2B,D), indicating that it might be part of large protein complexes. 3D reconstruction of confocal scans (Fig. 2F) revealed that the Z-line-associated CryAB is mainly localized at the external level of Z-discs. This developmental expression pattern of Drosophila CryAB is reminiscent of that of its human ortholog (Dubin et al., 1990).

To assess the role of CryAB in larval muscles we applied RNAi-mediated muscle-specific gene attenuation (Fig. 3I) using two different UAS-CryAB-RNAi lines (VDRC 40532 and 107305) crossed to the Mef2-Gal4 driver (Fig. 3; supplementary material Fig. S2). We observed that most Mef2>CryAB-RNAi larvae (from both RNAi lines examined) exhibited defects in sarcomeric organization, characterized by an irregular, fuzzy pattern of phalloidin staining in the large muscle segments (Fig. 3C,H; supplementary material Fig. S2). This aberrant organization of sarcomeric actin is consistent with the fact that CryAB has an actin-binding domain (ABD) (see Fig. 1), which appears to play an important role in actin filament organization assuming that muscle-targeted expression of CryAB carrying a mutated ABD leads to RNAi-like phenotypes (supplementary material Fig. S3). Moreover, CryAB attenuation also leads to muscle splitting (Fig. 3B; supplementary material Fig. S4A), which in some cases is associated with altered muscle attachments (Fig. 3B) or even with the loss of affected muscle (supplementary material Fig. S4B). The observed muscle defects are not due to apoptotic events, as we do not detect Caspase 3 activation associated with muscles in the CryAB-RNAi context (supplementary material Fig. S2I). Altogether, these findings demonstrate a role of CryAB in the maintenance of muscle fiber integrity.

To better characterize the influence of CryAB attenuation on the sarcomeric pattern we used Msp300 antibody to reveal the Z-bands (Fig. 3D,E). We found that, in the muscle segments displaying fuzzy phalloidin staining, the Z-band-associated signal of Msp300 was weak or lacking (Fig. 3E). We also noted that Msp300 was irregularly distributed around the myonuclei (Fig. 3D,E), suggesting that it might potentially interact with CryAB. We tested this possibility by co-immunoprecipitation (co-IP) (supplementary material Fig. S5) and found that there is no physical interaction between these two Z-disk-associated proteins. Gaps in Z-bands were also detected when examining EM sections of CryAB attenuated muscles (Fig. 3F,G), supporting the view that CryAB acts to stabilize Z-band-associated sarcomeric components.

Assuming that each of the larval muscles displays a specific number of sarcomeres (Bate, 1990), we also tested whether muscle-targeted CryAB knockdown impacted on sarcomere number. We focused on three distinct muscles (VL3, VL4 and SBM) and found that, in all cases, the number of sarcomeres was significantly reduced (supplementary material Fig. S4A), whereas, except for the fuzzy pattern areas, the size of sarcomeres was unaffected (supplementary material Fig. S4B). Thus, CryAB also appears to play a role in larval muscle growth, which occurs by the addition of sarcomere units.

Next we investigated whether the cellular organization and the performance of muscles are affected by the reduction of CryAB function. Previous studies demonstrated that the localization of nuclei is crucial for proper muscle performance (Metzger et al., 2012) and that mitochondria are highly organized in muscles (Picard et al., 2012). The accumulation of CryAB around the nuclei (Fig. 2) prompted us to test whether it was involved in the correct distribution of myonuclei within the muscle fibers. We found that in the CryAB-RNAi context they were irregularly distributed along the muscle fibers (Fig. 4A,B). Nuclei are also misarranged in muscles overexpressing CryAB with a mutated ABD (supplementary material Fig. S3), providing evidence that the actin binding function of CryAB is essential for nuclei localization.

Cheerio, which makes a link between perinuclear actin and cytoplasmic actin cables, is involved in nuclei positioning in Drosophila nurse cells (Huelsmann et al., 2013). It is also well known that in vertebrate muscles the Cheerio ortholog Filamin C is associated with Z-bands and when mutated leads to myofibrilar myopathies (Goldfarb and Dalakas, 2009). We tested whether Cheerio is expressed in larval muscles and whether CryAB influences its localization. Cheerio was detected aligned with Z-bands (Fig. 4C), but its pattern was more expansive than the sharp Z-line revealed by phalloidin. As in nurse cells, it also accumulated in the perinuclear area. However, in muscles attenuated for CryAB the sarcomeric localization of Cheerio was severely affected, with pronounced accumulation around the mislocalized nuclei (Fig. 4D), strongly suggesting that CryAB is required for Cheerio localization. In addition, the mitochondrial signal appeared significantly reduced in CryAB-RNAi muscles compared with the wild type (Fig. 4E,F), indicating impaired respiratory functions. This possibility was supported by EM analysis revealing mitochondrial swelling, broken cristae and glycogen deposits (Fig 4G,H).

To test whether the described morphological alterations affected the locomotive abilities of larvae, we carried out three behavioral tests. We measured the speed of crawling and found that CryAB-RNAi larvae needed ∼20% more time to travel the appointed distance (Fig. 4I). We also measured the time needed for each larva to right itself from the dorsal to ventral position. In this test, CryAB-attenuated individuals took twice as long (Fig. 4J). Finally, the number of peristaltic movements of crawling larvae was counted: during a 30 s period, CryAB-RNAi specimens executed at least 20% fewer peristaltic movements than controls (Fig. 4L). To confirm the contraction disturbance, we also compared the length of muscle fibers relaxed by EDTA with that of contracted muscle fibers. This showed that the analyzed muscles from CryAB-RNAi individuals had a significantly lower contractility index (Fig. 4K). Thus, these data show that CryAB is required for proper muscle contraction and efficient motility. Finally, we found that these functions of CryAB in muscles also influence the survival of flies (supplementary material Fig. S3C).

DRM is caused by the missense R120G mutation within the α-crystallin domain of CRYAB. It is due to intracellular aggregations of desmin and Vimentin IFs, which abnormally interact with the mutated CRYAB (Simon et al., 2007; Zobel et al., 2003). Comparison of amino acid sequences of human and Drosophila CryAB proteins showed conservation of arginine at position 120 (Fig. 1A) and prompted us to test whether the R120G mutation in Drosophila CryAB could mimic muscle defects observed in DRM patients. We generated transgenic lines carrying either the wild-type or mutated form of CryAB fused to GFP (Fig. 5A-D″).

Compared with endogenous CryAB (Fig. 2), both CryAB-GFP and CryAB^R120G-GFP displayed irregular accumulations along the muscle fibers in addition to their localization in sarcomeres (Fig. 5A′-D′). We therefore tested whether these accumulations affected sarcomeric organization. In the area where the wild-type CryAB-GFP accumulated, the sarcomeric pattern marked with phalloidin appeared normal, although the intensity of actin bands was attenuated (Fig. 5A-A″, asterisk). Occasionally, we observed misalignments of myofibrils (Fig. 5A, arrowheads), suggesting that the overexpression of CryAB leads to a local destabilization of myofibrillar organization. In contrast to these mild phenotypes, the accumulations of CryAB^R120G resulted in severely altered sarcomeric patterns (Fig. 5B-D″), with virtually all muscles affected. We observed a striking complementary distribution of CryAB^R120G with respect to sarcomeric actin along the muscles. Large segments of muscle fibers with high levels of CryAB^R120G displayed reduced phalloidin staining with abnormally interspaced and irregular Z-bands (Fig. 5B, arrowheads), whereas the neighboring CryAB^R120G-free regions were characterized by a high-intensity phalloidin signal probably resulting from the disorganization and local compaction of sarcomeres (Fig. 5B, arrows). In addition to localizing to the Z-bands, CryAB^R120G accumulated in either a fuzzy pattern (Fig. 5B′,B″) or a punctate pattern suggesting the formation of aggregates (Fig. 5C′,C″). We also observed small patches of accumulated CryAB^R120G (Fig. 5D-D″) in which the sarcomeric pattern was not particularly affected. However, we noted that within each sarcomere CryAB^R120G was not only present in the Z-bands but also in two additional stripes between the Z-bands (Fig. 5D″), and that myofibrils were misaligned (Fig. 5D, arrowheads).

These alterations prompted us to test whether the mutated form of CryAB also affected larval muscle function. We tested muscle performance using the righting, motility and journey tests. In all these tests (Fig. 5E-G), Mef2>CryAB^R120G larvae showed markedly reduced muscle performance than those expressing wild-type CryAB in muscles. We also found that muscle-targeted expression of CryAB^R120G affected muscle contractility (Fig. 5H). The fact that several phenotypes of CryAB^R120G overexpression are similar to those observed after knocking down CryAB suggests that CryAB^R120G acts as a dominant negative.

A recent study demonstrated that when the mutated human CRYAB is expressed in adult Drosophila heart it mimics DRM-related cardiac defects such as arrhythmia and an increase in systolic heart diameter (Xie et al., 2013). However, it remained unclear whether these defects were due to reduced functions of the endogenous Drosophila CryAB gene. We therefore tested whether CryAB is expressed in the adult Drosophila heart and if it is required for normal heart function.

We found that CryAB protein is present in a striated pattern in both the transverse myofibrils of the myocardium and in the layer of the longitudinal muscle fibers that is associated with the heart (Fig. 6A). This CryAB expression plays an important role in the adult heart because the heart-specific attenuation of CryAB resulted in an accelerated cardiac rhythm (Fig. 6B,E). The video-captured M-modes of hearts revealed a reduction in the heart period following RNAi-mediated CryAB knockdown, as compared with control flies (Fig. 6E). The heart period was strongly reduced in young flies (Fig. 6B,E) and continued to be significantly shortened in aged Hand>CryAB-RNAi animals (data not shown). The shortened heart periods in CryAB-attenuated flies were mainly due to decreased diastolic and systolic intervals (Fig. 6E; supplementary material Fig. S6). Conversely, the heart periods of flies overexpressing the wild-type or R120G form of CryAB were unaffected or only slightly (non-significantly) decreased (Fig. 6B).

However, heart-specific expression of CryAB^R120G led to a significant increase of arrhythmia in 1-week-old flies (Fig. 6C), a phenotype similar to that observed in transgenic Drosophila lines expressing the mutated form of human CRYAB in the heart (Xie et al., 2013). Additionally, altering the expression of CryAB or introducing CryAB^R120G in the heart resulted in enlarged heart diameter (Fig. 6D; supplementary material Fig. S6) and significantly reduced the fractional shortening in CryAB^R120G flies, impairing systolic function (supplementary material Fig. S6). Thus, this analysis reveals a novel role of CryAB in the regulation of heartbeat and demonstrates that introducing the R120G mutation into Drosophila CryAB leads to pathological cardiac arrhythmia and mimics cardiac defects observed in DRM patients.

The altered Cheerio pattern in larval muscles with attenuated CryAB prompted us to test whether these two proteins physically interact. We first performed a co-IP experiment with anti-CryAB antibody followed by proteomics analyses. Peptide sequencing of co-immunoprecipitated proteins ranging from 30 to 170 kDa revealed that not only actin but also Tropomyosin 1 and 2 and a few other sarcomeric components, including α-Actinin, Myosin heavy chain and Paramyosin, were strongly represented in CryAB-containing protein complexes (supplementary material Table S1). Importantly, mass spectrometry data also revealed nine matches in Cheerio, corresponding to six different peptides all located in the C-terminal region of the protein (Fig. 7A). cheerio (cher) is the Drosophila homolog of human filamin (Sokol and Cooley, 1999) and encodes several isoforms ranging from 90 to ∼260 kDa, all of which are present in third instar larvae (Fig. 7B). The cher-90 isoform is devoid of an ABD and the Rod1 repeat region, and is made up of the C-terminal part of cher-240 (Fig. 7A). Since all the identified peptide sequences matched this common C-terminal part, each Cheerio isoform can potentially interact with CryAB.

To confirm the proteomics data, we performed co-IP experiments with anti-CryAB antibody using protein extracts from the dissected third instar larvae expressing GFP-tagged cher-90 in muscles. We found that immunoprecipitated CryAB protein complexes did indeed contain Cheerio (Fig. 7B). We also tested whether CryAB colocalized with cher-240 in sarcomeres using the cher^CPTI0847 line, which carries a Venus insertion that does not affect cheerio function (Huelsmann et al., 2013). We observed that Venus-tagged Cheerio colocalized at the level of Z-bands with CryAB (Fig. 7C). Moreover, when we overexpressed cher-90-GFP in muscles, it colocalized with CryAB (Fig. 7D), supporting the view that CryAB interacts with Cheerio. Importantly, Cheerio not only interacts with CryAB but also appears to stabilize sarcomeric cytoarchitecture in a manner similar to that of CryAB: knockdown of cheerio leads to the fuzzy actin pattern associated with the aggregate-like accumulations of CryAB (Fig. 7E).

To test functional relationships between CryAB and Cheerio we generated a double UAS-cherRNAi;UAS-CryABRNAi line that allows attenuation of both genes simultaneously and a rescue line by combining the UAS-cherRNAi line with UAS-CryAB. The sarcomeric actin pattern in cheerio CryAB double-knockdown muscles is severely affected, with areas of weak phalloidin staining complemented by an abnormal Z-band-associated accumulation of Kettin (Sallimus – FlyBase) (supplementary material Fig. S7A-A″), a phenotype that has not been observed in a simple CryAB-RNAi context (Fig. 4B). Targeted overexpression of CryAB in muscles with attenuated cheerio partially rescues proper sarcomeric actin organization and fully restores the normal Kettin pattern (supplementary material Fig. S7B-B″). Thus, CryAB and Cheerio not only bind to each other but also interact genetically to stabilize sarcomeric and, potentially, perinuclear actin. Altogether, these experiments show that CryAB-Cheerio interactions are crucial for the maintenance of the contractile apparatus in muscles and suggest that loss of this interaction contributes to the muscle defects and affected muscle performance observed in CryAB-RNAi larvae.

It has been demonstrated that unfolded filamin interacts with Drosophila/mammalian co-chaperone Starvin/BAG3, which plays an important role in the maintenance of Z-disk components (Arndt et al., 2010). We tested whether knocking down starvin in muscles influences CryAB sarcomeric localization. The attenuation of starvin led to Z-band breaks and an irregular sarcomeric actin pattern associated with clustering of myonuclei (supplementary material Fig. S8). However, the CryAB sarcomeric pattern and perinuclear localization appeared unaffected (supplementary material Fig. S8), supporting a view that CryAB-Cheerio and Starvin-Cheerio interact independently. Since the CRYAB-filamin C interaction has not as yet been tested in vertebrate skeletal muscles, whether this interaction is a conserved feature of CryAB proteins remains to be elucidated.

The physiological response to stressful conditions in muscle tissue involves high-molecular-mass Hsps such as Hsp70 (McArdle et al., 2004), together with a set of sHsps (Koh and Escobedo, 2004; Haslbeck et al., 2005). However, besides this classical role, in recent years stress-independent and tissue-specific expression and function have been reported for several members of the Hsp gene family (Michaud and Tanguay, 2003; Rupik et al., 2011).

Here, we provide evidence that in Drosophila, the sHsp subfamily member CryAB [which is also known as l(2)efl (Kurzik-Dumke and Lohmann, 1995)], which encodes the fruit fly ortholog of mammalian CRYAB, is specifically expressed in developing larval somatic muscles and is required for myofibril integrity and muscle performance.

In the developing mammalian heart and skeletal muscles, CRYAB and six other sHsps (HSPB1, HSPB2, HSPB3, HSPB6, HSPB7 and HSPB8) are expressed at a relatively high levels (Davidson et al., 2002). It is not known whether all these genes diverged from a common ancestor with developmental functions, but our phylogenetic analysis indicates that two sHsps involved in muscle development, Hspb8 and CryAB, do have a common ancestor. We also showed that, like its vertebrate counterpart (Doran et al., 2007), the fruit fly CryAB displays muscle- and heart-specific expression and accumulates at the level of the Z-bands and around the nuclei. Interestingly, Drosophila CryAB is also localized in between the Z-bands and muscle cell membranes and displays a punctate expression pattern that suggests that it might be a component of Z-band-associated protein complexes. It is well known that vertebrate CRYAB is also expressed at the level of Z-bands and, by acting as a chaperone of the type III IF protein desmin, helps to maintain cytoskeletal integrity in skeletal and cardiac muscle cells (Golenhofen et al., 1999). Desmin IFs are mainly associated with Z-bands and interact with other sarcomeric proteins to form a continuous cytoskeletal network that maintains the spatial organization of contractile apparatus (Goldfarb and Dalakas, 2009). However, neither desmin nor vimentin (another type III IF protein) orthologs exist in Drosophila (Sparrow and Schöck, 2009), suggesting that other cytoskeletal components of muscle cells replace them functionally to ensure myofibrillar integrity. The subcellular localization of Drosophila CryAB strongly suggests that it interacts with sarcomeric components and, in particular, with Z-band-associated proteins, thereby helping to maintain myofibrillar architecture.

Point mutations in human sHsps lead to several aggregation diseases (Clark and Muchowski, 2000). For example, mutations in αA-crystallin (CRYAA) lead to cataract (Litt et al., 1998), missense mutations in HSP27 (HSPB1) are associated with Charcot-Marie-Tooth disease (Evgrafov et al., 2004), and point mutations in CRYAB cause DRM (Vicart et al., 1998; Sacconi et al., 2012). Yet, the stress-independent functions of individual members of the sHsp gene family have not been systematically assessed in animal models because of functional redundancies and compensation effects. In the case of Cryab, mice knocked out for this gene also lack the adjacent Hspb2 gene and display no obvious developmental defects (Brady et al., 2001). However, as they become older, Cryab Hspb2 homozygous knockout mice show postural defects and other health problems that appear to stem from progressive myopathy (Brady et al., 2001). For Drosophila CryAB, the impact of its loss of function and its lethality have not yet been confirmed.

Here, we analyzed the effects of muscle- and heart-targeted knockdown of Drosophila CryAB to better evaluate its stress-independent functions. We focused on larval musculature and on the adult heart, and found that attenuation of CryAB led to a number of morphological and ultrastructural defects in body wall muscles, impaired muscle performance and accelerated heartbeat. Importantly, muscle alterations lead to local Z-band misalignment, which could be responsible for the observed muscle fiber splitting. Affected myofibrillar integrity results in a markedly altered contractility index in Mef2>CryAB-RNAi larvae, which also display lower motility compared with age-matched wild-type individuals. The sarcomeric alterations are also associated with the clustering/mislocalization of nuclei in muscle fibers and with abnormal swelling of mitochondria, together revealing the important developmental role of CryAB in Drosophila in the maintenance of correct muscle function and structural integrity. Some of the phenotypes resulting from the muscle-specific attenuation of Drosophila CryAB, including disruption of myofibrillar organization and muscle weakness, are reminiscent of those observed in myofibrillar myopathies in human (Goldfarb and Dalakas, 2009).

In human pathological cases, when the IF network does not assemble correctly and creates aggregates, CRYAB can reorganize the malfolded IF proteins into a normal filamentous network (Koyama and Goldman, 1999). Some mutations in CRYAB, such as R120G (Vicart et al., 1998) or D109H, are claimed to affect dimerization of the protein (Sacconi et al., 2012) leading to loss of its chaperone activity and ultimately to pathological muscle defects. Interestingly, protein sequence alignment revealed that both these residues (R120 and D109) are conserved in Drosophila CryAB. When we generated a Drosophila strain carrying a mutation in one of these residues, CryAB^R120G, we found that in larval muscle cells with accumulations of CryAB^R120G the sarcomeric architecture and, in particular, the actin filament pattern were severely affected. We also observed a misalignment of myofibrils and the appearance of protein aggregates containing CryAB^R120G, symptoms also seen in DRM patients (Goldfarb and Dalakas, 2009). The affected myofibrillar integrity and irregular sarcomeric pattern correlated with reduced muscle performance as measured by motility assays in Mef2>CryAB^R120G third instar larvae. Taken together, these observations suggest that the R120G mutation in Drosophila CryAB has an impact on larva body wall muscles similar to that which the analogous mutation in CRYAB has on human skeletal muscles leading to disruption of myofibrillar integrity and muscle weakness. In DRM patients carrying the CRYAB^R120G mutation, muscle defects appear in mid-age adults, whereas in our Drosophila model, following forced expression of CryAB^R120G in muscles, they are apparent in third instar larva. At this late stage of larval development muscles are fully functional and differentiated, like those in adult humans.

Another important phenotype observed in patients carrying the R120G mutation in CRYAB is increased systolic heart diameter, dilated cardiomyopathy and arrhythmia. Here again, targeted cardiac expression of CryAB^R120G resulted in arrhythmic heart beating in the adult fly, reminiscent of that observed in the DRM mouse model and in DRM patients (Wang et al., 2001).

Thus, expressing the R120G-mutated form of CryAB in body wall or in cardiac muscles in Drosophila leads to pathological defects similar to those observed in patients harboring the CRYAB^R120G mutation. This suggests that this mutation, which potentially impairs CryAB dimerization (Sacconi et al., 2012), might have a more general impact on the chaperone function of CryAB and not only on interactions with desmin IFs.

As reported above, muscle-specific expression of Drosophila CryAB^R120G leads to DRM-like phenotypes with Z-band disruption and adversely affected muscle performance. However, as revealed by sequencing of the Drosophila genome (Adams et al., 2000), no gene orthologous to desmin has been identified in the fruit fly, suggesting that CryAB interacts with other sarcomeric components to stabilize the contractile apparatus. To identify potential CryAB-interacting proteins we performed co-IP experiments using dissected third instar larvae. Highly enriched sarcomeric proteins bound by CryAB include actins, Tropomyosin 1 and 2, α-Actinin, Paramyosin, Myosin heavy chain and Cheerio.

Drosophila CryAB accumulates in Z-bands and to a lesser extent in M-lines, and thus can potentially interact with all these sarcomeric components. Identifying actin in the IP material is consistent with the fact that CryAB has an ABD and that the actin pattern is affected in muscle fibers with attenuated CryAB as it is in muscles expressing CryAB carrying a mutation in the ABD. However, the finding that CryAB can potentially interact with Z-band-associated Cheerio was of particular interest, as in humans mutations in filamin C lead to myofibrillar myopathy with dissolution of myofibrils similar to that observed in DRM (Kley et al., 2007; Luan et al., 2010). Filamins are actin-crosslinking proteins consisting of an N-terminal ABD followed by 24 immunoglobulin-like repeats (Stossel et al., 2001) and are involved in a number of cellular processes including cell-matrix adhesion, mechanoprotection and actin remodeling (Feng and Walsh, 2004). In vertebrates, filamin C is muscle specific and localizes at myotendinous junctions and also at Z-bands and costameres (Ohashi et al., 2005; van der Ven et al., 2000a,b). It has been shown that filamin C interacts with two major protein complexes at the sarcolemma, namely the dystrophin-associated glycoprotein complex (DGC) (Thompson et al., 2000) and the integrin complex (Gontier et al., 2005; Loo et al., 1998), both of which are known to have important roles in affording mechanical integrity to striated muscle. However, through its C-terminal region, filamin C also binds the Z-band proteins myotilin (van der Ven et al., 2000b) and myopodin (synaptopodin 2) (Linnemann et al., 2010), suggesting that it ensures a link between Z-bands and sarcolemmal DGC and integrin complexes, thus maintaining the mechanical integrity of muscle cells. This possibility is supported by the detachment of myofibrils from sarcolemma and intercalated Z-bands in muscles of zacro (filamin C) medaka mutants (Fujita et al., 2012). In Drosophila, overall filamin protein structure its actin-crosslinking function are conserved in Cheerio. It plays a key role in ring canal formation during oogenesis (Sokol and Cooley, 1999) and is involved in the positioning of nuclei in ovarian nurse cells (Huelsmann et al., 2013), but the role of Cheerio in muscles had remained elusive.

Here, we show that in third instar larval muscles Cheerio protein accumulates between myofibrils and the muscle cell membrane at the level of Z-bands and around the nuclei, a subcellular localization reminiscent of that of filamin C. We also found that the sarcomeric Cheerio pattern was disrupted in larvae with attenuated CryAB, suggesting that the identified interactions between CryAB and Cheerio might help to maintain myofibrillar integrity. Thus, we hypothesize that CryAB exerts a chaperone function on Cheerio in Drosophila muscles, which are devoid of desmin IFs, and that this interaction helps to stabilize the contractile apparatus.

Recently, it was reported that the conserved BAG3/Starvin complex detects mechanical unfolding of filamins caused by cell stretching (Arndt et al., 2010; Ulbricht et al., 2013). This complex binds unfolded filamins and targets them to autophagosomes for degradation. It is unclear whether Drosophila CryAB is part of the same complex or of another sHsp complex that regulates filamin and muscle cell function. The fact that Drosophila CryAB can interact with cher-90, the Cheerio isoform that does not contain an ABD and thus might not be stretched and mechanically opened up, suggests that the CryAB complex might bind the folded C-terminus of Cheerio and thus not be part of the BAG3/Starvin complex. Future experiments are required to determine whether the CryAB and BAG3/Starvin complexes bound to filamins have overlapping or specific roles during the development and function of skeletal and cardiac muscle cells.

These findings raise the question as to the extent to which Cheerio is able to replace the IF network in Drosophila muscles. Human CRYAB is known to interact with desmin (Goldfarb and Dalakas, 2009), with sarcomeric actin (Bennardini et al., 1992) and with titin (Golenhofen et al., 2002); however, whether the identified interaction between Drosophila CryAB and Cheerio is conserved in human muscles and, if so, whether it helps to stabilize myofibrillar architecture remains to be elucidated.

The following Drosophila stocks were used: w¹¹¹⁸ strain as wild type; Mef2-Gal4 [Bloomington Stock Center (BSC) BL27390]; Hand-Gal4 (kindly provided by L. Perrin, TAGC, Marseille, France); UAS-lacZ (BSC, BL1776); UAS-RNAi-cher [Vienna Drosophila RNAi Center (VDRC) 107451]; UAS-RNAi-starvin (VDRC 42564); two UAS-RNAi-CryAB lines (VDRC 40532 and 107305); cherCPTI0847 protein trap line [Kyoto Drosophila Genomics Resource Center (DGRC)]; and UASp-cher90-GFP (generated by S. Huelsmann, Gurdon Institute, Cambridge, UK). The UASp-Venus-CryAB, UASp-Venus-CryAB ABDmut and UASp-Venus-CryAB R120G lines were generated in the K.J. lab. The generation of CryAB and cheerio transgenic lines is described in the supplementary Materials and Methods.

Primary antibodies: goat anti-GFP (1:500; Abcam, ab5450); rat anti-CryAB (1:500; generated in the K.J. lab); rabbit anti-Mef2 (1:1000; gift from H. Nguyen, Erlangen University, Germany); rat anti-Kettin (1:25; Abcam, ab50585); mouse anti-LamC28.26 [1:1000; Developmental Studies Hybridoma Bank (DSHB)]; rabbit anti-β3-tubulin (1:10,000; gift from R. Renkawitz-Pohl, Marburg University, Germany); mouse anti-β-actin (1:10; Invitrogen, AC-15); guinea pig anti-Msp300 (1:300; gift from T. Volk, Weitzmann Institute, Israel); rat anti-Cheerio (1:200; Sokol and Cooley, 1999); and rabbit anti-active Caspase 3 (1:1000; Abcam, ab13847). We also used different secondary fluorescent antibodies as well as 18 nm colloidal gold-AffiniPure goat anti-rat IgG (whole molecule) (Jackson ImmunoResearch; 1:20) and 10 nm goat anti-rat IgG (whole molecule)-gold (Sigma-Aldrich; 1:100) conjugated to Alexa Fluor 488, CY3 or CY5 (1:300; Jackson ImmunoResearch). Phalloidin-TRITC (1:1000; Sigma) was used for revealing sarcomeric actin.

Analyses of CryAB transcript levels after RNAi-based attenuation are described in the supplementary Materials and Methods.

Tests of larval viability and muscle performance using the righting, motility and journey tests, and characterization of muscle morphology and heart physiology are described in the supplementary Materials and Methods. The heart physiology assay was performed as described previously (Fink et al. 2009).

Third instar larvae were dissected as described previously (Budnik et al., 2006) in ice-cold Ca²⁺-free saline buffer containing 128 mM NaCl, 2 mM KCl, 4 mM MgCl₂, 1 mM EGTA, 35 mM sucrose and 5 mM HEPES pH 7.2 (Demontis and Perrimon, 2009). Buffer without EGTA was used for measurements of contracted muscles. Body wall muscles were fixed with 4% formaldehyde in PBS for 15 min and then rinsed three times for 5 min each in PBS with 0.5% Tween 20 (PBT). Muscles were blocked for 30 min in 20% horse serum in PBT at room temperature. Dissection and staining of adult flies were performed as described previously (Taghli-Lamallem et al., 2008). Primary antibodies were applied overnight at 4°C. After three washes in PBT, secondary antibodies were applied for 2 h at room temperature with or without phalloidin-TRITC as appropriate. MitoTracker Red CMXRos (Invitrogen) was used at 1.5 µM in ice-cold Ca²⁺-free saline buffer; it was incubated for 15 min at 37°C with dissected larvae. Subsequently, larvae were fixed in 4% formaldehyde for 15 min and washed in PBS three times for 10 min each. Muscle and heart preparations were mounted in Fluoromount-G anti-fade reagent (Southern Biotech) and analyzed using SP5 or SP8 (Leica) confocal microscopes. 3D models of labeled muscles were generated using Imaris software (Bitplane).

TEM of dissected larvae and immunogold detection of CryAB in TEM sections are described in the supplementary Materials and Methods.

For the co-IP assay, 300 dissected and frozen Mef2>cher90-GFP (for CryAB-Cheerio) or w¹¹¹⁸ (for CryAB-Msp300) larvae were homogenized at 4°C in 500 µl HEB extraction buffer [containing 50 mM Tris (pH 7.6), 140 mM NaCl, 5 mM EDTA, 1% (V/V) NP40 and 0.5% (W/V) sodium deoxycholate] as described previously (Perng et al., 2006) and supplemented with a protease inhibitor cocktail (Roche). After 30 min incubation on ice, lysate was centrifuged at 17,000 g for 10 min at 4°C. Supernatant was pre-absorbed 1 h at 4°C with 50 µl protein G-Sepharose (Amersham) to eliminate non-specific binding of the proteins to the beads. The protein G-Sepharose beads were separated from the lysate by centrifugation 1 min at 13,000 g at 4°C. The supernatant was incubated 4 h at 4°C with 5 µl rat anti-CryAB antibody or with 5 µl rat non-immune serum as a negative control. The protein G-Sepharose was subsequently added to each lysate/antibody mixture and incubated overnight at 4°C with gentle agitation. The complexes were then washed three times with washing buffer (50 mM Tris-HCl pH 7.5, 500 mM NaCl, 0.1% NP40) and proteins eluted by boiling the beads in 2× SDS-PAGE buffer for 10 min. Eluted proteins were separated on a 4-10% SDS-PAGE gradient gel and blotted to PDF membrane (Whatman), blocked with 5% non-fat milk. The membrane was incubated with anti-GFP antibody (1:5000; for CryAB-Cheerio) or with anti-Msp300 (1:1000; for CryAB-Msp300) followed by anti-mouse or anti-guinea pig HRP-conjugated secondary antibody (1:10,000) at 25°C for 1 h, and subsequently washed three times with TBST (TBS containing 0.5% Tween-20). Protein bands were visualized using the ECL detection kit (Amersham) and analyzed with the Chemi Doc imaging system (Bio-Rad).

Mass spectrometry to identify CryAB-interacting proteins is described in the supplementary Materials and Methods.

Bioinformatics analysis of phylogenic relationships among sHsps is described in the supplementary Materials and Methods.

We thank H. Nguyen, T. Volk, R. Renkawitz-Pohl, B. Suter, L. Cooley, the BSC, VDRC and DGRC for providing reagents and fly stocks; Fly-Facility for generation of CryAB transgenic stocks; and J. Overton for injection of the cheerio construct.