DAAM Is Required for Thin Filament Formation and Sarcomerogenesis during Muscle Development in Drosophila

During muscle development, myosin and actin containing filaments assemble into the highly organized sarcomeric structure critical for muscle function. Although sarcomerogenesis clearly involves the de novo formation of actin filaments, this process remained poorly understood. Here we show that mouse and Drosophila members of the DAAM formin family are sarcomere-associated actin assembly factors enriched at the Z-disc and M-band. Analysis of dDAAM mutants revealed a pivotal role in myofibrillogenesis of larval somatic muscles, indirect flight muscles and the heart. We found that loss of dDAAM function results in multiple defects in sarcomere development including thin and thick filament disorganization, Z-disc and M-band formation, and a near complete absence of the myofibrillar lattice. Collectively, our data suggest that dDAAM is required for the initial assembly of thin filaments, and subsequently it promotes filament elongation by assembling short actin polymers that anneal to the pointed end of the growing filaments, and by antagonizing the capping protein Tropomodulin.


Introduction
Striated muscles contain cylindrical structures, myofibrils, composed of repeating elements called sarcomeres, the basic contractile units of muscle. A sarcomere, defined as the region between neighboring Z-discs, contains two filament systems, the actin-containing thin filaments and the myosin II-containing thick filaments, and their associated proteins. The thin filaments are anchored into the Z-disc where they are cross-linked by dimeric aactinin and a number of other proteins [1]. These filaments extend in both directions from the Z-disc into neighboring sarcomeres. They consist of a filamentous actin (F-actin) core decorated with the regulatory proteins Tropomyosin (TM) and Troponin. Interdigitated with thin filaments are the bipolar thick filaments, composed largely of myosin molecules, that are at the middle of the sarcomere and crosslinked by the M-band proteins. Whereas the structural properties of these macromolecular complexes have been determined in detail in recent decades, much less is known about the in vivo assembly of the filaments and Z-discs to form the very regular sarcomeric structures [2]. In particular, the initial assembly of thin filaments and the regulation of actin dynamics during myofibril formation and maintenance remains poorly understood.
Owing to the regular assembly of actin monomers (G-actin) into F-actin, these filaments display a polarized morphology and dynamics with barbed (+) and pointed (2) ends. In vivo filament growth likely occurs only at the barbed end, whereas the pointed end is favored for depolymerization [3]. New actin filament formation critically requires a nucleation step, during which a few actin monomers combine to form a nucleation seed, prior to elongation. As nucleation is not favored kinetically, and spontaneous in vivo nucleation would lead to anarchic filament assembly, this step is promoted by nucleation factors. Nucleation factors described so far include the Arp 2/3 complex, formins, Spire, Cordon-bleu and Leimodin (Lmod) [4,5]. Although actin nucleation factors have been extensively studied in many different model systems, the essential nucleation factors in developing muscles have not been clearly identified. Lmod and the mammalian formin Fhod3 have both been implicated in actin assembly in vertebrate striated muscles [6,7] but subsequent work concluded that they are unlikely to contribute to actin nucleation during the initial stages of myofibril assembly [8,9,10,11]. In fruit flies, the genome harbors no clear Lmod ortholog, and genetic analysis of the Drosophila Fhod ortholog, Fhos, and other members of the formin family, such as Diaphanous, Cappuccino or Form3, revealed no clear role in muscle development [12,13,14,15].
Regulation of thin filament elongation and length, thought to be controlled by elongation factors and capping proteins, are also important aspects of actin dynamics in muscles. Elongation factors, such as Ena/VASP proteins or the barbed end binding formins that also function as nucleation factors, promote filament growth, whereas capping protein binding blocks polymerization. In contrast to non-muscle cells where thin filament growth is restricted to the barbed ends, sarcomeric actin filaments elongate from their pointed ends [16]. In each half sarcomere the thin filaments are aligned with the same polarity and their barbed ends are within the Z-discs, where they are capped by CapZ, whereas their pointed ends are capped by Tropomodulin (Tmod). So far Tmod, TM, Lmod and the Sarcomere Length Short (SALS) proteins have all been implicated in thin filament length regulation. Of these, Tmod binding causes thin filament shortening; conversely, loss of Tmod function causes lengthening of actin filaments [16,17]. TM enhances Tmod binding affinity, whereas Lmod and SALS seem to antagonize the capping activity of Tmod and promote filament elongation from their pointed ends [9,18]. Surprisingly, instead of promoting elongation, SALS appears to inhibit filament elongation in vitro. These results together with the observation that no protein was yet isolated which would catalyze F-actin assembly at the pointed end, mean that the mechanism which enables muscle thin filaments to elongate from their pointed ends remains mysterious.
Here we show that the Drosophila formin DAAM (Dishevelled associated activator of morphogenesis) plays an important role in sarcomerogenesis. The absence of dDAAM reduces larval motility, causes a flightless phenotype and complex defects in sarcomere organization. The latter include shorter and thinner sarcomeres with reduced thin filament levels and an absence of both Z-disc and M-band organization. Our protein localization studies revealed that, despite being a barbed end binding protein in non-muscle cells, dDAAM is highly enriched near the thin filament pointed ends both in Drosophila and mouse muscle cells.
We propose that members of the DAAM family of formins are very good candidates for the long sought-after muscle actin/thin filament nucleators.
Electron microscopy (EM) of the IFM of dDAAM Ex1 , UDT mutants ( Figure 2) confirmed and extended all the major myofibrillar defects seen in the confocal images. Notably, in longitudinal sections (Figure 2A-D) we revealed irregularly shaped, thin myofibrils with frayed edges, strong Z-disc defects, absence of M-lines and shorter sarcomeres. The thick and thin

Author Summary
Sarcomeres, the smallest contractile units of muscle, are formed by two major filament systems, the myosin containing thick and the actin containing thin filaments. Although it is well established that sarcomerogenesis involves the formation of novel actin filaments, so far it remained largely unclear how these filaments form. In this study, we show that the Drosophila and mouse members of the DAAM formin subfamily are sarcomere associated actin assembly factors. Genetic analysis revealed that dDAAM plays an essential role in thin filament formation and sarcomere organization. In addition, we demonstrate that mDaam1 is an early determinant of myofibrillogenesis. Our data suggest that besides a role at the barbed end of the thin filaments, dDAAM also functions at the pointed end where it antagonizes the capping protein Tropomodulin. Based on these observations, we propose that DAAM family formins are very good candidates for being the long sought-after muscle actin nucleators, that also promote filament elongation by assembling short actin polymers that anneal to the Z-disc anchored growing filament. Given that cardiomyopathies, muscular dystrophies and the cardiovascular disease related heart muscle degenerations belong to the major health problems worldwide, understanding the mechanism of how muscles normally form is of immense biomedical relevance. filament organization was also severely altered. Thick filaments rarely ran parallel to each other, the average thick filament number per sarcomere was strongly reduced compared to controls, and filament packing was much looser than wild type (Figure 2A-D). Most strikingly, instead of thin filaments running in parallel between the myosin filaments ( Figure 2C), this space was filled with many thin filaments that formed a meshwork in some places ( Figure 2D). Although the organization of these filaments was very different from wild type, their dimensions argue that they are disordered thin filaments. Possibly, but we consider it unlikely, they are connecting filaments. Such filaments containing the Sallimus/Kettin proteins link the Z-disc to the thick filaments in insect flight muscle. The transverse sections of dDAAM mutant IFMs confirmed the presence of irregularly shaped myofibrils consisting of thick filament clusters with grayish material in between of them, whereas individual thin filaments could hardly be seen ( Figure 2E-H). Clearly the regular myofibrillar lattice was missing. Unlike wild type thick filaments which appear ringshaped or hollow in transverse sections (except at the level of the M-line) ( Figure 2G) [20], dDAAM mutant thick filaments were very dark, irregular in shape and almost never hollow ( Figure 2H).
At 48 hours after puparium formation (APF) (at 29uC) the pupal IFM of dDAAM Ex1 , UDT flies already showed all the muscle phenotypes observed in adults. These include irregular myofibrillar organization, reduced F-actin levels, lack of visible M-lines and disorganized, unequally spaced Z-discs ( Figure S2A-B0). Accordingly, EM analysis revealed sarcomere shortening (2.360.05 mm, n = 22 in wild type; 1.5360.08 mm, mean6SD, n = 22 in mutants, p,0.001), absence of M-lines, erratic filament packing and strong Z-disc defects ( Figure 2I-J). Together, these data suggest that the IFM phenotypes observed in newly hatched dDAAM mutant adults were likely to be a consequence of loss of dDAAM function during early muscle development.
To test whether the structural alterations observed in dDAAM mutant myofibrils affect their mechanical properties an Atomic Force Microscope (AFM) was used to measure the transverse elasticity of individual myofibrils in rigor conditions ( Figure S2C). The elasticity (Young's modulus) of dDAAM Ex1 and dDAAM Ex1 , UDT mutant myofibrils was significantly lower, 661.63 kPa (n = 35) and 461.24 kPa (n = 15), than that of wild type, 2264.91 kPa (n = 25).
In summary, the genetic impairment of dDAAM function severely affects the structural and mechanical properties of the flight muscles. These results argue that this formin is an important regulator of muscle development affecting multiple aspects of myofibril formation in flies. dDAAM impairs somatic muscle formation and heart development To ask whether dDAAM plays a role generally in muscle development, larval body wall muscles and the heart tube were examined. The body size and somatic musculature of dDAAM Ex68 null mutant early third instar (L3) appeared normal, but late in L3, 100 hours after eggs laying (AEL), the larvae were shorter (2.0860.31 mm; n = 30) than wild type (3.2460.25 mm; n = 30; p,0.001; Figure 3E,I). Although gross alterations were not evident in the overall structure of the musculature, mutant muscles were also smaller, some myofibers were split and their general organization was looser than in wild type ( Figure 3A-D).
Measurements of the ventral longitudinal 3 (VL3) muscle showed a 53% length reduction and 38% reduction in width ( Figure 3K,L) compared to wild type. Shortening of VL3 in dDAAM mutants arises both by sarcomere shortening and a reduction in sarcomere numbers ( Figure 3M,N). The mean sarcomere length of wild type VL3 muscles at 100 hours AEL was 6.261.6 mm (n = 477 sarcomeres; 12 muscles), but was decreased in dDAAM mutants to 3.860.7 mm (n = 241 sarcomeres; 8 muscles; p,0.001). The serial sarcomere number of VL3 was also decreased in dDAAM mutants (30.162.1; n = 8) compared to wild type (39.764.3; n = 12; p,0.001).
To investigate the physiological relevance of the muscular defects observed, we examined the larval motility of dDAAM mutant larvae. Until the early L3 stages there were no differences between the wild type and the dDAAM mutant larvae, possibly due to maternally derived dDAAM (in ,10% of dDAAM Ex68 larvae the dDAAM protein could still be clearly detected at 100 hours AEL, Figure S3E,F). Consistent with the findings of the structural analysis, kinematic studies of linear larval crawling at 72 hours AEL showed that velocities of wild type and mutant larvae did not significantly differ ( Figure 3F). Subsequently at 100 hours AEL their velocity was decreased by ,60% compared to wild type ( Figure 3F,J). Although, we observed a strong correlation between larval body length and crawling velocity ( Figure 3G,H), the dDAAM mutant larvae are much slower than their reduced size would indicate. Rescue experiments with DMef2-Gal4 driven expression of UAS-DAAM constructs confirmed that the observed phenotypes are specific to loss of dDAAM function. Western blot analysis revealed that the IFM expresses two dDAAM protein isoforms, a short (130 kD) minor isoform and a long (163 kD) major isoform ( Figure 1B). These correspond respectively to the predicted DAAM-PB and DAAM-PD proteins (Flybase annotation). The rescue experiments (above) were performed with UAS-DAAM-PB as well as with UAS-DAAM-PD. UAS-DAAM-PB expression partly rescued the velocity decrease and almost fully rescued the body and muscle size of dDAAM Ex68 mutant larvae, whereas UAS-DAAM-PD expression almost completely rescued all the phenotypic traits ( Figure 3I-N). Moreover, muscle-specific expression of these constructs not only rescued the larval muscle defects, but partly rescued the lethality of dDAAM Ex68 to adulthood (3% for PB and 6.1% for PD). Importantly, unlike the wild type constructs, the actin polymerization incompetent mutant forms, UAS-DAAM-PB I732A and UAS-DAAM-PD I1042A mimicking the Bni1 I1431A mutation [21], failed to rescue ( Figure 3I-N). These data demonstrate that the effect of dDAAM on muscle structure and larval motility is muscle autonomous, and that the actinassembling activity of dDAAM is essential for normal muscle development. Additionally, it appears that the two muscle-specific dDAAM isoforms play largely, but not completely, redundant roles in larval muscle.
Muscle-specific expression of UAS-DAAM-PB and UAS-DAAM-PD, in a wild type background, produced significantly longer larvae (PB: 4.2660.15 mm, n = 10, p,0.001; PD: 4.2460.19 mm, n = 10, p,0.001) than wild type. Their VL3 muscles were longer, although in both cases sarcomere size was slightly shorter than wild type ( Figure 3I,K,M). Muscle lengthening occurred by significantly increasing sarcomere number compared to wild type (PB: 5662.8, n = 14, p,0.001; PD: 5462.5, n = 12, p,0.001) ( Figure 3N). Interestingly, the aforementioned structural aspects were almost identical in larvae overexpressing either isoform. Nevertheless, larvae expressing the PB isoform were much faster (,55% faster, n = 10) than wild type larvae ( Figure 3J), while the velocity of larvae expressing PD (,5% faster, n = 10) and the controls ( Figure 3J) were not show that, as compared to the wild type, highly ordered and tightly packed sarcomeres (A, C), the dDAAM mutant myofibrils (B, D) display Zdisc and M-band defects, and shortened sarcomeres with loosely organized thin and thick filaments. Transverse sections of wild type (E, G) muscles reveal the hexagonal lattice organization of thin and thick filaments, which is almost entirely lost in dDAAM mutant myofibrils (F, H). Instead, the mutant fibrils are irregularly shaped, consisting of clusters of thick filaments, and individual thin filaments are hardly detectable. Note: wild type thick filaments are hollow (G), while those of the dDAAM mutant are very dark, irregularly shaped and almost never hollow (H). Longitudinal sections of pupal IFM (48 hours APF) (I, J) show that, as compared to wild type (I), mutants (J) have strong Z-disc and Mline defects, shorter sarcomeres and irregular filament organisation. Arrows mark the Z-discs, asterisks mark the M-bands, m labels the mitochondria. Bars, 500 nm. doi:10.1371/journal.pgen.1004166.g002 significantly different. Lengths of PB and PD overexpressing larvae were indistinguishable but PB larvae had significantly wider VL3 muscles compared to PD larvae. Thus increasing dDAAM isoform levels is sufficient to enhance the number of sarcomeres initiated, but efficient sarcomere elongation may require cooperation of both isoforms and regulation of their ratio.
Larval heart tube size was also reduced in dDAAM mutants compared to wild type (,40% reduction in diameter). In 100 hour old wild type larvae the maximum heart diameter was 100.3367.39 mm; n = 9 whereas in dDAAM mutants 60.446 6.18 mm; n = 9, p,0.001 and they displayed reduced F-actin levels ( Figure S3C,D). Many mutant myofibrils appeared thinner than in wild type and often deviated from the normal orientation ( Figure  S3D). These observations strongly suggest that the formin dDAAM may be a crucial regulator of muscle development in Drosophila with an effect in every muscle type and developmental stage examined.

Sarcomeric localization of dDAAM
To further characterize the role of dDAAM in myofibril formation, we examined its localization pattern in the IFM. In newly eclosed adults the anti-dDAAM serum [22] produced a strong staining in the middle of IFM sarcomeres in the M-line region and a weaker staining was evident at the Z-disc and within the sarcoplasm ( Figure 4C). This pattern persisted from 48 hours APF (the earliest analyzable pupal developmental timepoint) ( Figure 4A-B9) until young adulthood. However, in slightly older adults the signal gradually decreased at the M-line and by 4 days after hatching, equally strong signals were detected at the M-line and Z-disc ( Figure 4D). In a dDAAM Ex1 , UDT mutant, which is nearly protein null for dDAAM ( Figure 1B), only background staining was detectable demonstrating the specificity of the antibody ( Figure S4B). To complement the immunostaining we created a C-terminally GFP tagged dDAAM knock-in allele (dDAAM EGFP ). The dDAAM EGFP allele is viable and fertile in either homo-or hemizygous states, and expression of this protein is entirely under the control of endogenous regulatory sequences. The dDAAM::EGFP fusion protein displayed a roughly equally strong enrichment at the M-line and Z-disc in young and 4 dayold IFMs (Figure S1D-E9). Thus, although the anti-dDAAM serum detects a partial difference between the early and late dDAAM pattern, which is not seen with dDAAM EGFP (presumably due to a difference in the accessibility of the native and the EGFP tagged C-termini), both tools confirm that sarcomeric dDAAM protein is present at both the Z-disc and the M-line.
As thin and thick filaments overlap almost entirely in Drosophila IFM, it was not possible to determine unambiguously whether dDAAM enrichment in the middle of wild type sarcomeres reflects binding to the M-line or to the thin filament ends that extend close to the M-line. In UH3-Gal4/+; UAS-Tmod/+ mutant flies excess Tmod resulted in shorter thin filaments that were not in perfect register and varied in length ( Figure 4G) while M-line organization remained largely normal ( Figure 4H), as judged by F-actin and Obscurin staining, respectively. In such IFMs the dDAAM protein no longer formed a distinct band at the M-line. Instead a punctate intra-sarcomeric staining occurred that mostly co-localized with the pointed end region of the actin filaments ( Figure 4G-G0). This suggests that the mid-sarcomeric dDAAM enrichment, seen in wild type, is likely to be thin filament binding and not an M-line association. Consistent with this conclusion, in developing larval body wall muscles (72 hours AEL) the dDAAM staining clearly resolves into two bands along the M-line ( Figure 4E). Interestingly, in full-grown larval myofibrils the dDAAM staining relocated to a region flanking the Z-disc ( Figure 4F), which is similar to the pattern observed for SALS and Tmod [18]. All together these localization data indicate that dDAAM is present in the growing sarcomeres at a location consistent with a role in thin filament regulation.

Evolutionary conservation
As many muscle proteins are evolutionary highly conserved, and the mouse Daam1 (mDaam1) gene was shown to be involved in heart development [23], we examined the localization of mDaam1 by immunostaining of skeletal muscle sections from 15 day-old animals. Interestingly, in the m. tibialis anterior two bands of sarcomeric enrichment occurred at either side of the M-line, whereas in m. vastus lateralis most protein was detected along the Z-discs ( Figure 5A-B0). To verify this mDaam1 localization pattern further and its development during the early phases of myofibrillogenesis, we used the mouse myogenic cell line C 2 C 12 [24] and a-actinin, known to be one of the earliest marker of myofibril formation [25]. In C 2 C 12 cells that were induced to differentiate for 24 hours, mDaam1 was detected in two broad bands in the sarcomeres between the Z-bodies and the M-line ( Figure 5E). In C 2 C 12 cells differentiated for 48 or 96 hours, the same mDaam1 distribution was detected as after 24 hours of differentiation ( Figure 5F). To resolve the sarcomeric position of the two bands labeled by anti-mDaam1, double staining was carried out with the anti-titin 9D10 and the anti-myomesin B4 antibodies in C 2 C 12 cells differentiated for 96 hours. The 9D10 antibody labels the PEVK region of the giant titin protein located in the I-band close to the I-A border [26,27], whereas B4 labels the M-line [28]. The mDaam1 staining did not significantly overlap with that of either 9D10 or B4 ( Figure 5C-D0) confirming that most of the mDaam1 protein is accumulated between the M-line and the I-A border, corresponding to the thin and thick filament overlap region.
The sarcomeric localization pattern of mDaam1 suggests two important conclusions. Firstly, despite some muscle-specific differences, the subsarcomeric localization of mDaam1 appears similar to that of Drosophila DAAM with regard to accumulation at the Z-disc and alongside the M-line. Secondly, because mDaam1 is recruited to sarcomeric complexes as early as the actin crosslinker a-actinin protein, this formin is likely to be an early determinant of myofibrillogenesis.

dDAAM interacts with thin filament mutants
To collect further evidence for our proposal that dDAAM has an important role in thin filament formation and regulation, genetic interactions with the IFM-specific Act88F KM88 and Tm2 3 mutations [29,30] were tested. IFM structure was analyzed in heterozygous mutants in wild type and dDAAM Ex1 mutant backgrounds. The results revealed that the mild dDAAM Ex1 IFM phenotype ( Figure 6A) is strongly enhanced by Act88F KM88 and enhanced by Tm2 3 . Myofibrils of Act88F KM88 heterozygotes were thinner than wild type and some Z-discs were not entirely straight ( Figure 6B), but the precisely repeating organization of the sarcomeres remained. In contrast, the IFM of dDAAM Ex1 ; Act88F KM88 /+ double mutants exhibited a network of very thin myofibrils often with a branched appearance, in which Z-disc and sarcomeric organization appeared to be completely abolished ( Figure 6C). Similarly, in dDAAM Ex1 ; Tm2 3 /+ mutants the myofibrils appeared disorganized, displaying strongly varying width, unequal sarcomere and thin filament length and the frequent appearance of mini-sarcomeres ( Figure 6E). As controls we examined Act5C null mutants, affecting the major non-muscle cell specific actin isoform [31] and a strong loss-of-function allele of the cytoplasmic Tm1 isoform, Tm1 02299 [32]. As expected, these mutations did not alter the IFM phenotype of dDAAM Ex1 ( Figure  S5). The strong dominant genetic interaction between dDAAM and the IFM-specific Act88F and Tm2 alleles, and the complete lack of interaction with the non-muscle cell specific isoforms, suggests that the major function of dDAAM during muscle development is indeed linked to the regulation of sarcomeric actin filament formation.

dDAAM is required for thin filament elongation
Under in vitro conditions the FH2 or FH1-FH2 domains of dDAAM behave as bona fide formins possessing both actin nucleation and elongation activities [33]. The observation that the thin filaments are shorter in dDAAM mutants than in wild type, suggested that dDAAM is a positive regulator of thin filament elongation. Consistent with the view that muscle thin filaments elongate from their pointed ends, dDAAM is present at the pointed end of actin filaments in the IFM, although, as expected for a formin, it also accumulates at barbed ends. To determine whether dDAAM is functionally important for pointed end elongation we investigated genetic interactions of dDAAM with mutations affecting the pointed end regulator proteins SALS and Tmod. SALS promotes filament elongation in vivo [18], whereas Tmod binding is thought to prevent elongation [16]. The presence of sals f07849 /+ in a dDAAM Ex1 mutant background had no obvious phenotypic effect. In contrast, the tmod 00848 mutation entirely suppressed the weak flightless phenotype of dDAAM Ex1 (4.960.5%, mean6SEM, n = 160, p = 0.027) ( Figure 1A) suggesting that dDAAM and Tmod may act antagonistically during thin filament growth.
To investigate the dDAAM/Tmod interaction in more detail we first examined the IFM-specific RNAi silencing of tmod, and we found that in most myofibrils it severely disrupted myofibrillogenesis ( Figure 7A). However, approximately 10% of the myofibrils had almost normal looking Z-discs allowing us to determine that these sarcomeres were shorter (2.6260.11 mm; n = 26; mean6SD; p,0.001) than wild type. Phalloidin staining revealed the presence of thin filaments in the mid-sarcomeric region ( Figure 7B) and impaired M-lines are evident by EM analysis ( Figure 7H). The strong effect on myofibrillogenesis is in accordance with previous reports that Tmod1 in mouse and Unc-94 (tmd-1) in C. elegans are required for myofibril assembly [34,35,36,37]. The decreased sarcomere length was unexpected as the inhibition of Tmod function increases sarcomere length in cultured cardiomyocytes [38] or in Drosophila primary muscles [18]. We noted however, that although sarcomere length of UH3-Gal4; UAS-tmod RNAi flight muscles was reduced, some of the thin filaments clearly failed to terminate in the H-zone of these mutant sarcomeres ( Figure 7H). Therefore, individual filament length can be longer than in wild type, which would be consistent with the known function of Tmod in filament length regulation. To study whether the tmod RNAi phenotype is sensitive to dDAAM protein level, tmod silencing was carried out in a dDAAM Ex1 mutant background. Most (,80%) myofibrils displayed a striated pattern with distinct M-lines and somewhat aberrant Z-discs, and nearly normal sarcomere length (2.860.13 mm; n = 30; mean6SD; p,0.001) ( Figure 7C). This phenotype suggests that the reduced dDAAM levels suppress the ''over elongation'' of the thin filaments seen in the IFM of tmod RNAi flies, and hence, these results further confirm that these two proteins have antagonistic activities in thin filament elongation.
Although dDAAM protein is detected in the vicinity of the pointed ends of sarcomeric thin filaments, former structural studies indicated that formins are strictly barbed end binding proteins [21,39,40]. This paradox would be resolved if pointed end elongation relies on the formation of short actin filaments that anneal sequentially to growing thin filaments anchored to the Z-disc. In this model, dDAAM would mediate the assembly of short actin filaments by acting as a classical barbed end binding formin, but would additionally either actively promote actin filament annealing, or at least not block it. To test this expectation, an in vitro F-actin annealing assay was carried out with the barbed end binding FH1-FH2 domains of dDAAM. We found that the presence of the FH1-FH2 fragment (100 nM) allowed the end-toend annealing of actin filaments ( Figure 7G), although in previous in vitro assays the FH1-FH2 domains of dDAAM significantly reduced barbed end assembly under similar conditions [33]. Capping protein and TM were used as controls. In accordance with former studies [41,42], the barbed end blocking capping protein had an inhibitory effect, whereas TM enhanced the endto-end annealing of actin filaments, and the combined effect of TM and dDAAM was even slightly higher than the one of TM alone ( Figure 7G). The annealing model suggests that, even if at the pointed end sarcomeric region, dDAAM acts as a barbed end binding protein. Hence it follows that dDAAM is unlikely to directly interfere with the binding of pointed end cappers, such as Tmod. To address this issue, we investigated the effect of dDAAM and Tmod in overexpression assays. The IFM specific overexpression of Tmod resulted in thin filament shortening [43] ( Figure 7D-D0), whereas the excess of dDAAM had no obvious phenotypic effect in the IFM ( Figure 7E-E0). When the two proteins were expressed together, we observed the same phenotypic effect as the overexpression of Tmod alone ( Figure 7F-F0). Therefore these results support the annealing model of dDAAM mediated thin filament elongation and the interaction studies are also consistent with the proposal that dDAAM affects thin filament assembly at pointed ends.

Discussion
The sarcomeric actin filaments are critical structural and functional elements of muscles, yet the mechanism of actin filament formation and its regulation during myofibrillogenesis remained unclear. The initial steps of actin filament formation require nucleation factors, of which Lmod and Fhod3 have been previously identified as muscle-specific nucleators [6,7]. However, functional analysis led to the conclusion that Lmod and Fhod3 are crucial to myofibril maintenance but are unlikely to contribute to filament nucleation during the initial stages of myofibril assembly. Recent work in C. elegans revealed that two members of the formin family, Cyk-1 (a Diaphanous ortholog) and Fhod-1, are both enriched at the Z-disc and promote filament lattice growth and its maintenance in striated muscles [44]. Surprisingly however, the muscle phenotypes displayed by cyk-1 or fhod-1 single mutants were relatively mild and it is unresolved whether other nucleation factors are required in worm muscles. Here we provide in vivo evidence that DAAM, another formin family member, is important for sarcomeric thin filament formation. We have found that dDAAM is required for thin filament elongation and that the actin-assembling activity of dDAAM is indispensable for formation of functional muscles. In addition, we have shown that in differentiating C 2 C 12 cells the mouse Daam1 ortholog is incorporated into sarcomeric complexes at least as early as a-actinin. Thus DAAM family formins are strong candidates for being involved in the initial assembly of myofibrillar actin filaments. Interestingly, although the F-actin content of dDAAM mutant muscles is reduced, some filaments still form. Notably however, the dDAAM mutants available for muscle studies are not protein null. This prevents us from determining whether an additional nucleation factor, such as Dia or Fhos, is involved or that residual dDAAM activity is sufficient to promote some level of F-actin formation. Nevertheless, our results demonstrate that dDAAM is a developmentally important sarcomere-associated actin assembly factor in Drosophila. Remarkably, expression of the vertebrate DAAM orthologs are known to be abundant in developing somites and heart [23,45], and genetic analysis of mDaam1 indicated a role in sarcomere organization in cardiomyocytes [23]. Overall this suggests that the regulation of sarcomeric actin filament formation is an evolutionary conserved DAAM function.
Our studies revealed that in the IFM the dDAAM protein is mostly enriched at either end of the thin filaments, the expected positions for proteins affecting thin filament assembly. We formerly showed that in vitro dDAAM behaves as a bona fide formin, possessing all the major properties reported for other formin family members [33]. Here we propose that at Z-discs dDAAM may regulate G-actin incorporation with the well described barbed end processive capping mechanism of formins. Given that the sarcomeric dDAAM expression in the IFM, including the Z-disc accumulation, is maintained during adulthood, it appears likely that dDAAM also contributes to the maintenance of normal muscle structure and function. Besides the Z-disc enrichment, dDAAM also accumulates at the pointed end region of the thin filaments. Since dDAAM promotes thin filament formation and acts antagonistically to the F-actin pointed end capping protein, Tmod, the simplest interpretation of these data is to assume that dDAAM is involved in filament elongation from the pointed end. This is in good accordance with the evidence that in cardiac myocytes and in Drosophila primary cultures actin dynamics predominate at the pointed ends [17,18], yet the presence at the pointed ends is unexpected for a formin, a barbed end binding protein. Because available structural studies exclude the possibility that a formin directly binds to the pointed end, dDAAM might be recruited to the pointed end by binding to a different protein than actin, or our findings indicate the presence of F-actin barbed ends in the vicinity of the pointed end of the thin filaments. Although we cannot strictly exclude the first possibility, at present the functional importance of such an association is unclear. Therefore we favor the second alternative that has interesting mechanistic implications. If barbed ends indeed exist in the region of the pointed ends, then pointed end elongation could be achieved through the end-to-end annealing of short actin filaments to the Z-disc anchored growing ''mother filament'' (Figure 8). Such a mechanism, demonstrated in vitro, would allow rapid filament elongation at the pointed ends. Our data are compatible with the model in which dDAAM promotes the formation of these short filaments by acting as an F-actin barbed end binding processive capper that also allows filament annealing. An important question is how long these short filaments are? In this regard, it is interesting to note that during contractile ring formation in fission yeast the formin Cdc12p was shown to nucleate short actin filaments that anneal to each other in the presence of TM [42], and consistently, TM increased the annealing process by ,2 fold in our in vitro assay. As TM is a major myofibrillar protein, and the IFM-specific Tm2 mutation dominantly enhanced the thin filament defects of dDAAM Ex1 , we propose that the length of the filaments involved in the annealing process is unlikely to be shorter, but could be equal to an F-actin fragment covered by one TM dimer which is about 37-38 nm or 14 actin monomers. Whereas the ability to anneal end-to-end is an intrinsic property of actin filaments, a better understanding of this mechanism during myofibril formation awaits future studies, most importantly the visualization of the short protofilaments. Nonetheless, it is remarkable that the formin Fhod3, implicated in myofibril maintenance and maturation  [10,46], also displays an accumulation in the pointed end region [7,47] and might regulate actin assembly with a similar mechanism as dDAAM.
Previously presented models of thin filament growth in Drosophila proposed a two-step mechanism [18,43]. According to this view, during the first step short filaments are assembled, and during the second step these filaments extend to their final length. Moreover, it is presumed that, at least in larval muscles, the initial phase is SALS-independent, whereas subsequent elongation from the pointed end requires SALS activity that is thought to antagonize the effect of Tmod [18]. The shorter sarcomeres observed in dDAAM mutant muscles argue that dDAAM is required during the second step of thin filament formation. On the other hand, the severe Z-disc organization defects and the reduced sarcomere number in larval muscles, that are also typical for dDAAM mutants, indicate an earlier function that may be linked directly to the initial steps of thin filament formation. Our mDaam1 protein localization data during C 2 C 12 cell differentiation is also consistent with an early function during sarcomerogenesis, therefore dDAAM is a good candidate for being involved already in the first steps of sarcomeric thin filament formation. Whether the annealing mechanism is at work during the first, second or both steps of actin filament formation/elongation, and whether SALS and dDAAM cooperate or act through independent mechanisms during the second step, remain open questions.
Interestingly, beyond the strong effect on thin filaments, dDAAM also affects thick filament and myofibrillar lattice organization. While these phenotypes can be the indirect consequences of the severe impairment of the sarcomeric thin filament system, another alternative could be that dDAAM plays a more complex role in sarcomerogenesis. In favor of this idea we note that the dDAAM mutants display a more poorly organized filament system than observed in Act88F null mutants which completely lack the sarcomeric thin filaments [48]. Additionally, dDAAM affects the shape of the thick filaments which is not reported for Act88F [48]. Moreover, we found that despite the lack of thin filaments, in Act88F mutants the dDAAM protein remains associated with the muscle fibers displaying a non-uniform distribution with foci that largely overlap with those of Myosin staining ( Figure S4C). Taking all these observations together with the unusually strong effect on lattice organization, we speculate that, besides actin binding, dDAAM might play an important role in the integration of the thin and thick filament systems during sarcomerogenesis. Remarkably, unlike the actin isoforms [49], overexpression of the wild type dDAAM protein in larval muscles significantly increased sarcomere number and muscle size while sarcomere length remained nearly normal. Therefore dDAAM appears to play an instructive role in sarcomere formation, and to our knowledge, this is the first example reported where overexpression of a single muscle protein results in such an effect on muscle development.
The UAS-dDAAM RNAi-T129M dDAAM specific RNAi line, targeting nucleotides 2562-3068 of the RE67944 dDAAM cDNA clone, was created by standard cloning and transformation techniques. To create a UAS-DAAM-PD clone, the PD isoform specific region was amplified from a cDNA pool generated by reverse transcription of mRNAs isolated from the L3 stage. We first created a pENTR3C-DAAM-PD clone that subsequently was used to create pTW-DAAM-PD (UAS-DAAM-PD) destination clones suitable for transgenesis.
The dDAAM EGFP knock-in mutant was created by a two-step Pelement mediated gene conversion experiment. First a targeting construct was assembled in a modified pBS vector where we inserted a 1.3 kb 39 dDAAM genomic region until the last codon, this was followed by a 2.3 kb Gal4::VP16 fragment flanked with I-SceI cut sites on both sides, next we inserted a 1150 bp fusion fragment containing the 39 dDAAM region encoding the last 83 Cterminal aminoacids fused to an EGFP coding sequence ending with a stop codon. This was followed with the entire 39 UTR of dDAAM and a 1.1 kb genomic region further downstream of it. This way, besides the genomic flanking sequences, the construct carries Gal4::VP16 that can be used as a marker gene which is flanked both by I-SceI sites and a ,250 bp long genomic duplication encoding the most C-terminal dDAAM coding sequences. This targeting construct was converted into the dDAAM genomic region after remobilizing the EP(1)1542 P-element insertion located 200 bp downstream of dDAAM (see Flybase for details). To this end, EP(1)1542 virgins were crossed to ry 502 Fab-7 1 D2-3 (gift from L. Sipos, BRC HAS, Szeged) males and the embryonic progeny of this cross was injected with the targeting construct. Offspring of the previous cross was crossed to w; UAS-EGFP flies en masse and put on egg laying medium. Embryos were collected on apple-juice plates, and the hatching larvae were screened for GFP fluorescence with an MZ FLIII stereo microscope (Leica, Switzerland). Larvae with GFP expression in the tracheal and nervous system were collected individually and  (2) is the end-to-end annealing of DAAM assembled short actin filaments (in orange) to the Z-disc anchored growing ''mother filament'' (in brown). doi:10.1371/journal.pgen.1004166.g007 used to set up stocks. Conversion events were confirmed by PCR and sequencing. Once the Gal4::VP16 containing construct has been successfully converted into the dDAAM gene, in a subsequent round of crosses I-SceI was used to induce DNA double strand breaks that could eventually be repaired through homologous recombination between the ,250 bp duplicated dDAAM regions. This event, confirmed by PCR and sequencing, led to the removal of all non Drosophila sequences with the exception of EGFP, and resulted in a C-terminally GFP-tagged dDAAM allele. dDAAM EGFP is fully viable and fertile in hemi-or homozygous state indicating that the presence of EGFP does not significantly alter dDAAM function.

Immunohistochemistry
Newly eclosed adult IFMs were dissected from bisected half thoraces in 4% paraformaldehyde (PF), incubated for 15 minutes, then washed with relaxing solution (6 mM MgCl 2 , 5 mM EGTA, 5 mM ATP, 90 mM potassium propionate, 20 mM NaPi, pH 7.0). The muscles were permeabilized overnight in Triton-X/glycerol solution (50% v/v glycerol, 0.5% Triton X-100, 20 mM NaPi, 2 mM MgCl 2 , 1 mM EGTA, 5 mM DTT, pH 7.0) at 4uC, and washed in PBS supplemented with 0.5% Triton X-100 (PBT), then labeled with primary and secondary antibodies. For pupal IFM preparations, timed pupae were removed from their puparia and pinned by the head on dry Sylgard (Dow Corning), dorsal side down, and then submerged into 4% PF in PBS. After dissection along the ventral midline, unattached material was flushed gently away using a syringe to expose the IFMs. These were detached and incubated in fixative (4% PF in PBS) for a further 15 minutes, then transferred back to relaxing solution. For larval heart tube dissections larvae were cut along the ventral midline in relaxing solution. Then fixed with 4% PF in PBS [53]. Fat bodies and other organs were removed, then labeled with primary and secondary antibodies. For developmental staging, white pre-pupae with everted spiracles were removed into fresh vials at 25uC and harvested at required time-points. Adult flies were selected as 'newly eclosed' between 0 and 8 hr post-eclosion.

Larval length measurements and crawling assays
Aged larvae were collected and rinsed with tap water, and then gently placed onto agar plates. The plates were placed under an Olympus SZX12 dissecting microscope equipped with an Olympus C7070 digital camera. Total illumination was applied, images were acquired at 25 Hz. The recording environment (temperature, humidity, illumination) was stationary. Twenty seconds of movie was recorded with DScaler (The DScaler Project Team) for each larva. During this period of time most wild-type larvae moved out from the field of view. ImageJ (NIH) software was used to analyze the image sequences. Persistent forward movements were selected to characterize larval crawling velocity and larval length. Larval length was calculated as the average of the minimum and the maximum head to tail distance for each larva. Maximum intensity projections were used to generate larval tracks. To calculate larval crawling velocity, the lengths of these tracks were divided by the time. Tracked larvae were dissected, stained and subjected to muscle measurements made on VL3 (ventral longitudinal 3) muscles. Muscle length was measured manually as the major axis of the VL3 muscles; muscle width was measured as the minor axis of the VL3 muscles. Sarcomere number and sarcomere length were measured from gray scale intensity plots across phalloidin stained sarcomeres, sarcomere size being the distance between adjacent peaks.

Flight tests
Flight tests were carried out with three day old flies [54]. Flies were released inside a perspex box illuminated from above, and scored for the ability to fly up, horizontally or down. Flies falling into the third category (down) were counted as flightless.

Tissue culture and mouse muscles
The mouse myogenic cell line, C 2 C 12 (ATCC), was maintained in growth medium (DMEM supplemented with 10% FBS; GIBCO/Life Technologies). Cells were initially plated into 100mm-diameter dishes (Greiner) at a density of 10 4 /cm 2 . When cultures reached ,80% confluence they were subcultured onto sterile glass coverslips in 35-mm-diameter dishes. Cultures were kept in growth medium until they reached 60% confluence and subsequently were switched to differentiation medium (DMEM containing 2% horse serum; GIBCO/Life Technologies). This medium was replaced every day, and samples were processed for immunostaining at selected time points. Cells were fixed in 4% formaldehyde in PBS for 10 minutes, and permeabilized in PBS+ 0.1% Triton-X100 for 3 minutes before staining. Primary antibodies were applied for 1 hr RT, and after 365 minutes washing in PBS, cells were incubated with secondary antibodies for another 1 hr. After washing three times for 5 minutes in PBS, samples were mounted in PBS:glycerin (1:4).
For sections of m. tibialis anterior and m. vastus lateralis, C57Bl/6 adult male mice were sacrificed by cervical dislocation. Leg muscle was dissected, embedded in Tissue-Tek O.C.T. compound (Sakura Finetek) and snap-frozen in isopentane cooled by liquid nitrogen. 10 mm cryosections were fixed in prechilled acetone and kept at 280uC.

AFM and force measurements
IFM muscle fibers, falling apart for individual myofibrils upon preparation, were mounted on a poly-L-lysine coated glass surface and measured in phosphate buffered saline. To identify the target points at which to perform individual force measurements, a rough and low resolution scan was taken (not shown). Experiments were carried out with Asylum MFP-3D head and controller (Asylum Research, Santa Barbara, CA). The driver program was written in IGOR Pro software (version 5.04, Wavemetrics, Lake Oswego, OR). Rectangular, gold coated, silicon nitride cantilevers were used, with a nominal spring constant of 30 pN/nm and a V shaped tip with radius of curvature of roughly 30 nm (Bio-Lever, BL-RC150 VB-C1, Olympus Optical Co. Ltd). The measurements were performed in contact mode in liquid, with the vertical piezo working in a closed loop. Constant speed of 0.6 mm/s (scan rate 0.1 Hz) and total load was kept below 1 nN during experiments.
Simple force curves were measured by lowering the probe until a desired deflection is reached and pulling it back. To calculate the sample's elasticity the contact region of the lowering part from force curves has been used. By subtracting a reference curve, recorded on a hard surface from those measured on the object of interest, the sample's force vs. indentation curve can be obtained, which provides the Young's modulus of the measured sample [55]. Several points were examined recording multiple force curves at each selected place; the average and standard deviation of which was calculated.

Protein gel electrophoresis and western blot analysis
Adult IFM samples were dissected as described above. Tissues were immediately placed in ice-cold RIPA lysis buffer and kept overnight. SDS-PAGE and Western blot analyses were carried out according to standard protocols. Membranes were stained with rabbit anti-dDaam (1:1000), and rabbit anti-glycogen phosphorylase (1:20000) (gift from A. Udvardy, BRC HAS, Szeged) used as a loading control. Secondary antibody was a-rabbit-HRPO (1:10000; Sigma). For chemiluminescent detection we used a Millipore Immobilon kit.

Actin filament annealing tests
To measure the annealing of actin filaments fluorescence microscopy assays were performed. Actin filaments (10 mM, Factin) were polymerized for 2 hr at room temperature in 4 mM Tris-HCl (pH 7.0), 0.1 mM CaCl2, 0.2 mM ATP, 0.5 mM DTT, 1 mM EGTA, 1 mM MgCl2 and 50 mM KCl (F-buffer). The Factin solution was then diluted to 1 mM using F-buffer in the absence or presence of actin-binding proteins (100 nM capping protein or 100 nM dDAAM FH1-FH2 or 1 mM skeletal muscle TM or 100 nM dDAAM FH1-FH2 and 1 mM skeletal muscle TM). The samples were incubated overnight. For investigation of the annealing, Alexa-488-phalloidin labeled samples were sheared five times through a 26 gauge needle. Samples were diluted 100 fold into microscopy buffer (F-buffer supplemented with 50 mMDTT, 5 mM DABCO and 0.5% (w/v) methylcellulose) 0 and 60 minutes after shearing and processed for microscopy observations. Single actin filaments were observed with an Olympus IX81 inverted fluorescence microscope using a 1006 objective (NA1.4) and a CCD camera (Orca ERG Hamamatsu). The length of the actin filaments was measured and analyzed with ImageJ. Under each condition 3-4 independent measurements were performed and 300-600 filaments were analyzed. Statistical analysis was carried out using Microsoft Excel or Microcal Origin 6.0.

Statistics
Excel (Microsoft) was used to collect and organise data. Statistical analysis was carried out using Prism 5 (GraphPad Software Inc.) and/or SigmaPlot 12 (Systat Software Inc.). Normality of the data was verified by Shapiro-Wilk test. Pairwise comparisons were made using the Student's t test or the Mann-Whitney U test according to the normality, p,0.05 was considered as statistically significant. In these sagittal sections of thoraces rhodamine-phalloidin was used to visualize the muscle F-actin. Note that mutant dorsolongitudinal muscle (DLM) fibers are shorter (arrows) and thinner than in wild type, and some of the muscles appear degenerated. (C-D) Phalloidin staining of a wild type (C) and dDAAM Ex68 mutant (D) larval heart tube to visualize F-actin (in green). Compared to wild type, the dDAAM Ex68 mutant has reduced F-actin levels, and heart tube diameter is smaller. In addition, many of the mutant myofibrils appear thinner than their wild type counterparts and often deviate from the typical wild type orientation. (E-F) A developing wild type (E) larval body wall muscle at 72 hours AEL clearly expresses and accumulates the dDAAM protein (in cyan) in its myofibrils. A similar but weaker dDAAM expression pattern can still be detected in ,10% of dDAAM Ex68 mutant larvae (F) even at 100 hours AEL. Kettin (in green) labels the Z-discs in E-F. Bars: 100 mm (A-B); 40 mm (C-D); 5 mm (E-F). (TIF) Figure S4 Sarcomeric localization of the dDAAM protein in wild type and mutant IFMs. Myofibrils of wild type (A-A0), dDAAM Ex1 , UDT (B-B0) and Act88F null mutant (C-C0) IFM from young adults stained for dDAAM (cyan, A-C0), actin (red, A9-B0) and Myosin (green, C9 and C0). Staining of wild type IFM reveals dDAAM accumulation at M-line and Z-disc, and in the sarcoplasm (A-A0). In contrast, in a dDAAM Ex1 , UDT mutant IFM only a weak background staining is evident (compare A to B). In Act88F null mutants, which completely lack sarcomeric thin filaments, dDAAM protein remains associated with muscle fibers and displays a partial colocalization with myosin (C-C0). Bars, 5 mm. (TIF) Figure S5 dDAAM shows no interaction with the non-muscle cell specific isoforms of actin and tropomyosin. Adult IFM myofibrils showing dDAAM Ex1 (A), Act5C G0025 /+ (B), dDAAM Ex1 ; Act5C G0025 /+ (C), Tm1 02299 /+ (D) and dDAAM Ex1 ; Tm1 02299 /+ (E) mutants stained for Kettin (green) and actin (red). Note that all mutant myofibrils look nearly wild type. Bars, 5 mm. (TIF)