Myosin isoform switching during assembly of the Drosophila flight muscle thick filament lattice

Each muscle sarcomere is a macromolecular complex containing a large number of structural proteins that are essential for its function (reviewed in Clark et al., 2002). Even though there are differences in sarcomeric structure between organisms and muscle types, the basic architecture of striated muscle sarcomeres is universal. A fundamental question in myogenesis is how the multitude of proteins is assembled to produce the very regular final sarcomeric structure and different models have been proposed to explain it (reviewed in Sanger et al., 2005). The most widely accepted of these is the premyofibril model (Rhee et al., 1994) which postulates a progressive process of maturation of sarcomeres within a myofibril from early, less complex structures called premyofibrils, by the sequential addition of sarcomeric proteins. Other models propose that the various sarcomeric sub-parts, such as early Z-disc and thin filament complexes called I-Z-I brushes, and thick filament A-band precursors, assemble individually and separately before they come together to form the final structure (Fürst et al., 1989; Schultheiss et al., 1990; Lu et al., 1992; Holtzer et al., 1997).

The Drosophila indirect flight muscles (IFM) are an established model genetic system for the study of muscle development (Fernandes et al., 1991; Vigoreaux, 2001). Many of the IFM sarcomeric proteins are present as IFM-specific isoforms, which often permit their genetic knockdown without compromising adult viability. The ultrastructural details of insect IFM sarcomere assembly have been described in detail in two dipteran species, Calliphora erythrocephala (Auber, 1969) and Drosophila melanogaster (Reedy and Beall, 1993). In both species sarcomere assembly begins with the establishment of nascent myofibrils, which mature synchronously throughout the fibre, in Drosophila without the later addition of extra sarcomeres (Reedy and Beall, 1993). This means that any given developmental time-point represents a stage in assembly for all the sarcomeres in the fibre. In Drosophila IFM sarcomere assembly takes around 3 days, permitting ready sampling at distinct time-points for the detailed study of gene expression and protein localisation.

Sarcomeric myosin is a hexamer, consisting of two myosin heavy chains, two essential light chains and two regulatory light chains. The N-terminal region of the heavy chain polypeptide forms the myosin head with its ATP-dependent motor domain, whereas the C-terminal region forms the rod. Extensive α-helical domains of the latter region self-dimerise to form coiled-coil rods, which further associate with each-other to form thick filaments. Most, if not all, animals express multiple sarcomeric myosin heavy chain isoforms to produce muscles with different physiological properties (Schiaffino and Reggiani, 1996; Weiss and Leinwand, 1996; Reggiani et al., 2000). In Drosophila, the different isoforms arise from the single myosin heavy chain (Mhc) gene (Bernstein et al., 1983) by alternative transcript splicing of multiple alternative exons at five positions within the final mRNA, and the inclusion or exclusion of exon 18 (George et al., 1989).

Myosin assembly into thick filaments involves three processes. First, myosin heavy chain monomers dimerise via their α-helical coiled regions (Cohen and Parry, 1998); second, fully formed dimeric myosin molecules associate in an anti-parallel fashion, through a sequence of charged amino-acids called the assembly competent domain, to form a thick filament nucleating centre (Sohn et al., 1997). Finally, further myosin molecules are added to the ends of the nucleating centre, elongating the nascent thick filament. In the sarcomeric context, it remains unclear how and where the process of assembling individual thick filaments, then organising them into the superstructure of the A-band takes place and how this is regulated in vivo. A-band substructures may form alone and become incorporated into the sarcomere later, or all three sequential processes of myosin assembly into thick filaments might take place within the developing sarcomere. The different sarcomeric assembly models proposed so far (reviewed in Sanger et al., 2005) allow for any of these possibilities.

In our search for genetic tools to investigate A-band assembly, we obtained a transgenic line expressing a GFP-tagged version of the Mhc gene, called weeP26. This line was generated in a ‘protein trap’ screen, in which a small P-element (weeP) containing a GFP exon was mobilised to obtain random insertions into the genome (Clyne et al., 2003). Early in our characterization of this transgenic line, we observed unusual differences in the pattern of myosin–GFP localisation in the sarcomeric A-bands of the IFMs compared to other muscles in the fly. We report here that these differences reveal the presence of a second myosin heavy chain (MHC) isoform in addition to the major IFM-specific myosin isoform previously described (Hastings and Emerson, 1991) during IFM sarcomere development. The differential expression and localisation of the two MHC isoforms shows that IFM A-bands assemble progressively, by the addition (and subsequent elongation) of new thick filaments around an initially established thick filament core. Furthermore, we show that the IFM-specific protein flightin assists in regulating this assembly process by reducing myosin exchange/turnover once the latter has been assembled into thick filaments.

The weeP26 transgenic line contains a transposon encoding a GFP exon inserted in the Mhc gene between exons 18 and 19 (Clyne et al., 2003), the last two exons of the gene; we have confirmed this independently by sequencing of weeP26 genomic DNA. Alternative splicing of Mhc transcripts produces variants that either include or exclude exon 18, with translation terminating either in exon 18 or 19, respectively, as both exons contain a stop codon (Fig. 1A). The large number of MHC isoforms can therefore be classified as either MHC-18 or MHC-19. Due to the location of the GFP exon, in the weeP26 line only the MHC-19 isoforms will include the GFP, as in MHC-18 isoforms translation termination happens upstream of the GFP exon (Fig. 1A). It was expected therefore that all embryonic/larval muscle isoforms, which are MHC-19, (Zhang and Bernstein, 2001), would be GFP-tagged in weeP26 muscles; the IFMs, which specifically express an exon-18-containing isoform (Hastings and Emerson, 1991) were not expected to have any GFP-tagged MHC. Confocal microscopy of weeP26 muscles showed that the GFP-tagged myosin (MHC-GFP) localises in the A-band region of the embryonic larval wall muscles (Fig. 1B) and the adult jump muscle (Fig. 1C), known as the tergal depressor of trochanter (TDT) or tergotrochanter (TTM) muscle. The unexpected appearance of MHC-GFP in IFM sarcomeres (Fig. 1D) also showed an unusual sarcomeric localisation pattern. While the GFP signal was found throughout the A-band region in TDT and larval muscle, in the IFM it was limited to the core region of the A-band. This was shown better by staining for the IFM-specific thick filament protein flightin, which was clearly localised in a much wider area than MHC-GFP (Fig. 1E). Fig. 1F is a schematic of the localisation of MHC-GFP in the core of the IFM sarcomere, in two foci flanking the M-line. Overall this result showed that there are two different populations of myosin molecules in the weeP26 IFMs, bearing or lacking the GFP tag, with the tagged molecules localised in the sarcomeric core. Since only MHC-19 variants are expected to be GFP- tagged in weeP26 flies, this suggests that an MHC-19 variant is present in the IFMs, but with a restricted sarcomeric location.

The GFP tag in weeP26 IFM could affect MHC properties and thick filament assembly. However, the normal Z-disk appearance in the confocal images above (Fig. 1A) suggests that sarcomere assembly in weeP26 homozygotes is unaffected. Electron micrographs of homozygous adult weeP26 IFMs show that the ultrastructure of the sarcomeres is indistinguishable from those of wild type (Fig. 2) including sarcomere length at 3.363± 0.136 µm (n = 30), confirming that myosin-containing thick filaments occur in regions outside the MHC-GFP signal, and that the GFP tag does not affect sarcomere assembly. Finally, flight-testing to assess IFM function showed that 3-day-old weeP26 homozygotes fly as well as the wild type [Student's t-test, t(18) = 1.364, P<0.001]. Therefore the GFP tag in weeP26 IFMs has no negative effects on sarcomere assembly or function.

The restricted pattern of MHC-GFP localisation in the adult IFMs could be explained if the protein was expressed and assembled into the early developing sarcomere, and not later, on the assumption that IFM A-bands grow by the addition of thick filaments around an early thick filament core. To study this, the localisation of MHC-GFP was followed in developing weeP26 IFM sarcomeres. WeeP26 IFMs were dissected out of early (48 hrs after puparium formation, APF) and mid-development (72 hrs APF) pupae, and immunostained for obscurin, a protein that in Drosophila IFM localises to the M-line from the earliest stages of sarcomere assembly (Katzemich et al., 2012) (Fig. 3). At 48 hrs APF, MHC-GFP is present throughout the early, narrow A-band core (including the region of the developing M-line). At 72 hrs APF, despite the growth of the sarcomere in length (along the myofibrillar axis) but also in width (orthogonal to the myofibrillar axis), as shown by the lateral extension of the obscurin-stained M-line, the MHC-GFP pattern remains largely unchanged. It is confined to the core, and mostly absent from the rest of the growing A-band. An apparent reduction in the signal intensity at the centres of the fluorescence corresponds to the M-line region. The GFP pattern in the adult IFM sarcomere is almost identical, except that the M-line fluorescence gap is slightly wider. These observations suggest that in weeP26 IFM the early A-band is assembled using the GFP-tagged MHC, which is expected to be an MHC-19 isoform (henceforth designated MHC-IFM19). Subsequently, thick filaments increase in length outwards from the core using the non-tagged MHC, the previously reported IFM MHC-18 isoform (henceforth designated MHC-IFM18), leaving MHC-IFM19 in the core. This not only confirms that the IFM A-band grows from the core outwards by the addition of thick filaments laterally, but also that the thick filaments grow at their tips, otherwise the GFP foci would progressively move away from each other.

To eliminate the possibility that the change in MHC isoform in weeP26 during development is due to the presence of the GFP-exon, Mhc expression during wild-type IFM development was characterised. mRNA was isolated from developing wild-type IFM at 36, 48 and 72 hrs APF. RT-PCR was performed using primers on exons 17 and 19, flanking exon 18, to test for the inclusion or exclusion of exon 18, thereby differentiating between Mhc-IFM18 and Mhc-IFM19 transcripts (Fig. 4A). At 36 hrs APF, only expression of Mhc-IFM19 was detected (Fig. 4B–C). At 48 hrs expression of Mhc-IFM19 was reduced, while Mhc-IFM18 was first detected. At 72 hrs APF, Mhc-IFM19 signal was no longer detectable, but Mhc-IFM18 was strongly expressed. The newly eclosed adult IFMs still strongly expressed the Mhc-IFM18 mRNA. The Mhc-IFM19 transcript was also detectable at very low levels in the adult IFMs, though this could also be due to a small contamination from other thoracic muscles. The results were confirmed by multiple mRNA isolation replicates, and PCR using different conditions (not shown). There is therefore clear evidence from fluorescence microscopy and expression data for a transition from expression of MHC-IFM19 to that of MHC-IFM18, around 48 hrs APF.

Adult IFM is reported to express only one MHC-IFM18 isoform, the alternative exon profile/composition of which is already known (Hastings and Emerson, 1991). In order to examine the exon profile of the earlier MHC-IFM19, the alternative Mhc exons (numbers 3, 7, 9, 11 and 15) were amplified from IFM cDNA isolated at 36 hrs APF and sequenced. According to the RT-PCR results above this should only contain the Mhc-IFM19 mRNA. The amplification was performed using primers on the constitutive exons, flanking the alternative exons, in order to detect all possible isoforms. For each reaction, sequencing produced a single clean sequence, showing that only one specific Mhc-IFM19 isoform is expressed at 36 hrs APF. The sequence was compared to the known sequence of the alternative exons. The comparison showed that the translation of Mhc-IFM19 uses exons 3b, 7d, 9a, 11e and 15a, the same profile as the adult Mhc-IFM18 isoform. Therefore the only difference between the early Mhc-IFM19 and late Mhc-IFM18 is in the choice to include, or not, exon 18, which determines the C-terminal sequence.

All Drosophila Mhc isoforms are of virtually identical size. As all MHC-19 isoforms in weeP26 are expected to be GFP-tagged, there is sufficient size difference to differentiate between the early MHC-IFM19-GFP and the untagged adult MHC-IFM18. The accumulation of the two isoforms at different developmental stages of homozygous weeP26 IFM was followed by western blot (Fig. 5). However, even at 48 hrs APF, when MHC-IFM19 is solely expressed, only a portion of the total MHC migrated to the GFP-tagged band size on the gel, with the untagged MHC being the major band. A western blot of larval muscles, which are known to only express MHC-19 isoforms (Zhang and Bernstein, 2001), similarly showed that only a portion of MHC was tagged. This can be explained by skipping of the GFP exon during transcript splicing, which has been shown before for other weeP lines (Clyne et al., 2003), or by proteolytic degradation of the GFP-tagged C-terminal region of MHC-IFM19-GFP.

The IFM-specific MHC-associated protein flightin has been implicated in A-band stability, as in its absence, thick filament lengths are misregulated, and sarcomeres break down as a result of contraction forces (Reedy et al., 2000). To examine whether MHC-IFM19-GFP in weeP26 IFMs might be differently localised in the absence of flightin, we crossed weeP26 to a flightin null mutant stock (fln⁰). For these experiments, we isolated adult IFMs immediately after eclosion, as flightin-null sarcomeres are known to become damaged following contractions in the newly emerged young adult. In flies with only one copy of flightin, the MHC-IFM19-GFP pattern was very mildly affected. However, in the flies lacking all flightin, MHC-IFM19-GFP was relocalised to the entirety of the A-band (Fig. 6A). As MHC-IFM19-GFP is only expressed during the early stages when the A-band core is formed, this localisation suggests that in the absence of flightin the protein diffuses freely within the A-band during development, rather than remain confined to the foci in the sarcomeric core. This suggests that flightin restricts MHC monomer exchange/diffusion in the A-band.

Another possible explanation for the distribution of the GFP signal in the weep26; fln⁰ sarcomere could be a cleavage of the MHC-IFM19-GFP and re-binding of the GFP moiety throughout the A-band. To ensure this was not the case, we probed western blots of 72 hr AFP IFM with anti-GFP antibody and detected no free GFP (not shown). Additionally expression of GFP in IFM of Dmef2-GAL4 >UAS-eGFP flies showed no association of the GFP with the sarcomeres. Furthermore, we explored the possibility that the fln⁰ mutation affects the Mhc isoform switch using PCR, as a very extended expression of the Mhc-IFM19-GFP mRNA could also give rise to the observed pattern, irrespective of the function of flightin. The data (supplementary material Fig. S1) show that the change in isoform expression in fln⁰ IFM is indeed delayed, but only by 24 hrs to around 72 hrs APF. However, this shift does not substantially affect our conclusions. Our unpublished work (in agreement with the work of Reedy and Beall, 1993) shows that the 72 hr timepoint is approximately half-way through sarcomere assembly. The 24 hr delay in expression could cause the MHC-IFM19-GFP localisation at the sarcomeric core to be wider, but not to extend throughout the A-band as we have described here.

In order for flightin to restrict the MHC-IFM19-GFP distribution to the sarcomeric core it would have to be expressed at least as late as the start of sarcomere widening, after 48–56 hrs APF. Flightin is known to begin accumulating half-way through pupation (Vigoreaux et al., 1993). However, the precise time of its first localisation in the developing A-band has not been previously reported. We therefore studied flightin localisation in the developing IFM in an Sls-GFP genetic background, which expresses a GFP-tagged version of the Z-disk protein sallimus as a Z-disk marker. Flightin could be detected as early as 48 hrs APF, localising at foci flanking the Z-disk (Fig. 6B). Since the IFM I-band is relatively narrow, and flightin is a myosin binding protein, this localisation would correspond to the edge of the nascent A-band, at the region of the putative A-I junction. This is confirmed in weeP26 IFMs of the same age, where flightin can be clearly seen localising at the edge of the nascent A-band (Fig. 6C). During intermediate developmental stages (56 and 72 hrs APF; Fig. 6B) flightin was also found in the A-I junction of the growing sarcomere, and since A-band thick filaments grow from their tips (see earlier), this argues that the filaments are being ‘decorated’ by flightin as they grow. However, the pattern had also expanded towards the M-line, with the latter expansion being more advanced at the sarcomeric core, than at the periphery. This shows that flightin, which by the end of sarcomere development is found in almost the entire A-band, for any given thick filament binds MHC in a progressive and possibly regulated manner, beginning at the A-I junction, and from there extending towards the M-line. It however appears to be excluded from the M-line itself, even in the fully developed adult samples.

To determine whether flightin binds both MHC-18 and MHC-19 isoforms indiscriminately, IFMs from a stock expressing only embryonic myosin (Wells et al., 1996), an MHC-19 isoform, were stained for flightin (Fig. 6D). Despite the absence of MHC-IFM18 in these muscles, flightin was found localised throughout the embryonic myosin A-bands. Flightin binding is therefore not restricted to MHC-IFM18, but can occur with the embryonic MHC isoform, which differs at all alternative Mhc exons from both adult IFM isoforms.

We have argued that flightin diminishes myosin motility within the A-band. Such an effect should be demonstrable in weeP26 IFM, in the relative localisation of flightin and MHC19-IFM-GFP. In the time following the switch from the fluorescent to the non-fluorescent myosin isoform, the portions of the A-band in which flightin is already present should retain their fluorescence. Conversely, in the areas lacking flightin, the signal should progressively weaken by turnover/exchange with the non-fluorescent myosin. For optimal comparison, average images of the two localisation patterns were generated, and intensity plots were drawn (Fig. 7). In the early stages (48 hrs APF) MHC-IFM19-GFP fluorescence in the putative A-band is most intense in the M-line region. This can be explained by new myosin molecules being added laterally there to form new thick filament nuclei. At that time-point, flightin largely colocalises with a portion of the fluorescence distal to the M-line. At 72 hrs APF, both localisation patterns have expanded, reflecting the growth of the A-band both in length and width, but the flightin pattern extends around that of MHC19-IFM-GFP. This is due to the MHC isoform change that takes place between 48 and 72 hrs APF, after which the A-band extends (as seen by the flightin pattern) without a fluorescent myosin. Besides its expansion in the growing part of the A-band, the flightin pattern also expands towards the M-line, in the already established part of the thick filaments, presumably by the addition of new flightin molecules there. The MHC19-IFM-GFP signal, however, progressively weakens in the M-line region and its surroundings (as seen by the intensity plots), where flightin is still absent. In young adults no GFP fluorescence is visible in the M-line region, but the two MHC19-IFM-GFP foci flanking it remain. Even though by that time flightin has reached the area immediately adjacent to the M-line, it is not early enough to rescue the GFP fluorescence there (for a schematic description of these events, see Fig. 8). Overall, these observations show that the MHC19-IFM-GFP fluorescence appears unaltered in the sarcomeric regions where flightin was already present at the time of the MHC isoform change. We attribute this effect to flightin, which prevents MHC19-IFM-GFP turnover and replacement by the non-tagged MHC. We propose that one stabilising effect of flightin on the thick filaments is a result of it reducing myosin monomer exchange/turnover within the A-band.

The unique localisation pattern of fluorescent MHC in the IFM of weeP26 flies argued for an unexpected alternate splicing pattern of the inserted GFP exon, and the likely presence of an early IFM MHC isoform (MHC-IFM19). We have confirmed that a novel Mhc mRNA is produced in early IFM myogenesis, which contains, except for exon 18, all of the alternate exons of the previously well-described mRNA (Hastings and Emerson, 1991) that encodes the major IFM MHC isoform (MHC-IFM18). This agrees with an observation made by Suggs et al. (Suggs et al., 2007), who previously detected an Mhc transcript lacking exon 18 in wild-type IFM. In weeP26, the inclusion of the GFP exon in the MHC-IFM19 splice variant would generate the fluorescent MHC-GFP signal in the IFM.

The MHC-GFP localisation pattern in weeP26 IFM could arise either by a targeted localisation of the GFP-tagged isoform in the sarcomere, or due to differences in timing of MHC isoform expression and assembly during development. We have shown here that in the wild-type expression of MHC-IFM19 mRNA precedes that of the major isoform message (MHC-IFM18), which itself appears as MHC-IFM19 mRNA is in decline. The mRNAs for these isoforms are expressed almost mutually exclusively. This suggests that at least part of the explanation of the IFM MHC-GFP sarcomeric pattern in weeP26 flies is the timing of expression of these MHC isoforms. The localisation of MHC-GFP in the sarcomere leads to two general conclusions about IFM A-band assembly. Firstly, the A-band assembles from the core outwards. Once the initial thick filament core with a defined nascent M-line centre is formed, further lattice growth takes place by the sequential addition of thick filaments radially around it. Secondly, once established, thick filaments elongate by addition of myosin monomers to their tips, otherwise the GFP foci in weeP26 IFM would be seen to progressively move away from the M-line, or become irregular.

Transitions in myosin isoforms during development, such as from embryonic and neonatal to adult myosin isoforms, have been demonstrated in various vertebrate striated muscles, such as rat (Whalen et al., 1981; Lyons et al., 1983, Weydert et al., 1987; Lyons et al., 1990; LaFramboise et al., 1991; Hughes et al., 1993) and chicken pectoral muscle (Taylor and Bandman, 1989). In the latter case, individual thick filaments containing both neonatal and adult isoforms were documented. However, such transitions in vertebrates most likely represent remodelling rather than functional isoform changes during de novo assembly of individual sarcomeres, as each myosin isoform is capable of making thick filaments alone. A different mechanism is found during the myogenesis of Caenorhabditis elegans body wall muscles, in which two myosin isoforms are found. These are encoded by different genes, myosin A (myo-3) and myosin B (unc-54) (Miller et al., 1983; Mackenzie et al., 1978). Both isoforms are found in every thick filament in those muscles, with the minor myosin A isoform restricted to the region of the bare zone, and the major myosin B isoform present in the flanking regions (Epstein et al., 1982; Miller et al., 1983). Epstein et al. (Epstein et al., 1985) proposed that during thick filament formation myosin A, along with other accessory proteins, forms a nucleation core. Paramyosin then assembles on that core, recruiting myosin B. This mechanism suggests that myosins A and B have differing functions, with the former required for thick filament nucleation, which is supported by the myosin A mutant phenotype (Waterston, 1989). This is different, however, from the case of IFM sarcomere assembly; as shown in the weeP26 IFM sarcomeres, there are regions where thick filaments contain one or both MHC isoforms, so the advantage, if any, of using two MHC isoforms in the assembly of individual thick filaments is not clear, but may be required for normal A-band initiation and assembly.

Our evidence is that IFM A-band establishment takes place using an MHC isoform encoded by an mRNA that is identical to the major IFM-specific MHC mRNA except for the exclusion of exon 18. This makes exon 19 the terminally translated exon in the early IFM isoform. Therefore, the early and late MHC isoforms differ only in their C-termini, which are translated from either exons 18 or 19. This suggests that the C-terminal polypeptide sequence encoded by exon 19 is either essential for early sarcomere assembly, or unable to sustain the specific function of the adult IFM. Translation of exon 18 adds only a single residue before the stop codon, whereas that of exon 19 produces a peptide tail of 27 residues; the difference between the two isoforms can be reduced to the inclusion (MHC-IFM19) or exclusion (MHC-IFM18) of the peptide tail. In most members of the myosin II protein family, the majority of the rod is α-helical with the exception of the C-terminus, which is a random coil and is frequently referred to as the ‘non-helical’ C-terminal tailpiece. In Drosophila MHC the predicted helical region of the rod stops within the first few amino-acids encoded by exon 19. Therefore, the non-helical tailpiece of the fly MHC is coded by exon 19, so that MHC-19 molecules contain the non-helical tailpiece, but MHC-18 molecules lack it.

Even though regions in the myosin rod upstream of the tailpiece (light-meromyosin region and assembly-competent domain) have been implicated in MHC dimerisation and thick filament formation (Sohn et al., 1997; Cohen and Parry, 1998), little is known about the function of the tailpiece itself. So far, most of our knowledge of it comes from non-muscle and smooth muscle myosins. Paracrystals of the two naturally occurring vertebrate smooth myosin isoforms (SM1 and SM2), which have tailpieces of differing sizes, show differences in thick filament packing (Rovner et al., 2002), suggesting that the tailpiece is involved in thick filament polymerisation. In non-muscle myosin the tailpiece can promote filament assembly (Hodge et al., 1992; Ronen and Ravid, 2009), while its removal affects the stability of in vitro rod assemblies either positively or negatively, depending on the isoform (Ronen et al., 2010). Conversely, some studies claim that the tailpiece is not directly important for thick filament assembly itself (Ikebe et al., 2001, Hoppe et al., 2003), but might be required for stable positioning of the filaments in the sarcomere (Hoppe et al., 2003).

Since the Mhc exon 18 is used only in some muscles [IFM, TDT and leg muscles; summarised in (Zhang and Bernstein, 2001)] the MHC expressed in most fly muscles will contain the exon 19 encoded tailpiece. The appearance of MHC-GFP throughout the A-band of weeP26 TDT (Fig. 1C) suggests that MHC-19 isoforms are also expressed in that muscle. However, the absence of the tailpiece in the major IFM MHC isoform and its inclusion in an early, relative minor MHC localised specifically only in the IFM A-band core argues that it may have a specific function. We cannot exclude a modified proposal that initiation of thick filament assembly during the earliest stages of IFM sarcomerogenesis requires the presence of MHC with a tailpiece, but the assembly of further thick filaments does not.

Wells et al. (Wells et al., 1996) ectopically expressed a wild type embryonic Mhc transgene containing exon-19 in the IFMs of an Mhc null strain (Mhc¹) and showed that muscles assemble normally, but progressively degenerate with use. This shows that the tailpiece does not prevent this myosin assembling into thick filaments in the periphery of IFM sarcomeres. Swank et al. (Swank et al., 2000) specifically addressed the importance of the embryonic versus adult myosin rod (Mhc exons 19 versus 18) in the IFM, by expressing in Mhc null flies a transgenic construct based on an embryonic Mhc cDNA, but containing genomic DNA (including introns) for exons 17-18-19. Splicing of the transgene transcripts occurred in the IFMs to produce an exon 18 containing message. This improved the IFM phenotype of the embryonic (exon-19) Mhc construct seen by Wells et al. (Wells et al., 1996), by increasing stability and reducing degeneration in the adult stages. Swank et al. (Swank et al., 2000) concluded that the exon 18 terminus is important for the stability of the adult IFM sarcomeres. This might indeed suggest that the exon-19 tailpiece might have a deleterious effect in the function of this specialised muscle, so that it is absent from the later, major MHC isoform. Unfortunately whether the construct by Swank et al. also expressed MHC containing exon 19 in the IFMs was not determined, so the question as to whether an MHC with a tailpiece is required for the initiation of A-filament assembly in the IFM remains unresolved. Further transgenic experiments with an Mhc cDNA construct lacking exon 19 were beyond the scope of the present study.

The MHC tailpiece could also affect myosin activity. The tail of smooth muscle myosin can modulate contractility through its binding to the light-meromyosin region of adjacent myosins within a thick filament (Cai et al., 1995) and phosphorylation of residues in the tailpiece of molluscan MHC reduces myosin ATPase activity (Kuznicki et al., 1985; Collins et al., 1982a; Collins et al., 1982b; Castellani and Cohen, 1987). It may be important to restrict contractility during early sarcomere assembly.

In addition to the switch of MHC isoforms between 48 and 72 hrs APF in IFM development shown here, other sarcomeric proteins including troponin I, troponin T (Nongthomba et al., 2007) and sallimus (Burkart et al., 2007) each show an isoform switch during a similar period of IFM myogenesis. All of these switches result from changes in alternate splicing patterns. Sallimus (also known as Drosophila titin), an Ig-domain-containing elastic protein that links the Z-discs to the A-filament ends (Lakey et al., 1990; Kulke et al., 2001), has two IFM isoforms localized differently within each sarcomere, with the shorter kettin found throughout the Z-disc and the longer Sls(700) found only in the Z-disk centre (Burkart et al., 2007). The latter isoform would then be co-axial with the central MHC-IFM19 containing A-filaments, although there is no evidence linking the binding of Sls(700) to the MHC tailpiece. The isoform changes in sallimus, and the other proteins mentioned above, suggest a coordinated regulation of alternate splicing after 40 hrs APF. It must reflect either changes required for sarcomere assembly, changes in muscle function during muscle development, as has been described in vertebrates, or as a consequence of the evolutionary changes that led to the specific muscle gene regulation programme of these highly specialised muscles.

The flightin localisation pattern during IFM sarcomere assembly has not previously been reported. Flightin is observed in the developing sarcomere as early as 48 hrs APF, beginning at the A/I junction of the nascent A-band. The flightin pattern then progresses inwards towards the M-line. This localisation pattern suggests two independent processes by which flightin ‘decorates’ thick filaments. Our evidence is that it first binds close to the A/I junction, where thick filaments appear to be decorated as they grow from their tips. The second process takes place beginning behind the already decorated thick filament tips, and progresses by the sequential addition of flightin to previously assembled regions of the thick filaments towards the M-line. This explains the concave shape of the flightin pattern across the myofibril diameter at mid-development (see Figs 7, 8) as this process begins earlier for the thick filaments in the core, and therefore the migration towards the M-line is more advanced there. This process may be due to a ‘domino’ effect of already-bound flightin facilitating further flightin binding to more proximal neighbouring regions of the same thick filament. Flightin phosphorylation is not the enabling mechanism in this progressive binding as phosphorylation occurs entirely post-eclosion (Vigoreaux et al., 1993). Note however that a visible gap remains in the M-line region, suggesting that flightin is excluded from the bare zone. This agrees with immuno-EM studies of the flightin homologue (zeelin-2) in the IFMs of the water bug Lethocerus indicus, where the protein was shown to be absent from the bare zone (Ferguson et al., 1994; Reedy et al., 2000; Qiu et al., 2005). These authors also showed that flightin was absent from the very tips of the thick filaments of the A/I junction, presumably due to the presence of the connecting filaments (consisting of sallimus and projectin) in that region (Ferguson et al., 1994), though this is not resolvable in our work due to limitations of confocal microscopy. The possible presence of the connecting filaments in that region might argue for an interaction between them and flightin, assisting the decoration of the thick filaments with flightin as they grow. However, in the sarcomeric actin null Act88F^KM88, which lacks Z-disks and organised connecting filaments, the scattered thick filaments are decorated with flightin throughout (Reedy et al., 2000), suggesting that flightin decoration does not require interactions with the connecting filaments.

In weeP26 flies homozygous for the flightin null mutant fln⁰, MHC-IFM19-GFP is found throughout the A-band rather than being restricted to the core. In this situation the GFP-labelling of the A-filaments is much less intense than that observed in the presence of flightin, which we interpret as indicating that the MHC-IFM19-GFP is mixed with the untagged MHC throughout the A-band, including regions normally assembled after MHC-IFM19 expression has ceased. The simplest explanation is that in wild-type flies flightin binding restricts MHC to the location where it is initially incorporated into a thick filament; in its absence myosin molecules show ‘diffusion’, probably by undergoing rounds of incorporation/dissociation while thick filament assembly continues. This is reflected in the fact that the MHC-IFM19-GFP pattern in weeP26 IFM appears to be prevented from diffusing only in the areas where it co-localises with flightin (Fig. 7). Flightin is known to have a role in the structural integrity of IFM sarcomeres since in fln⁰ null mutants sarcomeres break during their earliest use (Reedy et al., 2000). Furthermore flightin appears to affect A-band assembly dynamics, as in its absence developing IFM sarcomeres are thinner and longer (Reedy et al., 2000) and in newly eclosed fln⁰ null flies contain thick filaments with more variable and overall greater lengths than wild type (Contompasis et al., 2010). We propose that flightin has a role in IFM assembly, where it reduces both thick filament association and dissociation rates of myosin molecules, thereby regulating thick filament assembly kinetics. Such an effect would lead to an overall greater stability of the IFM thick filaments, agreeing with previous proposals for flightin in enhancing thick filament stability, while regulating their length (Contompasis et al., 2010).

The results are consistent with the premyofibril model of sarcomere assembly, insofar as myofibrils develop without any formation of large pre-assembly complexes that are subsequently stitched into the complete structure. However, there are a number of fundamental differences between the IFMs and vertebrate sarcomere assembly. Firstly, IFM sarcomeres grow individually and synchronously from the nascent myofibril stage all the way to adulthood (Reedy and Beall, 1993). Secondly, Drosophila titin also referred to as sallimus, is not long enough and lacks extensive homologous C-terminal domains to act as a half sarcomere ruler protein such as proposed for vertebrate titin (reviewed in Trinick, 1996). Thirdly, the Drosophila genome does not include homologues to cytoplasmic intermediate filament proteins (Goldstein and Gunawardena, 2000) so such structures and proteins cannot play a part in myogenesis. Finally, important processes such as lateral fusion of nascent myofibrils to form more mature myofibrils and Z-disk splitting to introduce new sarcomeres, observed in vertebrate muscles and cells in culture (reviewed in Sanger et al., 2005) are not seen in the IFM (Reedy and Beall, 1993). The isoform transitions of MHC-IFM19 to MHC-IFM18 and of other proteins such as sallimus, highlight the establishment of the sarcomeric core and its subsequent growth as two different steps – core assembly and peripheral growth. We propose that these transitions reflect different requirements between early nascent myofibril establishment and later adult function of this specialist muscle.

The WeeP26 MHC-GFP stock (w¹¹¹⁸; P[weeP 26]) was generated in the screen of Clyne et al. (Clyne et al., 2003). The transposon insertion location was confirmed by genomic sequencing using primers Plac1 (CACCCAAGGCTCTGCTCCCACAAT) and Pry4 (CAATCATATCGCTGTCTCACTCA) in conjunction with Mhc exon 17 and 19 primers (described below). The flightin null stock fln⁰ (w; +; e fln⁰) was kindly provided by Dr Jim Vigoreaux (Reedy et al., 2000). The weeP26/fln⁰ stock was generated by crossing the weeP26 and fln⁰ stocks via a double balancer (w; CyO/If; MKRS/TM6,Tb), to yield double homozygotes (w; weeP26; e fln⁰). The Sls-GFP stock (w; +; P{PTT-un1}slsZCL2144; line G53 – Flytrap #ZCL2144; Morin et al., 2001) which has GFP-labelled Z-discs was used for the flightin localisation study. The Oregon-R wild-type strain (Bloomington Stock Center) was used as wild-type control. For developmental staging, white pre-pupae with everted spiracles (Bainbridge and Bownes, 1981) were removed into fresh vials at 25°C (time 0 hrs after puparium formation – APF) and harvested at required time-points. Adult flies were selected as ‘newly eclosed’ between 0 and 8 hrs post-eclosion.

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 MgCl₂, 5 mM EGTA, 5 mM ATP, 90 mM potassium propionate, 20 mM NaPi, pH 7.0). For early 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 in 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 for a further 15 minutes, then transferred back to relaxing solution. The IFMs from both dissections were permeabilised overnight in Triton-X/glycerol solution (50% v/v glycerol, 0.5% Triton X-100, 20 mM NaPi, 2 mM MgCl₂, 1 mM EGTA, 5 mM DTT, pH 7.0) at 4°C, and then washed twice with relaxing solution. Primary antibodies [anti-flightin rabbit polyclonal, (Reedy et al., 2000); anti-obscurin, rabbit polyclonal (Burkart et al., 2007); anti-kettin, MAC155, rat monoclonal (Lakey et al., 1990)] were applied overnight at 4°C. Muscles were washed twice, secondary antibodies were applied for 3 hrs, then the muscles were rinsed twice again. Image averages of the flightin localisation patterns in the IFM were generated in ImageJ (http://rsbweb.nih.gov/ij/), by cropping, exporting and manually aligning single sarcomere images of identical size. Images were compiled into a stack and their average was calculated using the Z-project tools. Intensity profiles were calculated using the ‘Plot Profile’ function, the values plotted in Microsoft Excel and further processed in CorelDraw.

The EM protocol largely followed that described by O'Donnell and Bernstein (O'Donnell and Bernstein, 1988). Adult half-thoraces were dissected in EM dissection buffer (100 mM sucrose, 10 mM sodium phosphate buffer, 2 mM EGTA, pH 7.2) and fixed overnight at 4°C in 3% paraformaldehyde, 2% glutaraldehyde in modified (100 mM sodium phosphate) dissection buffer. Specimens were washed (×3) in 100 mM sodium phosphate buffer pH 7.2, fixed for 45 minutes in 1% 100 mM OsO₄ on ice and washed (×3) in phosphate buffer. Samples were dehydrated in an alcohol series, washed (×3) in 100% ethanol and then in epoxypropane (×2). Samples were infiltrated with mixtures of araldite resin and epoxypropane 25% (30 minutes) then 50% (overnight), before transfer to 100% and resin for 2 hrs at 37°C, followed by polymerisation for 48 hrs at 60°C. Samples were sectioned and post-stained with lead citrate before being examined, and images were recorded in a FEI Tecnai 12 Bio Twin electron microscope.

Muscles were dissected as described above in PBS containing protease inhibitors (Complete Mini EDTA-free, Roche, No. 04693159001) and mRNA was extracted (Qiagen RNeasy Mini kit, animal tissue protocol). For 36, 48, 72 hrs APF pupae and newly eclosed flies, 40, 40, 20 and 10 sets of muscles were dissected, respectively (for further confirmation replicates were made separately). cDNA synthesis was performed using the Stratagene First Strand cDNA Synthesis kit. RT-PCR was performed using primers GACGAACTCCTGAACGAAGC (exon 17 forward) and TCAGGAGCAAGGTCGAATCT (Exon 19 reverse), multiplexed with primers for the endogenous reference ribosomal protein 49 (forward TCCTACCAGCTTCAAGATGAC, reverse GTGTATTCCGACCACGTTACA). Reactions were carried out using the Qiagen PCR Multiplex kit (No. 206143), using the program 95°C for 15 minutes, 30 or 34 cycles of 94°C for 30 seconds, 57°C for 90 seconds and 72°C for 120 seconds, and final extension 72°C for 5 minutes. PCR products were separated on 1.2% agarose gel and band intensities were calculated using GeneTools v 4.01 (Syngene). Relative band intensities were calculated by dividing the intensity value of the Mhc bands with that of the endogenous reference. The exon profile of the early (36 hr APF) IFM MHC isoform was determined by sequencing the alternative exons of the 36 hr cDNA using the following primers for the flanking constitutive exons; Exon 3 (MhcEx2F: CCAGTCGCAAATCAGGAG; MhcEx4R: CAGAGATGGCGAAAATATGG), Exon 7 (MhcEx5F: GGCTGGTGCTGATATTGAGA; MhcEx8R: TTGTACAGCTCGGCGGTATCG), Exon 9 (MhcEx8F: CGATACCGCCGAGCTGTACAA; MhcEx10R: TCGAACGCAGAGTGGTCAT), Exon 11 (MhcEx10F: ATGACCACTCTGCGTTCGA; MhcEx12R: GATCTTGCCCAGACCCTCATC), Exon 15 (MhcEx14F: CTCAAGCTCACCCAGGAGGCT; MhcEx16R: GGTCTCATCCAGTTTCGACTG).

Adult and pupal IFM samples were dissected as above. Tissues were immediately placed in loading buffer (312.5 mM Tris-HCl pH 6.8, 10% SDS, 0.5% bromophenol blue, 50% glycerol, 6 M urea) and denatured for 3 minutes at 95°C. Samples were run on ‘loose’ agarose-reinforced acrylamide SDS gels (Tatsumi and Hattori, 1995), but with an acrylamide concentration of 3%. Proteins were transferred for 3 hrs at 50 V. Membranes were stained for MHC (MAC147, rat monoclonal; Qiu et al., 2005).

We would like to thank Sean Sweeney for his support and the weeP26 line, and Belinda Bullard, Kevin Leonard, Peter O'Toole, Meg Stark, Betsy Pownall and Adam Middleton for their suggestions and help. Also, Dieter Fürst, Jörg Höhfeld, Riga Tawo and Jan Daerr for their support during the revision of this manuscript, and Farinaz Afsari for help with western blots.