A targeted deletion of the C-terminal end of titin, including the titin kinase domain, impairs myofibrillogenesis

Skeletal muscle fibres are formed by the fusion of mononucleated myoblasts into multinucleated myotubes. Just prior to, and shortly after fusion, the first myofibrils are assembled from newly expressed skeletal muscle-specific proteins such as skeletal myosin, skeletal actin and titin. As this happens, the actin-based cytoskeleton completely re-organises from one similar to that found in crawling fibroblasts to the highly ordered structure found in skeletal muscle. The assembly of skeletal muscle proteins into myofibrils is a complex process that is not well understood.

Titin, one of the largest known proteins (over 3 MDa) (Kurzban and Wang, 1988; Maruyama et al., 1984) is thought to play a pivotal role in myofibrillogenesis in a number of ways. It is important in the organisation of the Z-line, in organising the myosin-containing thick filaments into muscle sarcomeres, and in the organisation of the M-line (reviewed by Gautel et al., 1999; Gregorio et al., 1999; Maruyama, 1997; Tskhovrebova and Trinick, 2002). The coding sequence is around 90 kb (Labeit and Kolmerer, 1995), the protein is 1 μm long and stretches from the Z-line to the M-line in the muscle sarcomere (Furst et al., 1988; Whiting et al., 1989). The amino-terminal region of titin is found at the Z-line (Labeit and Kolmerer, 1995; Yajima et al., 1996), the I-band region is the main source of muscle elasticity and the A-band region has been suggested to regulate thick filament length (Labeit et al., 1992; Whiting et al., 1989). The M-line region of titin binds to M-line proteins such as myomesin (Obermann et al., 1995).

At the end of the A-band, close to the M-line, titin contains a kinase domain (TK) first described in 1992 (Labeit et al., 1992). This kinase consists of a catalytic domain and a regulatory domain (which blocks ATP binding). It is activated both by phosphorylation of Y170 (titin kinase residue number) (Mayans et al., 1998) by an unknown tyrosine kinase and by Ca²⁺/calmodulin binding (Mayans et al., 1998). One of the known substrates for this kinase is telethonin [also called t-cap (Gregorio et al., 1998; Mues et al., 1998)]. To phosphorylate telethonin, TK must be activated, and be close to telethonin, which means either telethonin has to be recruited to the M-line to be phosphorylated, or TK phosphorylates telethonin before it is assembled into the muscle sarcomere early in development. This tight spatial and temporal control of TK activity suggests that it may be an important early event in myofibrillogenesis, but it may also be important in control and regulation of turnover at later stages. Overexpression of the isolated TK leads to a breakdown in myofibrillar structure that supports the idea that TK plays a crucial role in the control of myofibrillogenesis (Mayans et al., 1998).

Mutations in titin are known to cause heart diseases such as dilated cardiomyopathy, skeletal muscle diseases such as tibial muscular dystrophy, as well as heart and skeletal muscle defects in zebrafish, Drosophila and cultured cells (reviewed by Hein et al., 2000; Hein and Schaper, 2002). Myofibrillogenesis is severely disrupted in cultured BHK cells that lack titin (Van der Ven et al., 2000), in Drosophila muscle where the titin gene has been deleted (Zhang et al., 2000) and in cultured cardiac cells transfected with antisense DNA to titin (Person et al., 2000). Most recently, a conditional deletion of two exons (Mex1 and 2) close to the M-line of titin, produced severe heart and skeletal muscle defects (Gotthardt et al., 2003). As titin is important for so many aspects of myofibrillogenesis and myofibrillar maintenance, this is not surprising. If the titin kinase is essential for initiating myofibrillogenesis, through its phosphorylation of telethonin and other substrates, then we might also predict the truncation of titin to exclude the kinase domain and downstream sequence, would disrupt myofibrillogenesis, as the titin kinase would no longer be able to phosphorylate its substrates and control myofibrillogenesis, and would interfere with the normal function of titin at the M-line. It is significant that mutations in telethonin are known to result in mild limb girdle muscular dystrophy type 2G (Moreira et al., 2000).

We used gene targeting in cultured myoblasts to truncate the titin gene in the middle of the kinase domain, and examined the effects of this disruption on primary myofibrillogenesis, which is not possible in the existing conditional M-line knockout. Only one titin allele is targeted in our experiments. However, the many cardiac and skeletal muscle diseases reported recently (reviewed by Hein and Schaper, 2002) are all autosomal dominant, and hence a single targeted allele is likely to affect muscle function.

A genomic clone, Ti2, 20 kb in size, was recovered by screening a λ phage library, prepared from H2k^b-tsA58 conditionally immortal myogenic cells (custom built, Stratagene), using radiolabelled cDNA probes corresponding to sequence in/around the kinase domain of human cardiac titin (a gift from Dr S. Labeit, EMBL). To build the targeting construct, the thymidine kinase gene (RSV promoter; a gift from Alastair Reith) was cloned into the NotI-HindIII sites of PCRII vector (Invitrogen), a 10 kb EcoRI subfragment from the genomic titin clone was subcloned into the EcoRI site upstream, and finally, a SalI-XhoI fragment, containing the neomycin gene under control of the Pgk promoter (a gift from Alastair Reith) was cloned into a unique SalI site in the titin genomic DNA, in the same transcriptional direction. The 5′ and 3′ ends of the 10 kb EcoRI fragment corresponding to bp 68843 and bp 76320, and the SalI site to bp 75085 in the human cDNA sequence (accession number X90568), where GTTGAC in the human sequence is replaced by GTCGAC in the mouse sequence, generating the SalI site. The SalI site is at residue 194 in the titin kinase domain [titin kinase residue number (Mayans et al., 1998)]. Interruption of titin at this site will remove the regulatory tail of the kinase and downstream sequence, and will probably result in an unfolded kinase. Presence of both the neomycin and thymidine kinase genes allows for positive/negative selection (Mansour et al., 1988).

A single myogenic clone at low passage (p4), isolated from the skeletal muscle of 1- to 2-day old immortal mice, was used. The cells were isolated and cultured as described previously (Miller et al., 2003; Morgan et al., 1994) in growth medium [DMEM (Glutamax) supplemented with 20% foetal calf serum (FCS), 2% chick embryo extract (CEE), 100 μg/ml penicillin/streptomycin (P/S), and 20 units/ml murine recombinant γ interferon (Gibco)] at 33°C. Prior to transfection, the ploidy of the cells was checked and found to be normal. This was done by incubating exponentially growing cultures overnight at 37°C in an open system, exposing the cultures to 2 μg/ml colcemid (Sigma) for 4 hours on the following day, followed by trypsinisation (0.25% (w/v) trypsin and 0.02% (w/v) EDTA in phosphate-buffered saline for 5 minutes), incubation in 0.075 M potassium chloride for 15 minutes, and finally washing 3 times with Carnoy's fixative. Slides were analysed by GTL-banding, and between 21 and 26 metaphases were examined for each culture.

Cells were transfected by electroporation. Exponentially growing cells were trypsinised, counted and diluted in DMEM supplemented with 10% FCS, at 2×10⁷ cells per ml. 50 μg of the targeting plasmid, linearised by a NotI digest, was added to 0.5 ml of cells in an electroporation cuvette. After 1 minute, the cells were electroporated at 250 V, 1500 μF using the EquiBio CellJect (Flowgen), and immediately plated out into fresh growth medium in two 20 cm dishes. To positively select for permanently transfected clones, 1 mg/ml G418 (500 μg/ml active) was added to both plates 48 hours later. For negative selection, 2 μM gancyclovir was added to one of the plates at the same time, and kept in the medium for the next 5 days. Individual G418^r, or G418^r, GANC^r clones were picked after 14-21 days and expanded in the absence of G418 and gancyclovir before freezing stocks and genomic DNA analysis.

To isolate genomic DNA, cells were washed with PBS and rocked at 37°C for 6 hours in lysis buffer (0.1 M Tris-HCl pH 8.5, 5 mM EDTA, 0.2% SDS, 0.2 M NaCl, 20 mg/ml pronase). The DNA was then precipitated with 1 volume of isopropanol spooled onto a pipette, then resuspended in 100 μl Tris EDTA (10 mM Tris HCl, pH 7.4; 1 mM EDTA, pH 8.0).

Initial screening for a single targeted allele was performed by PCR on the genomic DNA, using a forward primer from the Pgk-Neo gene, and a reverse primer in the genomic DNA 3′ to the interruption site, designed using sequence from a genomic subclone in this region. PCR on genomic DNA using these two primers was expected to generate an amplified product for a correctly targeted locus of 2.1 kb. 181 clones were screened by PCR (77 G419^r, and 104 G419^r and GANC^r) and 4 potential targeted clones were identified (frequency of 2%). The cycling regime used in the PCR was 94°C for 2.5 minutes, followed by 35 cycles of 94°C for 1 minute, 67°C for 1 minute, 72°C for 2.5 minutes, followed by 1 cycle of 72°C for 10 minutes.

Each of the 4 clones that gave a band of the predicted size in the PCR screen was then further analysed by Southern blotting. A minimum of 15 μg of genomic DNA was digested by HindIII, and the bands separated by electrophoresis through an 0.8% agarose gel. The DNA was then transferred to nylon membranes (Hybond N+) and hybridised, under modified Church conditions with a 1.5 kb BamHI-SalI titin probe, external to the introduced DNA fragment (Fig. 1). We expected that this probe would detect a band of 5.7 kb in an untargeted allele, and that the size of this band would increase to 7.2 kb, if the allele had been targeted. DNA was also prepared from normal untransfected cells, for comparison. The karyotype of the targeted clones was also determined, as described above.

To examine the effect of a single targeted allele on myofibrillogenesis, myoblasts were cultured on laminin-coated coverslips and differentiated by replacing growth medium with differentiation medium (DMEM supplemented with 1% FCS, 1% CEE, P/S), removing γ-interferon, and increasing the temperature to 37°C. After 5 days, the cells were fixed in 4% paraformaldehyde in PBS for 30 minutes. Paraformaldehyde-fixed cells were permeabilised using 0.5% Triton X-100 prior to staining.

The antibodies used were: A1025, which recognises all skeletal myosin (Cho et al., 1994); anti-α-sarcomeric actinin (Sigma; A5044), anti-Z-line titin (T12; Boehringer Mannheim); anti-myomesin [an M-line protein (Obermann et al., 1996)] and anti-M-line titin [Ti51(Van der Ven et al., 1999), both gifts from Dieter Furst]; anti-telethonin, which localises at the Z-disc in cultured human myotubes (Mues et al., 1998), anti-MURF2, which associates with microtubules, titin and myosin early in myofibrillogenesis and can be found at the M-line and in the nucleus (Pizon et al., 2002) and anti-obscurin (Young et al., 2001), which is associated with the amino Z-disc region of titin in adult skeletal and cardiac muscle, but can also be observed at the M-line in developing muscle. Secondary antibodies, conjugated to Alexa 488, or 564 were purchased from Molecular Probes. Stained cells were mounted in pro-long antifade (Molecular Probes), and imaged using fluorescence and confocal microscopy, using either a Zeiss LSM Pascal, or a Zeiss LSM 510 Meta microscope.

To determine whether the targeted cells expressed a smaller titin isoform, as a result of the targeted disruption of the titin gene, protein samples from differentiated myotubes were analysed by SDS-PAGE. 5-day myotubes were washed in PBS, and directly lysed into modified Laemelli buffer (63 mM Tris-HCl pH 6.8, 2% SDS, 10% glycerol, 5% β-mercaptoethanol, 6 M urea, 0.005% Bromophenol Blue) for protein gel analysis. Three samples were made from three different cultures for wild-type and targeted cells. The samples were stored in aliquots at -80°C. Prior to loading on the gel, samples were heated to 80°C for 5 minutes. Proteins were separated by SDS-PAGE using gradient gels (2-7.5%) that were cast in a mini-protean III system (Bio-Rad). The gels were run at 100 V for 2-3 hours. Gels were either silver stained using a silver stain kit (Bio-Rad), Coomassie stained or transferred to a nitrocellulose membrane, at 250 mA for either 2 hours when probed with the Z1/Z2 antibody or 3 hours when probed with the anti-kinase antibody (Gautel et al., 1995), for western analysis. The blots were boiled for 30 minutes in H₂O to enhance antibody sensitivity, blocked (0.1% BSA, 0.2% gelatin and 0.1% Tween 20 in PBS) for 1 hour and incubated overnight in primary antibody at room temperature. Blots were washed with PBS-Tween 20, incubated in horseradish peroxidase-conjugated goat anti-rabbit antibody (Pierce, UK) for 1 hour and then exposed to substrate (Pierce). The antibodies used were as above, and in addition Z1/Z2 antibody (a gift from S. Labeit) or anti-kinase. Western analysis involved blocking overnight in blocking solution and exposure of the blot to the anti-kinase antibody for only 1 hour.

The length of myotubes was measured on digital phase contrast images from the wild-type and targeted (c20) clones, differentiated for 5 days under the same conditions. Approximately 50 myotubes from wild-type and targeted clones were measured.

Following electroporation of the targeting construct into a single wild-type myoblast clone, 181 G418^r or G418^r GANC^r clones were recovered and analysed to determine if the titin allele had been targeted. In the initial PCR screen, we detected a band of the correct size in 4 clones, a targeting frequency of 2% (Fig. 2A). Analysis of these 4 clones by southern blotting showed that only one of the 4 clones (clone 20) had a single extra band of the molecular mass expected if one of the titin alleles had been targeted (Fig. 2B, lane 1). This suggests that only clone 20 had been correctly targeted. While clone 25 (Fig. 2B lane 2) also had an additional band of the correct size, it had a third band of intermediate size. This clone also showed additional PCR bands (Fig. 2A, lanes 5 and 6), suggesting that this clone may not be correctly targeted.

The intensity of the targeted band for clone 20 was lower than that of the untargeted band (Fig. 2B, lane 1). One explanation for this reduced intensity is that the cells are no longer diploid, although they had been diploid when the cells were transfected. Analysis of the ploidy of the 4 potentially targeted clones identified from the PCR screen as well as the control cells (the original wild-type clone used for the electroporation) that had not been targeted showed that none of these clones was now 100% diploid. Clone 20, the correctly targeted clone, was 55% diploid (Table 1). All the cultures, including the original wild-type clone, contained a mixture of diploid cells (with 40 chromosomes) and tetraploid cells (with 80 chromosomes). This change in ploidy is likely to be a consequence of long term tissue culture, and is commonly seen in other muscle cell lines such as C2 cells. However, the change of ploidy in both the targeted and untargeted wild-type cells demonstrates that the change of ploidy in the targeted cells is not a direct result of targeting the titin allele.

Clone 20 not only showed the clearest evidence that the titin allele had been targeted on the southern blot, but also was the clone that had the fewest tetraploid cells, so we have investigated the effect of the targeted truncation of the titin gene in this clone on myofibrillogenesis, in comparison with myofibrillogenesis in the untargeted wild-type cells. However, it should be pointed out that we think it is unlikely that any differences described below are the result of the change of ploidy, as both wild-type and targeted cells have become tetraploid.

As a result of targeting one of the titin alleles, we expected that a truncated titin protein would be expressed that was missing the C-terminal 1895 amino acids, equivalent to a reduction in molecular mass of approximately 0.2 MDa. However, there was also a possibility that the truncated protein would be unstable, and rapidly degraded, such that the targeting event would result in an overall reduction in titin expression levels. To determine which was the case, we analysed titin expression in targeted and control (wt) cells by SDS-gel electrophoresis and western blotting.

We found that there were two differently sized bands for titin expressed in wild-type cells, which are likely to be two different isoforms. Their molecular masses were similar to those described for skeletal and cardiac isoforms (approximately 3.3 MDa and 3.0 MDa), from silver stained gels (Fig. 3a, lane 4). Western analysis showed that both of these bands react with anti-titin antibodies for epitopes at the Z-line (Z1/Z2, Fig. 3b, lane 2) and the kinase domain (Fig. 3c, lanes 1 and 3), suggesting that these are both full-length titin isoforms, and not breakdown products. However further analysis using a panel of titin antibodies would be needed to determine which exons are expressed for the two isoforms to determine whether the shorter isoform is a 'cardiac' titin, or a short form of the skeletal titin. Both cardiac and skeletal titin are expressed from a single gene, but tissue-specific exon skipping events can result in different sized isoforms (Freiburg et al., 2000). Titin isoform expression by cultured myotubes has not been previously described, but it is possible that the developing myotubes can express more than one isoform in common with other proteins. An example is the co-expression of skeletal and cardiac actin in fusing myoblasts (Cox et al., 1990).

In contrast, in the targeted cells, although we again observed two bands (Fig. 3a, lanes 1-3), the lower of the two bands had a lower molecular mass than that in the wild-type clone. Furthermore, while the anti-Z-line epitope antibody reacted with both bands (Fig. 3b, lane 2), the anti-kinase antibody only reacted with the higher molecular mass band (Fig. 3c, lanes 1 and 3), suggesting that the lower molecular mass band is a truncated titin isoform that has lost the kinase domain. This suggests that a truncated titin protein is expressed from the targeted titin allele.

However, it is also clear that the expression levels of titin in the targeted clone are lower that in the wild-type clone (Table 2). From equally loaded protein gels, we quantified the total intensity of the two bands obtained with the Z1/Z2 antibody and for the kinase antibody, for wild-type and targeted myotubes and we found that the total intensity was about 25-30% less in the targeted clone than in the wild-type clone (Table 2). Most of this reduction appears to arise from a reduction in expression levels of the lower band, from our analysis of the relative intensities of the upper and lower bands in the wild-type and targeted cells, using the Z1/Z2 antibody (Table 2). Furthermore, we also found that the levels of skeletal myosin expression were also reduced in the targeted clone, compared to the wild-type clone (Table 2). This suggests that truncation of the titin gene also results in reduction of both titin and myosin expression levels, and this could be explained by a failure of the targeted cells to differentiate as well as the wild-type cells (see below).

We found that myofibrillogenesis was markedly poorer in the targeted clone than in the wild-type clone in two major ways. First, the targeted cells did not form myotubes as well as wild-type cells. The majority of the myotubes in the targeted clones are short (Fig. 5) even though almost all the cells in the culture have differentiated, as judged by their positive immunostaining for skeletal myosin (Fig. 5). We measured the length of the myotubes for wild-type and targeted clones, and found that the average length in the targeted clone (c20) was 151.9±10.2 μm, which is approximately half the length of that of the wild-type cells; 325.7±16.6 μm (Fig. 4). This suggests that the targeted cells may not fuse as well as wild-type cells. The fluorescent images of myotubes stained with antibodies against skeletal myosin formed by the targeted cells had an 'emptier' appearance than those formed by the control cells.

Second the staining patterns for titin, α-actinin and myosin and other sarcomeric proteins demonstrate that the myofibrils are less well organised in the targeted clone than in wild-type cells (Figs 6 and 7). Overall, although there is evidence that myofibrils form reasonably well, as striations can be observed, the sarcomeres are generally less well ordered and in particular, their lateral alignment is poorer (Figs 6, 7). Using antibodies against the Z-line region of titin (T12), or antibodies against skeletal α-actinin, part of the Z-line, we found that the Z-lines are less well ordered, or in register, in the targeted cells (c20) as in the wild-type cells. Using an antibody against the M-line region of titin (Ti51), again we found that the M-line staining was less regular and sharply defined in the targeted clone as in the wild-type cells (Fig. 6). Furthermore, using an antibody against skeletal myosin, we also found that the muscle sarcomeres were less well ordered in targeted cells, than in wild-type cells (Fig. 6).

To look further into sarcomeric organisation in wild-type and targeted (c20) cells, we co-immunostained for myomesin and telethonin (Fig. 7A-F), skeletal myosin and MURF2 (Fig. 7G-L) and skeletal myosin and obscurin (Fig. 7M-R). We found well organised stripes of myomesin staining in wild-type cells as expected (Fig. 7A), but the staining pattern in targeted cells (c20) was remarkably unaffected (Fig. 7D). Telethonin staining did not show a sarcomeric pattern in wild-type cells but it was evenly distributed in a punctuate staining pattern throughout the cytoplasm. However, in targeted cells, while the telethonin staining was still cytoplasmic it was less evenly distributed, and in particularly there was an increased perinuclear staining (Fig. 7E), suggesting that some of the telethonin may be mistargeted.

Staining for MURF2 in wild-type cells appeared partly filamentous, which is probably because of its binding to microtubules (Pizon et al., 2002) as well as M-line staining, whereas the MURF2 staining in targeted cells was more disorganised (Fig. 7G-L). Again, the sarcomeric organisation of skeletal myosin was less well ordered in targeted cells than in wild type (Fig. 7J,P) as shown in singly stained cells (Fig. 6). Finally, in myotubes co- stained for skeletal myosin and obscurin (Fig. 7M-R), obscurin was localised at the M-line in wild-type cells (see arrows in Fig. 7O), but a less well ordered pattern was observed in c20 cells.

These results demonstrate that a targeted truncation of the titin gene inside the kinase domain, which disrupts the titin kinase and alters the reading frame of all the titin sequence downstream of the titin kinase, interferes with myofibrillogenesis. This is despite the fact that one normal copy of the titin gene is still expressed, and that there is probably expression of a truncated titin protein from the targeted allele. These results clearly demonstrate that disruption of the titin gene, close to its 3′ end, is sufficient to disrupt myofibrillogenesis and underlines the importance of all of the functional domains of titin in organising the contractile proteins into muscle sarcomeres.

Interruption of the titin gene in the kinase domain could either have resulted in loss of titin protein, due to an unstable transcript, or in a truncated protein, containing all of titin up to the kinase domain. We found that there is both expression of a truncated protein, and a reduction in expression levels. This will result in a reduction in the levels of titin kinase, and domains C-terminal to the titin kinase, which includes the M-line region of titin important for binding to myomesin (Labeit et al., 1992; Trinick, 1994) as well as KSP sites that are developmentally phosphorylated (Gautel et al., 1993). We have not demonstrated that the truncated titin is able to integrate into the muscle sarcomere normally, but we suspect that it should be able to do so, at least up to the existing A-band section.

As well as the reduction in titin expression levels, we also found that the expression levels of myosin were reduced, and that myoblast fusion was less efficient in the targeted cells, resulting in shorter myotubes. Mutations in titin have also been found to affect fusion in Drosophila (Zhang et al., 2000). Furthermore, the targeted deletion of the titin exons Mex1 and Mex2, which contain the titin kinase domain, resulted in fewer muscle fibres (Gotthardt et al., 2003). Taken together these results may suggest that the C-terminal region of titin is important in a signalling pathway that is involved in the control of myofibrillogenesis and protein turnover in the myotubes and in muscle. This could be both through the activity of the titin kinase, and/or through the binding of M-line proteins, such as myomesin, to the COOH-terminal domains of titin.

It is not surprising that the integration of the truncated titin protein, lacking the M-line domains, into the muscle sarcomere results in a less well ordered M-line, as we have observed here by immunolocalisation of M-line region of titin and the kinase associated MURF2. Surprisingly, obscurin is also mislocalised suggesting that its M-line targeting may depend on a yet unidentified titin interaction. The de novo assembly of muscle-specific proteins into muscle sarcomeres is still poorly understood, and there are conflicting ideas on how this might happen. Either pre-myofibrils form first, which consist of Z-lines and muscle-specific actin together with non-muscle myosin and titin (Sanger et al., 2002), and sarcomeric thick filament proteins are subsequently exchanged into the pre-myofibril to form sarcomeres. In an expansion of this concept, stress-fibre-like structures containing titin and other thick filament proteins except sarcomeric myosin form first, and then sarcomeric myosin is inserted to form muscle sarcomeres (Van der Ven et al., 1999). In this second model, the formation of a cytoskeletal scaffold is initiated by the association of titin, actin and α-actinin at the Z-line, followed by association of titin with myomesin at the M-line, before myosin filaments are assembled into the sarcomere (Ehler et al., 1999; Van der Ven et al., 1999). Therefore, the disruption of the M-line might be expected to affect integration of myosin into the muscle sarcomeres, as the scaffold may not be fully formed, and this could explain the less well ordered sarcomeres that we observed when we stained for skeletal myosin. A similar disruption of the M-line was reported for mice with conditional knockout of the Mex1 and Mex2 exons of titin that contain the titin kinase domain (Gotthardt et al., 2003), although myomesin staining was not discussed. Interestingly, we found that the staining for myomesin, which binds to titin at the M-line was less affected. This may suggest that assembly of myomesin onto the M-line is an early event that does not depend on a full complement of titin molecules in the A-band and that one normal allele is sufficient for targeting.

We also expect that disruption of the titin kinase will probably result in a reduced phosphorylation of its substrates such as telethonin (Mayans et al., 1998) which is found at the Z-line in adult cardiac and skeletal muscle (Mues et al., 1998). However, from our immunolocalisation studies, we found that telethonin is not at the Z-line in the developing myotubes, but has a punctate cytoplasmic localisation. This cytosolic localisation was also observed in an embryonic mouse heart in-situ (E. Ehler and M. Gautel, unpublished). Myostatin is a member of the TGFβ family that is a negative regulator of muscle growth (Grobet et al., 1997; McPherron and Lee, 1997) and is found in cardiac as well as skeletal muscle (Sharma et al., 1999) Secretion of myostatin leads to decreased myoblast proliferation and differentiation (Joulia et al., 2003). This leads to the interesting possibility that one of the downstream effects of interrupting the titin kinase exon, probably through a reduction in telethonin phosphorylation, is mistargeting of telethonin, which in turn affects myostatin secretion in these early myotubes. This would be expected to have a general knock on effect on myoblast fusion and early differentiation, as we observed here.

Surprisingly, we also found that disruption of the C-terminal region of titin affected Z-line organisation. It also affects the organisation of obscurin, which can localise to both Z-discs and M-bands (Young et al., 2001). As the formation of Z-lines is an early process, and occurs before sarcomeres can be observed, we might expect that Z-lines would form normally in the targeted cells. The disruption we observe suggests that the C-terminal region of titin is also somehow important in organisation of the Z-line. A potential explanation is that a reduced kinase activity, results in decreased phosphorylation of telethonin, and a decrease in its binding at the Z-line through its interactions with FATZ, a protein that also binds to α-actinin and filamin at the Z-line (Faulkner et al., 2000), and to titin at the Z-line (Zou et al., 2003). However, this is unlikely as we did not find clear localisation of telethonin at the Z-line in either wild-type or targeted cells. The effects could be due to an altered activity for an unknown substrate, for the kinase important for Z-line formation, or some other mechanism. A large number of proteins are now being identified that bind at the Z-line (reviewed by Sanger and Sanger, 2001) and that also bind to titin, and it is clear that we have much to learn about how the Z-line assembles, how this assembly is controlled and how the Z-lines connect up with focal adhesions at the cell membrane during myofibrillogenesis.

In summary, we have found that truncation of the titin gene, resulting in a protein that lacks the titin kinase and M-line domain, affects myofibrillogenesis, disrupts the organisation of Z-lines and causes some mistargeting of telethonin and obscurin. This suggests that C-terminal domains of titin, in particular the titin kinase, may be important in organising the Z-line, possibly though phosphorylation of telethonin or other unidentified substrates.

We would like to thank Dr Paul Roberts for helping us to perform the karyotyping of the mouse cells before and after targeting, and Dorothy Wright, who originally recovered the titin genomic clone in Dr Peckham's laboratory. We would like to thank Siegfried Labeit and Dieter Furst for generous gifts of antibodies and DNA probes. This work was funded by Human Frontiers, BBSRC, the Royal Society, the Medical Research Council (M.G.) and the Wellcome Trust. The Zeiss 510 Meta confocal microscope was funded by SRIF.