Sequential myofibrillar breakdown accompanies mitotic division of mammalian cardiomyocytes

The contractile tissue of the heart is composed of individual cells, the cardiomyocytes. To ensure proper function, two multiprotein cytoskeletal complexes have to be assembled correctly, the myofibrils and the intercalated disks. The myofibrils consist of actin and myosin filaments that are organized in a paracristalline fashion to ensure maximal contractile force. Their basic unit is the sarcomere, which is defined as the region between two neighboring Z-disks, where the thin (actin) filaments are inserted. The thick (myosin) filaments are located in the middle of the sarcomere and held in place by another transverse structure, the M-band. A third filament system, the elastic (titin) filaments, seems to serve as a template for myofibril assembly as well as an elastic spring for the central positioning of the thick filaments during contraction (for review see Clark et al., 2002; Tskhovrebova and Trinick, 2003).

The intercalated disk is a specialized site of different types of cell-cell contacts that ensures intercellular communication, mechanical integration and force transmission between neighboring cardiomyocytes to guarantee optimal contractile work of the cardiac tissue (for review see Perriard et al., 2003). The heart is the first functioning organ in the embryo and any impairment of its function leads to early lethality. Nevertheless, cardiac growth has to be achieved while the heart is already involved in blood pumping activity. Before birth, this increase in mass is mainly due to division of the cardiomyocytes (hyperplasia), while after birth cardiac growth is caused by an increase in volume in the individual cardiomyocyte, a process called hypertrophy (Oparil et al., 1984; Li et al., 1996; Leu et al., 2001).

It is difficult to imagine how cardiomyocytes can handle both these highly dynamic processes, cell division and contraction of the myofibrils. Most cell types disassemble their cytoskeletal filaments before entering mitosis and microtubules reorganize to form the spindle apparatus, while actin and myosin make up the contractile ring that leads to cytokinesis (Sanger et al., 1989) (for review see Sanger and Sanger, 2000; Straight and Field, 2000; Nigg, 2001).

So far, comparatively little is known about how cardiomyocytes manage to divide and beat in the developing heart. To date, no stem-cell-like population of less differentiated cardiomyocytes has been identified in mammals that might serve as a continuously replicating population while fully differentiated cardiomyocytes do the contractile work (Rumyantsev, 1977). In the newt heart, where regeneration is possible also in the adult (Oberpriller and Oberpriller, 1971), differences in the proliferative potential of adult cardiomyocytes were detected recently (Bettencourt-Dias et al., 2003). Slowed proliferation was ascribed to cells from the future cardiac conduction system versus ventricular cardiomyocytes in the murine embryonic heart (Sedmera et al., 2003). There are few reports in the literature that describe seemingly intact myofibrils next to condensed chromosomes (Manasek, 1968; Kelly and Chacko, 1976), while others claim that myofibrils have to disassemble before cell division can occur (Goode, 1975; Rumyantsev, 1977; Kaneko et al., 1984). However, most of these studies were performed by electron microscopy, therefore presenting only a restricted view and a correlative study of proliferative events in the entire embryonic heart together with the investigation of cardiomyocyte cytoarchitecture is missing so far. To date, one of the most conclusive studies that deals with the fate of myofibrils in dividing cardiomyocytes used live birefringence microscopy to show that the cross-striated pattern was completely lost in newt cardiomyocytes undergoing cytokinesis, suggesting that myofibril disassembly has to occur before cardiomyocytes can divide (Kaneko et al., 1984).

We wanted to investigate this question by high-resolution confocal microscopy on triple stained specimens of cultured cardiomyocytes as well as by analysis of dividing cardiomyocytes in situ by using whole mount preparations of embryonic hearts. Using this method we were able to determine the state of myofibrillar organization in cardiomyocytes that had been ascertained as undergoing division by virtue of their expression of proliferation markers. Our results show that myofibrils are indeed disassembled in dividing cardiomyocytes but that this myofibrillar breakdown happens in a biphasic manner. Z-disk and thin-filament-associated proteins appear in a diffuse localization pattern before M-band and thick-filament-associated proteins. At the same time, cellular shape and cell-cell contacts remain remarkably similar to non-dividing cells. These results suggest that because of their demanding task of dividing in a contractile tissue and their elaborate cytoskeleton, embryonic cardiomyocytes have developed a specific strategy of coping with their cytoskeletal elements during cell division.

Embryonic (day 14) rat hearts were dissected, digested with collagenase (Worthington Biochemical Corp., Freehold, NJ) and pancreatin (Gibco Laboratories, Grand Island, NY) and cultured in maintenance medium (20% medium M199, 75% DBSS-K, 4% horse serum, 4 mM glutamine, 1% penicillin/streptomycin, 0.1 mM phenylephrine; DBSS-K: 6.8 g/l NaCl, 0.14 mM NaH₂PO₄, 0.2 mM CaCl₂, 0.2 mM MgSO₄, 1 mM dextrose, 2.7 mM NaHCO₃) in fibronectin-coated (10 μg/ml; Sigma) plastic dishes (Nunc) (Auerbach et al., 1997).

The proteasome inhibitor MG132 (Sigma) was used at a concentration of 20 μM for 3 hours immediately before fixation.

The cells were rinsed briefly in MP buffer (microtubule protecting buffer; 65 mM PIPES, 25 mM HEPES, 10 mM EGTA, 3 mM MgCl₂ (Schliwa et al., 1981), fixed for 10 minutes in 4% paraformaldehyde (PFA) in MP buffer and were then washed with MP buffer again. The cells were permeabilized with 0.2% Triton X-100 in MP for 5 minutes. Primary and secondary antibodies were diluted using 1% BSA in PBS and incubations were carried out at room temperature for 1 hour. After final washing three times with PBS, the cells were mounted with coverslips in 0.1 M Tris-HCl (pH 9.5)-glycerol (3:7) including 50 mg/ml n-propyl gallate as anti-fading reagent (Messerli et al., 1993).

The hearts of E14.5 mouse embryos were dissected in PBS and rinsed briefly with MP buffer followed by fixation in 4% PFA in MP buffer for 90 minutes. After several washes in PBS, the hearts were treated with hyaluronidase (1 mg/ml; Sigma) in PBS for 45 minutes at RT (room temperature) in order to remove the cardiac jelly and to ensure access for the antibodies to the inner myocardial wall (Tokuyasu and Maher, 1987). This was followed by permeabilization with 0.2% Triton X-100 in PBS for 45 minutes. After further washes in PBS and blocking with 5% NGS (normal pre-immune goat serum), 1% BSA (bovine serum albumin) in PBS for 45 minutes, the hearts were incubated with the primary antibody mixtures, diluted in blocking solution, shaking overnight at 4°C. After 4 ×2 hours washing in PBS with 0.002% Triton X-100 (PBT), the secondary antibodies were applied for overnight at 4°C. The hearts were washed with PBT for 6×1 hour and mounted on slides as described above (Ehler et al., 1999).

The monoclonal mouse anti myomesin (clone B4) antibody and the polyclonal rabbit anti MyBP-C were raised and characterized in our laboratory (Bähler et al., 1985; Grove et al., 1984). The monoclonal mouse antibodies against sarcomeric α-actinin (clone EA53), mouse anti α-tubulin (clone DM1A), polyclonal rabbit anti beta-catenin and polyclonal rabbit anti ubiquitin were from Sigma and the monoclonal rat anti α-tubulin (cloneYOL1/34) was from Abcam, UK. The polyclonal rabbit anti phosphorylated histone H3 antibody was from Upstate Biotechnology (via Lucernachem, Luzern, Switzerland). The monoclonal mouse anti sarcomeric myosin heavy chain (clone A4.1025) antibody was obtained from the Developmental Studies Hybridoma Bank, maintained at the University of Iowa, USA. The monoclonal mouse anti-titin (clones T51 and T12; M-line epitope and Z-disk epitope, respectively) antibodies were generously donated by Dieter Fürst (University of Potsdam, Germany) and the polyclonal rat antibody anti cardiac myosin binding protein-C was a kind gift from Mathias Gautel (King's College London, UK (van der Ven et al., 1999). The monoclonal mouse antibody against α-cardiac actin was obtained from Progen (Heidelberg, Germany).

For the triple immunofluorescence stainings, combinations of Cy3 anti rat (no cross reaction with mouse Ig), Cy2 anti rabbit and Cy5 anti mouse (no cross reaction with rat Ig) conjugated secondary antibodies were used. The secondary antibodies were purchased from Jackson ImmunoResearch via Milan (La Roche, Switzerland).

The specimen were analysed using confocal microscopy on an inverted microscope DM IRB/E equipped with a true confocal scanner TCS SP1, a PL APO 63×/1.32 oil and a PL APO 100×/1.40 oil immersion objective (Leica) as well as argon, helium-neon lasers. Image processing was done on a Silicon Graphics workstation using Imaris® (Bitplane AG, Zürich), a 3D multichannel image processing software specialized for confocal microscopy data sets.

To study the changes taking place in the cytoskeleton during cell division we first investigated primary cultures of embryonic cardiomyocytes by triple immunofluorescence. Dividing cells were unambiguously identified by staining with an antibody specific for phosphorylated histone H3 (Fig. 1B,E,H,K,N). Phosphorylation of Serine 10 in histone H3 is a prerequisite for the condensation of chromosomes during mitosis (Wei et al., 1998). The behaviour of the microtubular cytoskeleton shows no difference in dividing cardiomyocytes compared with other cultured cells (Fig. 1C,F,I,L,O). The microtubules are assembled to a spindle (Fig. 1F,I,L), which serves for segregation of the chromosomes and finally concentrated in the midzone region before the cells pinch off each other (Fig. 1O, arrow). A dramatic difference can be observed when the organization of the myofibrils is observed during cell division (Fig. 1A,D,G,J,M). Sarcomeric α-actinin, which is a component of the Z-disk, is localized in a cross-striated pattern in interphase cardiomyocytes as well as in cardiomyocytes in early prophase when the chromosomes are not yet condensed (right hand cell in Fig. 1A). In metaphase, however, the localization of α-actinin becomes completely diffuse and stays that way during telophase (Fig. 1D,G,J, left hand cell in A). Only in late cytokinesis as demonstrated in Fig. 1M, is the cross-striated localization pattern of α-actinin reestablished. These results suggest that the myofibrils undergo a disassembly-reassembly cycle during cell division in cultured cardiomyocytes.

To find out whether all components of the sarcomere are affected in a similar way during division we stained cultured cardiomyocytes for α-actinin, a Z-disk and an M-band epitope of titin, cardiac α-actin, sarcomeric myosin heavy chain, myosin binding protein-C and for the M-band protein myomesin. In dividing cardiomyocytes in metaphase [as identified by the visualization of condensed chromosomes with the staining for phosphorylated histone H3 (Fig. 2B,D,F,H,J,L,N)] interesting differences in the organization can be observed. The Z-disk-associated proteins, such as α-actinin and also Z-disk epitopes of titin, have a diffuse localization pattern at this stage of metaphase (Fig. 2A,C), whereas the organization of thick filaments remains remarkably intact despite the presence of condensed chromosomes in these cells [as indicated by the localization of sarcomeric myosin heavy chain (Fig. 2G), myosin binding protein-C (Fig. 2I) of the M-band with myomesin (Fig. 2K) and an M-band epitope of titin (Fig. 2M)]. It is very difficult to identify well-separated I-bands in cultured cardiomyocytes because of the different states of contraction; nevertheless the localization of cardiac actin appears more diffuse in metaphase cardiomyocytes as well (Fig. 2E). Therefore, there is an interesting delay in the way different parts of the sarcomere are disassembled during cell division with Z-disks and thin filaments attaining a diffuse localization pattern before A-band components.

To determine whether thick filaments and M-bands remain intact throughout cell division and how myofibril reassembly occurs, we compared the localization pattern of different sarcomeric proteins in cultured cardiomyocytes in anaphase, telophase and late cytokinesis. The different stages of the cell cycle were identified either by staining for phosphorylated histone H3 (Fig. 3D-F) or for tubulin (Fig. 3J-L,P-R). While cardiac α-actin shows a completely diffuse localization pattern in anaphase cardiomyocytes (Fig. 3A), M-bands and A-bands only start to get disassembled at this stage as indicated by the partially still-cross-striated staining pattern obtained for myomesin and sarcomeric myosin heavy chain (Fig. 3B and C, respectively). Complete myofibril disassembly is only seen in cardiomyocytes in telophase, as indicated by diffuse staining for M-band, thin and thick filament proteins (Fig. 3G-I). Reassembly of myofibrils occurs quite fast after cytokinesis because cross-striations for all investigated sarcomeric proteins can be seen soon after the cells have started to segregate from each other (Fig. 3M-O). These results suggest that a biphasic disassembly of myofibrils occurs in dividing cardiomyocytes, with Z-disk and thin-filament-associated components being disassembled before A- and M-bands. Nevertheless, disassembly of the entire myofibrils seems to occur in cardiomyocytes so that they can procede through telophase and complete cytokinesis. After cell division, reassembly of the myofibrils happens soon, which leads to a cross-striated localization pattern that is indistinguishable from neighboring non-dividing cells.

We determined that the observed differences in myofibril assembly of different parts of the sarcomere are not simply a reflection of slight temporal differences between the analyzed cardiomyocytes, by investigating the organization of the Z-disk protein α-actinin and MyBP-C as marker of the thick filaments in the same cell (Fig. 4). In interphase cells cross-striations can be seen for both sarcomeric components (Fig. 4A-C). However, in metaphase a clear distinction in the organization is apparent with α-actinin being mainly diffuse (Fig. 4E), while MyBP-C displays still distinct double bands in the same cell (Fig. 4F). By anaphase, however, both proteins have attained a diffuse localization pattern with some aggregated material (Fig. 4H,I), which is retained in early telophase (Fig. 4K,L). These observations show clearly that myofibril disassembly happens in at least two steps, with Z-disk material being disassembled before the thick filaments. A compilation of the state of assembly for different parts of the sarcomere at a given stage of cell division can be found in Table 1.

Cultured cardiomyocytes display important differences compared with cardiomyocytes in situ. There are obvious differences in cellular shape and in the surrounding of the cells, but there are also differences in cellular processes, such as responsiveness to growth factors (Armstrong et al., 2000) and in myofibrillogenesis (Ehler et al., 1999). Therefore, we wanted to find out whether the disassembly of myofibrils with the delay between different parts of the sarcomere could also be observed in cardiomyocytes in situ. Whole mount preparations of embryonic mouse hearts were stained for phosphorylated histone H3 and, at the same time, for the cell-cell contact protein beta-catenin to visualize the cell borders. In addition, different components of the sarcomere were stained and the whole mount preparations were subsequently analyzed by confocal microscopy. Interestingly, myofibrils are also disassembled in dividing cardiomyocytes in situ, with a similar delay between different parts of the sarcomere as in cultured cardiomyocytes. In metaphase cardiomyocytes the Z-disk-associated epitopes, such as α-actinin or titin T12, are already completely diffuse (Fig. 5B,E) while the M-band protein component (myomesin) still shows a partially cross-striated localization pattern (Fig. 5H, small arrows). It is only in late anaphase cardiomyocytes that M-band epitopes (such as titin T51, myomesin and sarcomeric myosin heavy chain) start to appear in a diffuse localization as well (Fig. 5K, data not shown). Therefore, myofibril disassembly also happens during cytokinesis in the developing heart in situ and shows a similar biphasic dynamic as in cultured cardiomyocytes.

In addition, dividing cardiomyocytes in situ stay tightly connected to their (presumably) contracting neighboring cells and there is only little change in the overall cellular shape. Staining for the adherens junction protein beta-catenin remains continuous in dividing cells, similar to the localization pattern seen in the surrounding non-dividing cardiomyocytes. At this stage of development, the segregation of intercalated disk proteins to the sites of terminal myofibril insertion has not yet been achieved and adherens junctions, as well as other types of cell-cell contacts, are still distributed all around the plasma membrane (Perriard et al., 2003).

To assess whether the myofibrils in dividing cardiomyocytes are merely disassembled and reassembled afterwards or whether protein degradation also plays a role in this process, we stained cultured cardiomyocytes with antibodies against ubiquitin (Fig. 6B,E,H) in combination with α-actinin to delineate the myofibrils (Fig. 6A,D,G) and with tubulin to assess the stage of cell division (Fig. 6C,F,I). Ubiquitination is the first step in non-lysosomal protein degradation (Hershko and Ciechanover, 1992) and we do indeed find an upregulation of ubiquitin expression in dividing cardiomyocytes. While in interphase cardiomyocytes the signal for the ubiquitin antibody is rather weak and mainly associated with the nuclei as reported previously (Hilenski et al., 1992) (Fig. 6B, cells in top left corner of E); once cardiomyocytes enter mitosis, ubiquitin starts to be spread throughout the cytoplasm (Fig. 6B, arrow). The ubiquitin fluorescence increases throughout telophase and remains high during cytokinesis, suggesting that ubiquitination of proteins takes place also in dividing cardiomyocytes (Fig. 6E,H). Owing to the low frequency of cell division in our cultures as well as in the developing heart in situ, we were unable to analyze by biochemical means whether myofibrillar proteins are ubiquitinated. However, the high intensity of the ubiquitin signal could mean that protein degradation of components of the sarcomere, or of proteins that control sarcomere integrity, might also happen during cell division.

Further evidence for the role of ubiquitin-mediated degradation in myofibril disassembly comes from experiments on dividing cardiomyocytes using inhibitors that interfere with the proteasome pathway. Treatment with MG132 leads to metaphase cells that still display cross-striations in the α-actinin staining, while control cells show a completely diffuse localization of this protein (Fig. 7).

We conclude that the myofibrils have to be disassembled to achieve successful cell division in cardiomyocytes and that this disassembly process is probably regulated by factors that are part of ubiquitination pathways.

By comparing the distribution of different components of the sarcomere in triple-stained dividing cardiomyocytes in culture and in the developing heart at different stages of mitosis, we were able to determine that myofibril disassembly occurs before cytokinesis in at least two steps. Disassembly of individual components of the myofibrillar apparatus like sarcomeric α-actinin or sarcomeric myosin heavy chain in cultured cardiomyocytes was described previously by others (Conrad et al., 1991; Li et al., 1996; Li et al., 1997; Du et al., 2003); however, so far no comparative analysis was performed. Our novel finding is that disassembly of the myofibrils shows biphasic dynamics. Proteins or epitopes that are associated with the Z-disk or the thin filaments display a diffuse localization pattern at a time when the thick filaments are still comparatively intact as indicated by a cross-striated pattern for proteins like myosin heavy chain, MyBP-C or myomesin. A representation of myofibril disassembly during cell division at the level of the individual sarcomere is depicted in Fig. 8. This process is exactly reversed to the time-course of myofibril assembly, where the first organized complexes consist of α-actinin, actin and Z-disk epitopes of titin at a time when all the thick filament components are still completely diffuse throughout the sarcomere (Ehler et al., 1999). Conversely, myofibril disassembly as it occurs during myopathies or muscle wasting is also characterized by degradation of Z-disk and I-band material before the rest of the sarcomere is affected (Taylor et al., 1995). During this process, degradation of myofibrils seems to be mainly regulated by the activity of muscle-specific calpain, which can associate with the I-band region of titin (Kinbara et al., 1998) and thereafter by the ubiquitin-proteasome pathway (Hasselgren and Fischer, 2001). A possible explanation for the prolonged existence of intact thick filaments comes from their structure and way of assembly. Myosin molecules are characterized by the ability to associate to bipolar filaments on their own in the test tube (Margossian et al., 1987). In addition, the putative M-band cross-linking molecule, myomesin, which seems to be important to provide a link between titin and myosin, binds so strongly to titin, that it can even be detected on isolated titin molecules (Nave et al., 1989). It appears that once the myosin molecules have been integrated by myomesin, they represent a rather stable structure. Also, during myofibrillogenesis in cultured cardiomyocytes, so-called floating A-bands have been observed, which were then integrated into nascent myofibrils (Schultheiss et al., 1990); suggesting, first, that the assembly of thick and thin filaments is an independent process and, second, that their disassembly might also be regulated autonomously.

How can M-bands still stay in register at a time when Z-disk epitopes of titin are already localized in a completely diffuse fashion? It has been thought that titin provides a basic cytoskeletal framework together with α-actinin and myomesin for myofibril assembly (Ehler et al., 1999). While this seems indeed to be the case for myofibrillogenesis, as indicated by the absence of properly formed myofibrils in cells that lack titin or express only truncated forms of it (van der Ven et al., 2000; Xu et al., 2002), there must be another means to hold the M-bands at least temporarily in place in dividing cardiomyocytes. One possibility is the intermediate filament network, consisting of desmin in cardiomyocytes. However, desmin filaments are mainly concentrated around the Z-disk and evidence for their continuation along the myofibrils to the M-band is still conflicting (Small et al., 1992). In addition, the cross-striated pattern that desmin filaments show in cardiomyocytes in situ is completely lost in cultured cardiomyocytes during the first days and reorganization only takes place after prolonged culture periods (Ehler and Perriard, 2000). Another possibility for alignment is a link from the myofibrils to the plasma membrane. The costameres (the major integrating complex between the myofibrils and the membrane) do show a cross-striated pattern in situ as well as in cultured cells, however, they are situated at the Z-disk region (Pardo et al., 1983). It is still not clear whether, and how, myofibrils are connected to the lateral membrane at the M-band level. Previously it was thought that skelemin serves as a linker between M-bands and the membrane (Price, 1987) but because skelemin has been identified as a splice variant of the integral M-band protein myomesin and embryonic M-bands consist exclusively of this isoform, this possibility is highly unlikely (Steiner et al., 1999; Agarkova et al., 2000). Another candidate for the linkage of M-bands to the membrane is spectrin, based on its localization pattern (Williams and Bloch, 1999; Flick and Konieczny, 2000). Recent evidence suggested an interaction between ankyrin, a binding partner of spectrin in erythrocytes and the giant M-band protein obscurin (Bagnato et al., 2003). It remains to be determined whether this multiprotein complex is indeed the filamentous material that provides a link between the M-band and the plasma membrane, as suggested by electron microscopy results (Pierobon-Bormioli, 1981; Nakamura et al., 1983).

In contrast to most cell lines, dividing cardiomyocytes in vitro, and even more so in situ, do not round up completely during mitosis and stay in relatively close association with their neighboring cells and the extracellular matrix. This was especially apparent when we studied dividing cells in whole mount preparations of embryonic hearts, where the cell-cell contact sites stayed absolutely intact in cells that were positive for phosphorylated histone H3. This is in agreement with pioneering ultrastructural studies on dividing cardiomyocytes, where it has also be shown that no loosening of cell-cell contacts occurs (Manasek, 1968). The fact that the contractile work of the cardiac tissue has to procede while individual cardiomyocytes divide, implies that this process does not interfere too much with the integrity of the tissue.

Currently it is not clear whether the myofibrils are only disassembled and the sarcomeric proteins are recycled again after cytokinesis, or whether protein degradation of myofibrillar components takes place as well. The drastic upregulation of ubiquitin expression in dividing cardiomyocytes suggests that protein degradation is important; however, possibly this is typical of proteins that are immediately involved in cell-cycle regulation like the cyclins and the cdks (Peters, 2002). Conversely, the increased expression levels of ubiquitin could also indicate a proteasome-independent function, as described recently (see review by Schnell and Hicke, 2003). However, the results from the MG132 experiment point towards a role of protein degradation in regulating myofibril disassembly, although not necessarily degradation of sarcomeric proteins. The fact that structural components, such as α-actinin, myomesin and titin (although no longer assembled into clear cross-striations), can still be stained by antibodies and show a subcellular localization that is distinct from the signal for ubiquitin supports the argument that these sarcomeric components are recycled. Similar recycling processes seem to occur when adult rod-shaped cardiomyocytes adapt to culture conditions. While they flatten and attach to the culture dish in the presence of serum, their myofibrillar material is aggregated to a clump; once the cells have spread, myofibrils are reassembled and beating activity is resumed (Messerli et al., 1993). In addition, it has been shown that myofibril assembly in cultured myocytes is not blocked by addition of cycloheximide, suggesting that protein synthesis is not essential (Rumyantsev, 1977).

At present the factors that regulate myofibril disassembly during mitosis have not been identified yet. The existence of a putative Z-disk degradation factor was postulated thirty years ago (Rumyantsev, 1977), however, its molecular nature is still unclear. Numerous proteins that are either integral components or transiently associated with the Z-disk have been identified in the recent years (Faulkner et al., 2001), but for none of them has a direct involvement in the organization of cardiomyocyte proliferation been shown so far. Also, at the M-band, numerous proteins that are potentially involved in signalling pathways have been identified, in addition to bona fide structural sarcomeric proteins. Among these signalling proteins are, for example, the MURFs (muscle-specific ring finger proteins) that are associated with the ubiquitin pathway (Bodine et al., 2001; McElhinny et al., 2002; Pizon et al., 2002) and have been shown to be involved in the regulation of muscle atrophy. Obscurin, is a giant M-band protein that possesses several domains that have been associated with different signalling pathways and also contains a Rho guanine exchange factor domain, suggesting that it could be involved in the activation of the small GTPase Rho (Young et al., 2001). Activation of the Rho signalling pathway seems to be important for cardiomyocyte proliferation during development, as demonstrated by the embryonic lethality of the conditional expression of the Rho inhibitor GDIα in the heart (Wei et al., 2002). In conclusion, these proteins might represent a sensory machine that is associated with the sarcomere under normal conditions, but it might adopt a second function as signalling molecules under conditions of additional work load, stress and myofibril disassembly during cytokinesis, or in disease. Unfortunately, the low rate of division in culture as well as in situ does not permit a thorough biochemical analysis of the identity of the ubiquitinated proteins at present; however based on results published on proteins for other cell types an ubiquitination of, for example, the MURFs is likely to occur. Preliminary experiments in our lab have shown that the signals for MURF-2 and for ubiquitin show a similar subcellular localization in dividing cultured cardiomyocytes (P.A., E.E., M. Gautel and J.-C.P., unpublished).

It is astonishing that the cells should break down such an elaborate structure as the myofibrils to undergo mitosis and it will be exciting to find out about the signalling pathways that lead to this disassembly and possibly degradation process. The observation that the myofibrils have to be disassembled for cytokinesis to occur might provide also a simple mechanistic explanation, why cardiomyocytes cease to divide after birth. With the hypertrophic growth that is caused by the increased workload on the heart after birth, this disassembly-reassembly process might be simply too costly from an energetic point of view. In addition, too many cytoskeletal elements in the form of myofibrils might physically impede cell division as well. This, together with the alterations in the expression levels of proteins that regulate the cell cycle and cytokinesis, respectively, might contribute to the uncoupling of karyokinetic and cytokinetic events as seen in postnatal rodent cardiomyocytes (Li et al., 1996; Georgescu et al., 1997; Poolman and Brooks, 1998).

This research was supported by grants from the Swiss National Foundation and by a PhD training grant from the ETH. We are grateful to Alain Hirschy, Stephan Lange and other members of the lab for their helpful comments. In addition, we thank Nicolas Dard (Institute of Biochemistry, ETH Zurich) for the kind gift of MG132. Antibodies against different titin epitopes were generously donated by Dieter Fürst (University of Potsdam, Germany). We also thank Mathias Gautel (King's College London, UK) for his support and for the kind gift of antibodies against cardiac myosin binding protein-C.