Apart from loss of cardiac tissue as for example by ischemia as caused by a myocardial infarction, two major forms of cardiomyopathy can be defined, hypertrophic cardiomyopathy (HCM) and dilated cardiomyopathy (DCM). Another type of cardiomyopathy, arrhythmogenic right ventricular cardiomyopathy (ARVC), is less frequent, but also associated with changes in cytoarchitecture (Basso et al. 2012). As we accumulate more data from next generation sequencing, the frequent hereditary component of these diseases becomes more evident and is no longer restricted to huge families. On a macroscopic basis HCM is defined as a thickening of the ventricular and/or septal wall and thus a reduction in chamber size and blood volume at filling (Seidman and Seidman 2001). DCM is characterised by a ballooning heart with an increased diameter of the ventricle (Seidman and Seidman 2001). As sudden cardiac death, especially in young athletes, can be the first sign of a HCM or an ARVC, more knowledge on the genetic cause of these diseases and screening for known variations can be potentially life-saving (Basso et al. 2012). Classically HCM is defined as being caused by mutation in proteins that constitute the myofibrils, while DCM is often associated by mutated cytoskeletal proteins (Chien 1999). The most frequent mutations that cause HCM are seen in sarcomeric myosin heavy chain and truncations of myosin binding protein-C (Frey et al. 2012), while truncating mutations in titin were initially seen in DCM (Gerull et al. 2002) and recently proposed to be among the main causes for DCM (Herman et al. 2012). However, the more sequence data are accumulated the more obvious it becomes that these simple associations are not correct (for recent review of disease causing mutations see McNally et al. 2013) and that for example proteins such as MLP and AMPK, which are not integral myofibrillar proteins, can be mutated in HCM, while mutations in classical myofibrillar proteins such as myosin can also lead to DCM. Truncating mutations in titin were also found to be associated with all types of cardiomyopathies (McNally et al. 2013). A more clear-cut link seems to exist at present for ARVC, which has been named the disease of the desmosome, a special type of cell–cell contact that links the intermediate filament cytoskeleton to the plasma membrane and is seen as mainly playing a structural role (Rickelt and Pieperhoff 2012). However, also for ARVC the situation will become more complex the more next generation sequencing (NGS) data are available and already now a link between the developing disease and Wnt signalling, which is usually active mainly in early embryonic heart development, is proposed (Asimaki et al. 2009; Swope et al. 2013).

At the level of the cardiomyocyte, HCM and DCM present quite different phenotypes (Fig. 1.4). HCM is usually already evident by Hematoxylin and Eosin staining and presents as myofibrillar disarray (Frey et al. 2012). The cardiomyocytes have a larger diameter and there is also a change in the arrangement of the tissue, with cell–cell contacts as characterised by the adhesion junction protein beta-catenin also being found at the sides of the cells. Apart from obvious regions of necrosis, DCM tissue can appear relatively unharmed at the histological and ultrastructural level and it required the careful analysis of the first mouse model for this disease, the MLP knockout mouse (Arber et al. 1997), to get a grip on the alterations that happen at the cellular level in this disease (Ehler et al. 2001). Freshly isolated cardiomyocytes from MLP knockout hearts are characterised by a more irregular shape compared to healthy hearts (see Fig. 1.1c), but there is no consistent cellular elongation, which might explain the ballooning ventricles but rather a general loss of size control (Leu et al. 2001). Similar observations were also made in other mouse models for DCM and even in samples from human DCM patients, suggesting that this may be a more general phenomenon of the disease (Hirschy et al. 2010; Pluess et al. 2015). With the obvious consequences that this has on the three-dimensional arrangement of heart tissue, it may provide some of the explanations for the alterations that are seen in the ways the cells make contact with their extracellular matrix. For example, the costameres become disorganised in the MLP knockout mouse and also the way the extracellular matrix is distributed over the cardiomyocytes changes in cardiomyopathy (Ehler 2010; Ehler et al. 2001; Hoskins et al. 2010). This probably reflects a loss in membrane organisation especially as far as triads are concerned and was also reported by ion conductance microscopy on chronic heart failure samples (Nikolaev et al. 2010). While the organisation and composition of the myofibrils is much less affected in DCM compared to HCM and there is for example no consistent re-expression of embryonic isoforms of the contractile proteins actin and myosin (MacLellan and Schneider 2000), there is a shift towards the expression of more compliant isoforms of elastic proteins that are needed to maintain a working sarcomere such as longer versions of titin and the EH-myomesin isoform (Makarenko et al. 2004; Schoenauer et al. 2011; Pluess et al. 2015). The major changes that are observed in the subcellular organisation of the cardiomyocytes in DCM happen at the intercalated disc. We consistently see an upregulation in the expression of actin-anchoring proteins such as adherens junction components (beta-catenin, cadherin) and also vinculin and N-RAP (Ehler et al. 2001; Pluess et al. 2015). This is potentially a response to an increased convolution of the plasma membrane and is also accompanied by a more intense signal for filamentous actin. Whether the elevated levels of an actin polymerising factor, the formin FHOD1, which are observed at the intercalated disc in DCM cause this increase in actin remains to be demonstrated (Dwyer et al. 2014). In mouse models for DCM this altered adherens junction composition is accompanied by a reduction in the numbers of gap junctions that are detected, suggesting impaired intercellular communication (Ehler et al. 2001). This observation can only be reproduced in a subset of the human patient material, indicating a difference between mouse and human phenotype and a potential compensatory effect by differential medication of human patients.

A multitude of studies have been carried out that investigate and compare the subcellular localisation of a variety of cytoskeletal proteins either in cardiomyocytes in culture or on heart sections using immunofluorescence with antibodies that specifically recognise a particular protein. Especially when combined with confocal microscopy, these investigations have told us a lot about the subcellular distribution of proteins that make up the cardiac cytoarchitecture during development and in disease. However, the researcher wants to go beyond these descriptive studies and to explore the role of different components in a more functional way. There the problem starts, because the ideal model system to study cardiac cytoarchitecture does not exist and what is available to us comes all with advantages and disadvantages (for a list comparing different systems see Table 1.1). Over the years many laboratories have used cultured cardiomyocytes for their analyses, which have the advantage that they retain some of their differentiated characteristics, for example they usually continue to beat rhythmically, but the obvious disadvantage that these cells are taken out of their environment and placed into a plastic dish in culture medium. This makes them extremely amenable for interference by addition of drugs and for microscopy, but also means that it cannot be excluded that some of the observations that are made are a culture artefact. Especially in the response to growth factors and also in the way myofibrils are assembled there seem to be dramatic differences between two-dimensional as in the petri dish and three-dimensional arrangements (Armstrong et al. 2000; Ehler et al. 1999). Another potential pitfall is that the stalwart system of cardiomyocytes in vitro are neonatal rat cardiomyocytes (NRC), which means that they are taken from an animal at a time when full maturity is not reached, yet. It is therefore entirely possible that the responsiveness to stimuli will be different compared to adult cardiomyocytes and their cytoarchitecture is definitely distinct (Rothen-Rutishauser et al. 1998). A lot of isoforms and proteins that are specific for either the embryonic or the adult heart are in the process of being switched around birth; therefore embryonic markers such as ANF expression and EH-myomesin isoform expression are still maintained at least in a subset of the NRC (Fig. 1.5a), while markers of sarcomere maturity such as the expression of telethonin or M-protein are not yet present in all the cells. In addition, a beating frequency in the hundreds per minute makes them dramatically different from the beating rate that would be expected from a human cardiomyocyte. Cardiomyocytes that were differentiated from human inducible pluripotent stem cells may become the method of choice for in vitro studies in the future, but are at the moment characterised by their costliness and also immature features as far as their physiology and cytoarchitecture is concerned, which makes them similar to differentiated human embryonic stem cells (Földes et al. 2014). Another inherent problem with cultured cardiomyocytes whether rodent or human is that their cytoarchitecture is different compared to in situ, especially as far as the cell–cell contacts are concerned. While in the postnatal heart cell–cell contacts would only be found where myofibrils terminate, they can also be situated laterally to myofibrils in cultured cardiomyocytes (see arrow in Fig. 1.5b). This means that the terminology intercalated disc-like structure would be more appropriate in this situation since while they contain all the major components, their ultrastructural architecture will be different to an intercalated disc in situ. Freshly isolated cardiomyocytes from adult mouse or rat hearts can only be obtained by retrograde perfusion of the tissue with collagenase using a Langendorff apparatus and are characterised by a cytoarchitecture that is as close to the tissue as possible. However, they only have a limited life span to do experiments on them, which is even shorter for the ones from mouse than for the ones from the rat heart. The dedifferentiation process starts immediately and for example the specialised plasma membrane organisation is lost, with slightly different dynamics between rat and mouse (Pavlovic et al. 2010). A possibility to maintain the characteristic subcellular distribution of proteins such as connexin-43 a bit longer may be to assemble the adult cardiomyocytes to three-dimensional spheres using gravity (Kelm et al. 2006).