There are a number of theories about how myofibrils are formed initially (see Sanger et al. 2005). Much of the work has been carried out on cultured cell lines or isolated neonatal cells. In such myocytes or myotubes, the observations have led to the idea of premyofibrils or stress fibre-like structures which contain sarcomeric proteins but only later become full length mature sarcomeres. These in vitro studies use essentially a two-dimensional system rather than a three-dimensional one. In three-dimensional or in vivo systems, there is little evidence for these intermediate sarcomeric structures (Ehler et al. 1999, 2004). There is, however, in agreement that some kind of scaffold of Z-disc and M-band proteins is laid down first and thick and thin filaments are then incorporated. Increase in length of the myofibrils in the in vitro systems appears to be by the addition of new sarcomeres at their ends.

In early heart, in vivo, only short myofibrils with a mature sarcomere length are seen (Hirschy et al. 2006). The question arises, ‘Where are new sarcomeres added?’ The evidence suggests that, at least in the mature heart, they are added at the ends of the fibrils near the intercalated disc.

Evidence for the growth of muscle cells occurring at their ends is reviewed in (Russell et al. 2010). In skeletal muscle, growth occurs at the ends of the fibres near the myotendinous junction (MTJ). In early work Williams and Goldspink (1971) labelled young rapidly growing skeletal muscle with tritiated adrenaline to follow where actin is deposited and found it accumulated at the ends of the fibres. This was supported by the observation in the electron microscope of copious ribosomes in these areas and bundles of thick filaments (no Z-discs) and narrow myofibrils at the periphery of the muscle. Dix and Eisenberg (1990) using mRNA detection of slow myosin heavy chain on stretched stimulated TA muscle in the rabbit which normally expresses the fast isoform, found a large accumulation of the slow isoform at the MTJ. In addition, in contrast to the control muscle where the regular sarcomeric structure is present up to the edge of the MTJ, in stretched muscle, bundles of filaments and thin myofibrils in a sea of new T-tubules, ribosomes and mitochondria were seen at the MTJ.

There is a constant turnover of sarcomeric proteins in muscle fibres (Willis et al. 2009). However, evidence for formation of new sarcomeres within the body of the fibre is seen only after eccentric exercise. Here, local changes in the distribution of desmin and Z-disc proteins such as α-actinin are seen and have been attributed to myofibril remodelling after damage inflicted by sudden large transient stretches (Yu et al. 2004). Such repair sites are not observed in control muscle. Similar repair or addition of sarcomeres has been seen in cardiomyocytes as a result of a major 10 % stretch (Yu and Russell 2005). In this case the stretch was permanent so the cells had to generate a number of additional sarcomeres in order to normalise the sarcomere length. The rate at which this occurred was about one sarcomere (2 μm) per hour much greater than normal even in early heart growth (see Sect. 2.1). In addition to fibril remodelling, the stretch also generated substantial changes in the morphology of the ID suggesting the involvement of this region in growth. This observation is supported by recent evidence that there are significant changes at the ID correlated with growth brought about by volume overload in the rabbit heart (Yoshida et al. 2010). In addition, evidence for sarcomere addition at the ID has been seen in both normal heart and those suffering from dilated cardiomyopathy (DCM) (Wilson et al. 2014). In order to understand this, and indeed the more extensive role of the ID, we have to look more closely at its structure and its relationship with the myofibril.

What exactly do we mean by the intercalated disc? The main body of the cardiomyocyte comprises the myofibrils interspersed with columns of mitochondria extending from one end of the cell to the other (Fig. 3.1b). These are held, together with the sarcoplasmic reticulum and t-tubules, within a cytoskeletal network of proteins such as desmin, the whole being anchored to the extracellular matrix (ECM) at the lateral plasma membrane by the costameres (Granger and Lazarides 1978; Craig and Pardo 1983; Pardo et al. 1983) (Fig. 3.2c). The ID is seen in the light microscope as a high density junction between the end of one heart muscle cell and the next. With the advent of electron microscopy, it became clear that it was a complicated structure with regions of highly folded membrane of variable amplitude (compare Fig. 3.2a, c). Here, we adopt the definition of the ID given by Forbes and Sperelakis (1985). It includes the membrane folds and their axial connections as well as the material within the folds delineated by the dotted lines in Fig. 3.2c.

The structure of the ID has been beautifully described in several excellent electron microscope investigations (see, for example, Fawcett and McNutt 1969; Forbes and Sperelakis 1985). The general form of the ID is a stepped structure. The treads of the steps are lateral stretches containing folded membrane (Fig. 3.2). In transverse section the folds (or peaks) are seen as circular sections so that the ID tread membrane can be thought of as having a structure like an egg box (Fig. 3.2d). Within the folds, the thin filaments from the terminal sarcomeres of the myofibrils run to the membrane (Fig. 3.3a). Since the myofibrils take up approximately two-thirds of the cross-sectional area of the cell, much of this membrane is associated with actin adhering junctions. However, since the mitochondria occupy typically most of the other third of the cell a significant portion of the ID structure is available for other functional regions such as desmosomes, signalling domains and possible some that relate to mitochondria. The risers of the ID steps are axial and usually have a length of one or more sarcomere lengths. Often in the mouse these steps contain gap junctions (Fig. 3.2c). It is worth noting that the close association of the membrane of the neighbouring cells at the ID means that there is little or no ECM here.

The main protein components of the ID plaque are those of adherens junctions (AJ) and desmosomes (Fig. 3.5) (Borrmann et al. 2006; Franke et al. 2006). AJs are similar in organisation to desmosomes. The transmembrane protein, N-cadherin, is bound to the armadillo proteins, p120, β-catenin and α-catenin. This complex is connected through vinculin, an actin binding protein, to the thin filaments. It seems that in the very early embryonic heart, desmosomes and adherens junctions are separate but by birth they are intimately combined (Pieperhoff and Franke 2007). The resulting junctions have been called areae compositae (Borrmann et al. 2006) or hybrid AJs (Li and Radice 2010). Here they will be called composite junctions (CJ). It is not clear what the functions of the desmosomal proteins are in the CJ. There are apparently obvious desmosomes where desmin filaments attach as described above, but it is not obvious that these filaments penetrate the dense plaque of the CJ and bind. The distribution of desmin with respect to N-cadherin at the ID seen in Fig. 3.4a suggests that there is not a strong correlation between the two. Possibly, the close association of AJ and desmosome gives rise to a stronger junction that is necessary to support the high tensions found in contracting heart. Of the proteins involved, one, plakoglobin (γ-catenin), is known to operate in both types of junction and can replace β-catenin in the adherens junctions (Zhurinsky et al. 2000). Another desmosomal protein, plakophilin-2, has been shown to interact with a specific region of αT-catenin, its adhesion modulation domain (Goossens et al. 2007). These two interactions might be the means of knitting the adherens and desmosomal proteins into a specialised domain.

Two primary cytoskeletal complexes are known to underlay and support membranes. One based on dystrophin and the other on spectrin. Both are found at the lateral plasma membrane in the cardiomyocyte, but only spectrin is seen at the ID (Stevenson et al. 2005). Several components of the spectrin complex have been located there, including αII spectrin, βII spectrin, ankyrin-G and proteins 4.1R and 4.1G (Mohler et al. 2004; Bennett et al. 2006; Pinder et al. 2012). The distribution of two of these at least, αII spectrin and 4.1R, does not overlap with β-catenin or connexin43, in support of their presence on the non-junctional membrane (Bennett et al. 2006; Bennett 2012; Pinder et al. 2012; Wilson et al. 2014). 4.1R is a multivalent protein stabilising the actin/spectrin interaction and interacting through its FERM domain with membrane signalling proteins (Baines et al. 2014). Ankyrin-G also links spectrin with membrane bound signalling proteins and has a pivotal role in locating the sodium voltage channel NaV_1.5 at the ID (Mohler et al. 2004).

Other, non-Z disc, proteins exhibit the same doublet distribution across the ID. N-RAP is a nebulin-like repeat protein with five nebulin super repeats suggesting it binds to actin/tropomyosin filaments (Luo et al. 1997; Zhang et al. 2001). It may replace the Z-disc protein, nebulette, at the TrJ. Another protein with the doublet spacing is αII spectrin which has been localised at the tops (peaks) of the ID folds (Fig. 3.4c) (Bennett et al. 2006; Wilson et al. 2014). This suggests that there is a membranous region associated with the TrJ and that there is a close structural and functional connection between them. A protein that may link between the spectrin-rich membrane folds and the myofibrillar TrJ is filamin C which exhibits the doublet distribution (van der Ven et al. 2000). Filamin C is known to bind between membrane proteins, sarcoglycans, present at the ID and the ID proteins, N-RAP and Xin (van der Ven et al. 2006).

We have seen that the increase in length of cardiomyocytes is likely to occur at the end of the cells, at the ID. An interesting feature of the ID is that the amplitude of the membrane folds varies (Figs. 3.2 and 3.3). Measurements of its amplitude show that it can vary from 20 nm to 2 μm with an average of 0.5 μm (Wilson et al. 2014). A myocytes 100 μm in length could therefore grow by about 1–2 % simply by increasing the ID amplitude. The heart grows as a result of local pressures to pump more blood. This is likely to be a small incremental effect over time so the ability of the cells to slowly change their length will be important. Equally so will be the ability to reverse the process when demands on the heart are reduced (Frenzel et al. 1988).

Increase in the ID amplitude involves both the elongation of the thin filaments and an increase in the surface area of the ID membrane. As described above, proteins that aid thin filament length extension, formins and MENA/VASP, are present in the ID. As the thin filaments extend the membrane also extends. In a normal muscle the ID amplitude is on average ~0.5 μm. A crude estimate suggests that the area of the ID membrane at this amplitude is about 1.6× that at its smallest amplitude. An increase in amplitude to 2 μm involves a ~5-fold increase in membrane area. Perhaps not surprisingly, high amplitude IDs are often the site of multiple vesicles (Fig. 3.3c). Both coated vesicles and calveolae have been previously described at the ID (Fawcett and McNutt 1969; McNutt and Fawcett 1969).

At higher amplitudes, there is a difference in the appearance of the CJ regions. At low amplitude the CJ covers essentially all the membrane slopes (Fig. 3.3a). As the amplitude increases, the CJ seems to fragment and smaller lengths of plaque are seen (Fig. 3.3b, c). This is perhaps not surprising since there are a fixed number of thin filaments coming from a sarcomere. We might assume that a single thin filament will require a given amount of plaque complex for a firm attachment to the membrane. Consequently, a smaller proportion of the membrane of high amplitude IDs will be needed for this function. As a result of the plaque breakup, there is now a significant amount of ‘bare’ membrane which could lead to an increase in signalling domains.

It is clear that there is a natural limit to the amplitude of the ID folds. They rarely grow longer than 2 μm. This corresponds to one sarcomere length and suggests that at this point, a new sarcomere could be inserted. Evidence for this comes from the work of Yoshida et al. (2010). Working with a model of volume overload by arterio/venous fistula in rabbits, they showed that there was hypertrophic growth and that the myocytes grew every 2 days by two sarcomeres. In the course of this time there was change in the ID morphology. In particular, starting from a normal ID amplitude the ID grew more convoluted until it reached a size of ~2 μm. It then returned to its original size. During this time it went through several morphological stages one of which the authors attributed to the insertion of a sarcomere on each side of ID. On removal of the overload the process reversed.

Wilson et al. (2014) have looked for evidence for this happening during normal growth. Since the normal rate of growth in the mouse heart is something like one sarcomere a week and it has been shown that sarcomeres can be inserted in neonatal rat cardiomyocytes at the rate of 1 per hour, it might be a rare event to capture in normal growth. In high amplitude folds in IDs, sarcomeres are seen occasionally (Fig. 3.7b). Furthermore, it was sometimes possible to see isolated A-bands, bundles of thick filaments, or single filaments within these folds (Figs. 3.3c and 3.7a). Interestingly there was no Z-disc associated with these partial structures. They were reminiscent of the nascent filament assemblies described at the ends of skeletal muscle fibres near the MTJ in young rapidly growing animals or after stretch (Williams and Goldspink 1971; Dix and Eisenberg 1990).

Since the long folds are symmetrical with respect to the neighbouring cells the insertion of sarcomeres on both sides of the ID are possible. But it is not clear that both fibrils that meet at the ID will grow.

These observations suggest a possible mechanism of ID growth and sarcomere addition in the ID (Fig. 3.7c) (Wilson et al. 2014). The TrJ is already a proto Z-disc containing many of the proteins necessary for Z-disc formation. In addition, there is continual turnover of sarcomeric proteins throughout the myocyte so the materials would be available (Willis et al. 2009). When the ID amplitude reaches a length of ~2 μm, thick filaments begin to appear in the ID fold. The proteins could be made in the ID region since ribosomes are found here or brought in the vesicles observed in the long ID folds. Alternatively, thick filaments could be transported in from elsewhere along microtubules as previously reported (Pizon et al. 2002). It certainly seems that the thick filaments are organised into bundles at the ID. This would involve the cooperation of M-line proteins, titin, myomesin and obscurin (Van der Ven et al. 1999). The new Z-disc based on the TrJ would require severing the old thin filaments at the TrJ and production of a new set of thin filaments of opposite polarity. Horowits and his colleagues have shown that N-RAP is involved in Z-disc formation (Manisastry et al. 2009). They suggest that dimers of N-RAP could lead to antiparallel organisation of thin filaments at nascent Z-discs (Crawford and Horowits 2011). The presence of N-RAP at the TrJ could therefore facilitate the formation of a new Z-disc there. As the Z-disc matures N-RAP apparently dissociates presumably to make way for nebulette together with the other Z-disc proteins, telethonin and FATZ. Telethonin, in particular, is known to be one of the last proteins to appear at the maturing Z-disc.

Inserted sarcomeres are surrounded by long tongues of membrane which would need to break up and be reabsorbed (Yoshida et al. 2010). It has been seen that peripheral junctional SR is often present at the top of the ID folds (Fig. 3.3b). Also, there are membrane associated proteins such as spectrin and myofibril/membrane bridging proteins such as filamin which are also found within the myocyte at T-tubule/Z-disc locations (Kostin et al. 1998). Possibly the tops of the ID folds together with the associated SR are involved in the formation of new t-tubules and SR at the new Z-disc (Fig. 3.7c).