|Home | About | Journals | Submit | Contact Us | Français|
Myosin binding protein C (MyBP-C) is a component of the thick filament of striated muscle. The importance of this protein is revealed by recent evidence that mutations in the cardiac gene are a major cause of familial hypertrophic cardiomyopathy. Here we investigate the distribution of MyBP-C in the A-bands of cardiac and skeletal muscles and compare this to the A-band structure in cardiac muscle of MyBP-C-deficient mice. We have used a novel averaging technique to obtain the axial density distribution of A-bands in electron micrographs of well-preserved specimens. We show that cardiac and skeletal A-bands are very similar, with a length of 1.58 ± 0.01 μm. In normal cardiac and skeletal muscle, the distributions are very similar, showing clearly the series of 11 prominent accessory protein stripes in each half of the A-band spaced axially at 43-nm intervals and starting at the edge of the bare zone. We show by antibody labelling that in cardiac muscle the distal nine stripes are the location of MyBP-C. These stripes are considerably suppressed in the knockout mouse hearts as expected. Myosin heads on the surface of the thick filament in relaxed muscle are thought to be arranged in a three-stranded quasi-helix with a mean 14.3-nm axial cross bridge spacing and a 43 nm helix repeat. Extra “forbidden” meridional reflections, at orders of 43 nm, in X-ray diffraction patterns of muscle have been interpreted as due to an axial perturbation of some levels of myosin heads. However, in the MyBP-C-deficient hearts these extra meridional reflections are weak or absent, suggesting that they are due to MyBP-C itself or to MyBP-C in combination with a head perturbation brought about by the presence of MyBP-C.
The vertebrate muscle sarcomere is composed of a highly ordered assembly of different proteins that interact to produce contraction. The general relationship of the major players, actin and myosin, in the vertebrate sarcomere is well established; however, our knowledge of the detailed organisation of the thick filament is lacking. Myosin binding protein C (MyBP-C), originally named C-protein by Offer et al.,1 is one important component of the thick filament that has not been directly visualised in isolated filaments.2–4 Interest in MyBP-C has increased recently because of the observation that mutations in the gene encoding for the cardiac isoform are a major cause of hypertrophic cardiomyopathy, a disease that affects 1 in 500 of the population.5
The striated muscle sarcomere in the relaxed state has exquisite three-dimensional (3D) order, as shown by X-ray diffraction of live muscle.6 The A-band is composed of an array of bipolar thick filaments and the overlapping part of the thin filaments. The thick filaments are composed of myosin molecules whose tail regions form the filament backbone and whose head regions emanate from the filament surface on a three-stranded quasi-helix with ~ 43 nm repeat. The myosin heads form “crowns” at ~ 14.3-nm axial intervals.7,8 In addition, there is in the A-band extra material that lies on a 43-nm axial repeat. In electron micrographs of A-bands in muscle sections9 and in isolated thick filament arrays (A-segments),10 the extra material produces prominent transverse stripes, 11 in all, starting at the edge of the bare zone. Immunolabelling of skeletal muscle showed that the 7–9 stripes distal from the M-line were contributed to by MyBP-C,11,12 the precise number depending on the muscle type. Although cardiac muscle shows accessory protein stripes in electron micrographs (e.g., Fig. 3), previous attempts to immunolabel cardiac muscle for electron microscopy have not been successful. In this study, we have been able to label cardiac muscle sufficiently strongly to determine the number of MyBP-C stripes and their precise location.
Skeletal muscle MyBP-C is composed of a sequence of 10 fibronectin III (Fn3) and immunoglobulin (Ig) domains (Fig. 1, C1–C10). Between C1 and C2 is a specific MyBP-C motif. The cardiac isoform of MyBP-C (cMyBP-C) has an extra Ig domain at the N-terminal (C0) and the cardiac sequence of the MyBP-C motif has three potential phosphorylation sites, which imply additional functions compared with the skeletal isoform.14 The C-terminal domains C8–C10 are involved in binding to the light meromyosin part of the myosin tail and to titin and the MyBP-C motif is thought to bind to myosin S2.15 In addition, intermolecular interactions between MyBP-C molecules themselves have been identified between domains 5 and 8 and 7 and 10, and this has led to a collar model for the arrangement of MyBP-C on the thick filament.16 An alternative, more axially oriented arrangement has been proposed by Squire et al.13
The development of a knockout (ko) mouse for cMyBP-C17 has enabled the investigation of the contribution of cMyBP-C to the thick filament structure. Two groups have studied the isolated filaments from the wild-type (wt) and ko muscles.2,3 Both show that the helical arrangement of the myosin heads in the ko filament is at least approximately maintained, although the heads are more prone to disorder. In addition, Zoghbi et al.3 identified a structure associated with the backbone in the wild-type filament that could be part of the MyBP-C molecule. However, the major part of the molecule was not located. These analyses do not identify interactions that may be stabilised by the muscle lattice. Therefore, we have investigated the contribution of MyBP-C to the accessory protein stripes in intact cardiac muscle as well as the effect of the loss of this protein on the axial distribution of the myosin heads in the cMyBP-C-null hearts. At the same time we have taken the opportunity to compare skeletal and cardiac muscle A-band structure in more detail than has been carried out previously. To do this, we have used electron micrographs of sections of well-preserved relaxed muscles in which the stripes and the axial distribution of myosin heads are clear. From these we have obtained the axial density distribution of the half-sarcomere. To increase the signal-to-noise ratio, we have developed a method to average several A-bands. A model for the contribution of MyBP-C in relation to the myosin head distribution is suggested. A preliminary report of this study has been presented.18
Previous attempts to label heart muscle with antibody to MyBP-C have been thwarted by the difficulty in getting the antibodies into the cells in sufficient concentration to be clearly seen in electron micrographs. The thick basal lamina inhibits the entry of the antibody. We circumvented this problem by using freshly prepared adult rat cardiomyocytes whose basal lamina is at least partly digested during preparation. In addition, a polyclonal antibody against cardiac MyBP-C17 with a very strong affinity was used and the labelling was enhanced with a secondary antibody to the first. Figure 2 shows electron micrographs of thin sections of adult rat cardiomyocytes labelled in this way; Fig. 2a shows a low-magnification overview of the labelling pattern in each half A-band. The enlarged sarcomeres in Fig. 2b show that the labelling occurs over nine stripes. The positions of these stripes correspond to those of the nine stripes (stripes 3 to 11) seen in rabbit soleus muscle12 (discussed later in relation to Fig. 4).
The axial distribution of MyBP-C and its relation to the myosin head array can be analyzed by averaging profile plots of the A-band. Figure 3 illustrates the procedure used for averaging profile plots of the A-band. Figure 3a shows an electron micrograph of a negative-stained cryosection of rat papillary muscle. The image shows clear M-line stripes and cross bridge bands. Several half A-band regions like the one boxed were selected; the profile plots were calculated and aligned by cross-correlation and the mean profile plot was calculated (Fig. 3b). After averaging, half A-band profile plots typically have well-defined symmetrical M-band profiles comprising three or five strong peaks and a marked end of the A-band (both indicated with a green line). The profile plot comprises distinct regular peaks in the cross bridge region. Starting from the edge of the bare zone, we can draw 11 equally spaced lines that coincide with the major peaks. The peaks correspond to the major stripes seen in the micrographs. They are labelled from 1 to 11 as in previous studies.12 We identify the range from stripes 3 to 11 as the cardiac C-zone. Hence, we refer to the remainder of the A-band, between stripes 1 and 3, as the P-zone (proximal) and the region distal from stripe 11 to the edge of the A-band as the D-zone (see Fig. 3 legend). Continuing the equally spaced lines distally in the D-zone takes us to line 17 just before the edge of the A-band. There are no obvious 43-nm features in this region. Electron microscopy studies of striated muscle are greatly aided by the fact that X-ray diffraction of live muscle gives a measure of the cross bridge periodicity present (43 nm).6 Using this value for internal calibration overcomes the problems of sample shrinkage due to fixation, dehydration and embedding and electron microscope radiation.19 Hence, the periodicity of the stripes is assumed to be 43 nm. In practice, we used the third order of the 43-nm reflection in the Fourier transform (see later), at 14.3 nm, for internal calibration, as it is much clearer.
Relaxed muscle samples were used in this study and it is important to note that MyBP-C stripes, which are quite labile, are only clearly seen in this state (authors' unpublished observations). Well-preserved samples show two sub-bands of ~ 14.3-nm periodicity within each 43-nm period. This is especially clear in the frog profile plot in Fig. 4c. We have identified these sub-bands as myosin head “crowns” by electron tomography of this sample (Luther et al., manuscript in preparation).
The average profile plot is ideal for measuring the length of the thick filament, a value not well known for cardiac muscle. In this study, we measure the length from the profile plot starting from the centre of the M-band to the A-band edge (half peak height). Our measurements show that the average A-band length in cardiac muscle [rat and wild type (wt) mouse] is 1.585 ± 0.011 μm.
The main results of this study are shown in Fig. 4, with Fig. 3 redrawn as Fig. 4d. Electron micrographs of the different samples are shown in the left panel. They are arranged so that the A-band sizes approximately match in each image. Each image in this panel was used along with others of the same sample to produce the average plot profiles shown in the right panel.
The plot profiles in the right panel are arranged to precisely line up the centre of the M-bands and the edges of the A-band as shown by the two continuous green lines drawn through the whole panel. Intervals of 43 nm, numbered 1 to 17, are drawn for each profile plot and they run continuously over the whole panel.
Figure 4a shows an electron micrograph of rabbit psoas muscle (a fast muscle) labelled with a polyclonal antibody against fast MyBP-C. This result was first shown by Bennett et al.;12 here we are re-evaluating the pattern in relation to the new data in this figure. The electron micrograph shows enhanced labelling over seven stripes from stripe 5 to 11 and hence identifies the C-zone in this muscle. Bennett et al. showed that slow muscle has a wider C-zone spanning nine stripes from 3 to 11.
Figure 4b shows the analysis for anti-cMyBP-C-labelled cardiac muscle from isolated rat cardiomyocytes. cMyBP-C is located at nine positions, from stripe 3 to 11. The positions of the outer seven labelled peaks match the positions of the peaks in the rabbit psoas (fast skeletal) muscle in (a). In Fig. 4b, the labelling at stripe 4 is located a little (~ 6 nm towards the Z-line) off the 43-nm banding pattern for all the other stripes. We have frequently observed weaker density and slightly variable location at stripe 4 in unlabelled cardiac and skeletal muscles.
Figure 4c shows the plot profile for fast skeletal muscle (frog sartorius). The plot is particularly clear, as this sample had the best preparation technique in this study (fast freezing and freeze substitution). The antibody labelling in (a) identifies the C-zone between stripes 5 and 11. Of special note here is that the native stripes in this muscle match precisely with the anti-MyBP-C peaks in Fig. 4a. This is an important result, as it is consistent with the conclusion that most of the MyBP-C molecule is located at the native 43-nm stripes. Between each pair of the 43-nm stripes in the C-zone are two minor peaks. We show elsewhere that these two minor peaks are due to the myosin cross bridge crowns, which we label crowns 2 and 3, with crown 1 being located at the 43-nm stripe (Luther et al., unpublished data). The A-band length in this muscle is 1.578 ± 0.007 μm.
The analysis for rat cardiac muscle cryosections is shown in Fig. 4d. The native peaks in this muscle up to stripe 11 match precisely the peaks in skeletal muscle (c). With the match of the M-line centre and A-band edge, this shows that cardiac and skeletal muscles have similar structure and organisation at this level. As in skeletal muscle, the 43-nm stripes coincide with the antibody-labelled positions for cMyBP-C (b), showing further similarity between skeletal and cardiac muscle.
An electron micrograph of a cryosection of mouse wt papillary muscle is shown in Fig. 4e. This figure has lower contrast overall and lower density in the D-zone, probably due to the staining method used (ammonium molybdate). However, when averaged over a large area, the C-zone shows peaks at locations that match the rat cardiac sample.
In Fig. 4f, we show the analysis for cryosections of mouse cardiac MyBP-C-ko muscle. Of the main stripes in the profile plot, only the first and second 43-nm stripes have strong peaks. In the C-zone, the major stripe values are greatly diminished and this occurs for all of the peaks 3 to 11. Hence, the densities at the positions identified as cMyBP-C location by immunolabelling (Fig. 4b), viz., stripes 3 to 11, are greatly reduced in MyBP-C-deficient mice.
The sub-banding attributed to myosin crowns is also seen in the rodent cardiac profiles. In Fig. 4c–f, C1, C2 and C3 identify examples of three peaks within a 43-nm period. It is important to note that C1 is much stronger than C2 and C3 in frog skeletal muscles and in the cardiac muscles. However, in the knockout, C1 is much weaker than C2 and C3, suggesting that the heads at this crown may be considerably disordered.
The mean length of the A-band in both cryosections and plastic sections (data not shown) in mouse wt cardiac and mouse cMyBP-C-ko muscle was found to be 1.584 ± 0.011 and 1.574 ± 0.008 μm, respectively. Hence, the A-band length is not affected by the ablation of cMyBP-C.
One-dimensional Fourier transforms of the mean profile plots of half A-bands (excluding M-band and starting at stripe 1 and ending just before stripe 17) are shown in Fig. 5. The purpose of this figure is to illustrate the so-called forbidden meridional reflections, at the first and second orders of 43 nm− 1 that occur in the X-ray diffraction patterns of muscle. They could arise from periodic deviations of the myosin heads from the axial repeat of 14.3 nm, from the contribution of MyBP-C or a combination of the two. The Fourier transforms for relaxed skeletal, rat and mouse wt cardiac muscles are shown in Fig. 5a, b and c, respectively. In these transforms, the reflections corresponding to 43 and 21.5 nm occur prominently as they do along the meridian in the X-ray patterns. The Fourier transforms of the cMyBP-C-ko muscle cryosection in Fig. 5d is dramatically different. The 21.5-nm reflection is completely absent and the 43-nm peak is greatly diminished, indicating that MyBP-C makes a major contribution to these reflections.
Also seen clearly in some of these transforms is the peak corresponding to 38.5 nm due to troponin on the thin filaments. This reflection is seen in both the mouse transforms (Fig. 5c and d) and suggests that the sarcomere length in all the areas used for analysis were rather similar so that the troponin repeat was also reinforced. On the other hand, this reflection is not seen in the frog transform (a) and only weakly in that from the rat (b). This can be explained if the sarcomere length varied in the areas used for analysis and the troponin repeat is therefore averaged out.
The presence in the Fourier transforms from the mouse cardiac samples of a strong reflection corresponding to 38.5 nm arising from the troponin repeat on the thin filaments suggests that the troponin mass may interfere with the interpretation of the myosin head positions especially in the MBP-C knock out. To decrease the contribution of the troponin stripes, we have averaged the different 43-nm repeats in the C-region. Since this repeat is ~ 4–5 nm longer than that of troponin, the troponin contribution will occur at different positions within each repeat and thence over several repeats will be averaged out. The positions of the nine MyBP-C stripes, 3–11, were determined and the eight repeats within that distance averaged. The centre of the sixth and seventh stripes, identified by Zoghbi et al.3 to be at 288 and 331 nm, are within a pixel of the start of the repeat defined here. The plots in Fig. 6 show one averaged repeat from essentially the centre of the MyBP-C peaks for frog and the three cardiac samples, rat and mouse wt and the ko mouse. The data from the frog are plotted displaced from the other three for clarity. All four plots show three peaks indicative of the three crowns of myosin heads. Rat and wt mouse plots are very similar, with the major peak at the origin in the MyBP-C position. The ko has three peaks in positions similar to that of the wt but the peak at the origin is very much reduced, confirming that the loss of stripe density corresponds to the absence of MyBP-C. It is also much weaker than the other two, further suggesting more disorder in this crown.
Averaging the repeats does not change the conclusions reached by looking at the raw density plots, suggesting that the troponin mass does not grossly affect the density distribution in general, but it does reinforce the conclusions reached, especially in the case of the ko where the density pattern of some of the repeats was more variable. In particular, one crown of myosin heads is hidden under the mass of the MyBP-C. Using the convention of Al-Khayat et al.,4 we label the three crowns of myosin heads 1, 2 and 3 from the M-line to the Z-line. Crown 1 is at the level of the MyBP-C stripe. Between the stripes there are two levels of heads. The one on the left, crown 2, is also partly subsumed in the MyBP-C mass, while crown 3 is a clear peak. MyBP-C seems to be associated axially with the space between crown 1 and 2 in the cardiac muscle. In frog, there is a similar arrangement, although the mass is more narrowly confined to crown 1.
The position of the myosin heads in the wt A-band cannot be measured with any certainty, since the presence of MyBP-C distorts their position. The position of the three individual peaks, seen in the plot from the ko, allows us to estimate the position and axial separation of the myosin heads. Using a crude measure of the positions of the peak density, we find that the separation between crowns differs. Crowns 2 and 3 are closer together (peak separation ~ 12 nm) than 1 and 3 (~ 13 nm), while 2 and 1 are the furthest separated (~ 18 nm).
In this study, the cardiac muscles were fixed in glutaraldehyde, cryoprotected and then cryosectioned. We have shown that this method preserves detail to nearly 4 nm.21 However, plot profiles of individual A-bands are noisy. We have developed a novel method of averaging several half A-bands by cross-correlation. This gives stable plots with typical M-band profiles, A-band ends and C-zone stripe patterns. One possible limitation is the accuracy of the side-by-side alignment of the filaments inherent in the samples that we have used. However, following careful selection of well-aligned areas, the detail in the density plots suggests that this is not a problem. Hence, regarding the method of sample preparation, we can conclude that the cross-links formed during glutaraldehyde fixation cause only small random changes and our method of averaging of different images improves the visibility of structural detail. For frog skeletal muscle, this chemical fixation step was avoided as rapid freezing/freeze substitution methods were used. This gives excellent preservation and even single A-bands gave good plot profiles.
We have compared the different samples in this study by precisely aligning the mean A-band profile plots. These profile plots shows that the A-bands and thick filaments are very similar in skeletal and cardiac muscle, with mean lengths of 1.578 ± 0.007 and 1.585 ± 0.011 μm, respectively. The mean length for both types was 1.58 ± 0.011 μm, which is close to the values found by Sjostrom and Squire9 and Page22 but a little different from the value of 1.63 μm found by Sosa et al.23 and Granzier et al.24 The P and C zones enclosed between stripes 1 and 11 are highly ordered and very similar in both.
This study shows that the length of the thick filament (A-band length) is not affected by the lack of MyBP-C, as the values in wt and ko are very similar (1.584 ± 0.011 and 1.574 ± 0.008 μm, respectively). Early work showed that synthetic thick filaments formed in the presence of physiological amounts of MyBP-C had more stable diameter and length than those formed in non-physiological amounts.25 It is thought that titin has the main role in determining thick filament length; our observations show that the interaction of MyBP-C with titin is not necessary to fulfil this role.
Here we discuss the axial structure of the averaged A-band profile plots in Fig. 4 in relation to the M-band, P-zone and D-zone. The M-band is very similar in rat and wt mouse cardiac muscle, comprising a five-line M-band as described by Pask et al.26 The cMyBP-C ko sample also has a five-line M-band (Fig. 4f), but the outer two lines are very weak. The reduction of the outer two may be due to the dynamical nature of the M-band.27 The same comment applies to the frog skeletal M-band in Fig. 4c, in which there are weak peaks outside the central one. The P-zone has strong peaks at stripes 1 and 2 in rat and mouse; the non-myosin cardiac protein located at these stripes has yet to be identified. The D-zone appears quite variable in individual samples. The consistent feature in the D-zone in all the averaged plots is a 14.3-nm periodicity due to the cross bridge repeat and the absence of the 43-nm periodicity stripes (data not shown).
Bennett et al. showed by antibody labelling that the number of MyBP-C locations in the A-band varied according to the muscle, between seven in fast rabbit psoas (stripes 5–11) and nine in slow rabbit soleus muscle (stripes 3–11).12 Furthermore, there were different isoforms and MyBP-C-related proteins such as MyBP-H, which filled some of the gaps. In heart muscle, it is known that there is only one cardiac isoform, cMyBP-C, and that in the cMyBP-C null mouse, other isoforms are not expressed to substitute for it.17 On this basis, we might expect that there are nine MyBP-C stripes in the heart. We have shown by immunolabelling that this is indeed the case and have unequivocally identified the location of cMyBP-C in cardiac muscle to be positions 3 to 11. The binding of MyBP-C to the thick filament is known to depend on titin and the myosin tail. Rabbit soleus muscle and heart both operate with slow myosin isoforms. Possibly, this is one of the factors that determines that the same arrangement of MyBP-C is found in both muscle types.
One slight proviso arises from the immunolabelling. One of the stripes, number 4, was sometimes weaker than the others. This was reflected in the more variable nature of this stripe in the unlabelled muscles. It is possible that other as yet unknown accessory proteins, such as are present at stripe 1 and 2, contribute to the MyBP-C position 4 in cardiac muscle. However, MyBP-C is a major contributor to the stripe density. We have shown this by comparing the fine structure of the A-bands in mouse MyBP-C-ko cardiac muscle with that in wt mouse, rat cardiac and frog skeletal muscle. The ordered arrangement of components along the sarcomere allows a detailed 1D analysis of the structures. We showed that the nine clear stripes present in wild-type cardiac muscle and enhanced by immunolabelling were absent from the MyBP-C-ko muscle.
The analysis carried out here allows us to look at the relationship of MyBP-C and myosin in the intact muscle so that relationships depending on the filament lattice are not lost. Within each 43-nm period in the C-zone, there are three crowns of cross bridges with separation of approximately 14.3 nm.3,8 Two of these are seen in our profile plots and in the averaged repeats as the two sub-bands between any two C-zone stripe peaks (Figs. 4 and 6). This is particularly clear in the frog (Fig. 4c). The third crown is subsumed under the peak of MyBP-C density. In the frog, the stripe density is relatively symmetrical about crown 1, suggesting that the bulk of the MyBP-C is at this level. In the rodent heart muscle it is not symmetrical. Crown 2 has become a shoulder on the major peak density, indicating that the MyBP-C, while having a major contribution at crown 1, also contributes density to the space between crowns 1 and 2. This interpretation is supported by our observation that this density is lost in the MyBP-C null filaments leaving only weak crown 1 density. This also supports original arguments that MyBP-C should lie approximately perpendicular to the filament axis because of the sharp stripes of the extra mass in the A-band and in A-segments as well as the clearly separated antibody stripes. We would therefore favour a model in which the bulk of MyBP-C is arranged around or extending from the thick filament at more or less the same level as crown 1. At greater than 32 nm in length, it would be able to reach to neighbouring thick or thin filaments. A relatively short region of the molecule extending along the filament is compatible with this view.3,13
We can compare our results with those on single filaments from cardiac muscle where changes due to removing the filaments from the muscle lattice environment cannot be controlled. Axial density distributions have been calculated for the 43-nm repeat in the MyBP-C region in rabbit cardiac filaments by single-particle analysis4 and from filtered images of cardiac muscle filaments from wt and ko mice.2 Again they both suggest that the stripe material is mostly associated with crown 1 but that some of the mass is located between 1 and 2 so that the crown 2 is seen as a shoulder on the main peak. In mouse filaments in the absence of MyBP-C, density at both crown 1 level and the intermediate peak is lost. Since these images are from filtered or averaged data sets, it is to be expected that only the ordered structure will be revealed. It is therefore difficult to know what of the crown 1 mass is due to C-protein and what proportion of the change may be due to loss of order of the myosin heads (see later).
In their reconstruction of the C-zone of the mouse cardiac filament by single-particle analysis, Zoghbi et al.3 identified a structure that is arguably part of the C-protein molecule. This is three 4-nm domains lying axially close to the filament backbone and titin. These three domains could be C8–C10, which have been found in solution studies to bind to the myosin rod and part of the 11-domain titin repeat (Fig. 1). It is unlikely that these three domains (molecular mass ~ 30 kDa) would contribute significantly to the strong 43-nm stripes. Furthermore, it is very close axially to the position of the myosin heads on crown 1 so it does not correlate with the mass between crowns 1 and 2. This leaves eight domains of the molecule to be accounted for. It seems likely that these domains contribute to our 43-nm stripe density.
The organisation of MyBP-C has some of the quality and elusiveness of the M-line proteins. These proteins are about the same molecular weight and are also composed of Ig and Fn3 domains and yet they are only seen in the electron microscope under some conditions. Their visualisation may be favoured by superimposition of different layers in a thicker section and by the presence of other proteins, such as the M-band part of titin and creatine kinase, which bind to myomesin and M-protein27 and enable the M-line bridges to be seen. It is possible that C-protein is also revealed in the same way.
The myosin filament is considered to form a three-stranded quasi-helix with a mean 14.3-nm axial cross bridge spacing and a 42.9-nm helix repeat. In X-ray diffraction patterns, the expected reflections occur on the meridian corresponding to the third, sixth, etc. orders of the 42.9-nm cross bridge repeat. However, there are additional meridional reflections that should not occur for a pure helical structure, that correspond to the first, second, fourth, etc. orders of the 42.9-nm cross bridge repeat, the so-called forbidden meridionals.2,6,8,28 Accounting for these forbidden meridionals has been the quest of many studies.28,29 We have considered the first and second forbidden meridionals in this study and found that in MyBP-C-ko muscles they are weak or absent, respectively. This was also found in a recent study on isolated filaments.2,3 Forbidden meridionals can arise from enhanced 43-nm periodicity (such as the dense 43-nm stripes labelled by MyBP-C) or from perturbation of the 14.3-nm cross bridge periodicity (unequal spacing of the three crowns in each 43-nm period) or from a combination of the two. In order to understand the origin of the forbidden reflections, we require the detailed structure of the thick filament to determine the relative contributions of MyBP-C labelling and variable crown spacing. Our opinion is that it is MyBP-C labelling that is the main cause of the forbidden meridionals.
It will be interesting to see whether there is a reduction or absence of the forbidden meridionals in hypertrophic cardiomyopathy samples that arise from mutations in cMyBP-C. This is because the majority of cMyBP-C mutations result in truncation of the myosin-binding region, and hence may be effective knockout samples.15
In the presence of MyBP-C it is not possible to determine precisely the axial position of the myosin heads from the axial density distribution. However, several studies of reconstructions of muscle thick filaments, both heart and skeletal, where the positions of the myosin heads can be identified, have indicated that there are axial deviations of the myosin heads from a strict 14.3-nm repeat. However, different filaments seem to give different results. In rabbit heart filaments, the distance between crowns 1 and 2 is the smallest (~ 13 nm) and between 2 and 3 the largest (~ 15 nm),2,4 whereas in fish skeletal filaments the 1–2 separation was found to be 15 nm and the 2–3 13.5 nm.8 In mouse heart, the situation is different again with the 2–3 distance being the shortest and the 1–3 the longest.3 (Note that in the Zoghbi et al. paper itself the crowns are numbered from the opposite direction, i.e., from the end of the filament. Crown 1 is still associated with the MyBP-C mass.) In the density profile of frog skeletal muscle filaments (Fig. 6), the two crowns between the C-protein stripes are clearly seen and they are separated by less than 14.3 nm, which could be in agreement with either the fish or the mouse heart data. Whether these differences are due to different methods of analysis and the unknown contribution of MyBP-C or is a true difference due to muscle type (cardiac or skeletal) or species is not clear.
In the absence of MyBP-C, there are two sub-bands in several 43-nm repeats in the profile plot for MyBP-C-ko muscle (Fig. 4f). This indicates that crowns 2 and 3 are axially ordered. As expected, the peak at each stripe position is absent because of lack of MyBP-C. We would expect a diminished peak equal in size to crowns 2 and 3, but it is much lower. Hence, the lack of MyBP-C at crown 1 apparently results in disordering of the heads at this level. This is borne out in the averaged repeat (Fig. 6) and is in agreement with the results of Zoghbi et al.3 We also found that there is an apparent difference in crown spacing in the ko similar to that seen in the wt and ko filaments by Zoghbi et al.3
Kensler and Harris2 and Zoghbi et al.3 have remarked on the greater sensitivity and loss of order exhibited by the filaments from the ko mouse hearts. We found it more difficult to obtain well-ordered cryosections but we were able to find good regions to analyse that exhibited the 14.3-nm repeat. Clearly then, C-protein contributes to the stability of the filament and of the sarcomere, an observation that is in line with the functional deficits and cardiac hypertrophy found in hearts of cMyBP-C-ko mice.17,30
For electron microscopy of striated muscles, one of the best methods for preserving the fine structure is the use of cryosections.9,21 We have shown that this method preserves the axial order to about 4 nm.21 The cryosections were prepared as follows. Cardiac muscle was dissected under Krebs solution with 30 mM 2,3-butanedione monoxime (BDM) added (inhibiting actomyosin interaction and thus preventing dissection-induced contracture)31 and pinned on Sylgard in a Petri dish. The only effect we observed due to BDM was improvement in orientation of the sections resulting from relaxation of the myocardium. The muscle was aerated for 30 min with 95% oxygen/5%carbon dioxide. Then the solution was replaced with Krebs (rat cardiac) and the aeration continued. The sample was fixed for 1 h with 3% glutaraldehyde in Krebs solution. A cryoprotectant solution was prepared composed of 2.3 M sucrose in Krebs. The fixed muscles were immersed in the sucrose solution for a few days in a refrigerator and then small, 0.5- to 1-mm-cube pieces were cut. The pieces were mounted on a cryopin, excess sucrose was blotted off and the pieces were frozen by plunging into liquid nitrogen. Cryosections ~ 100 nm thick were cut with an RMC MT7 ultramicrotome fitted with a CR20 cryoattachment‡, transferred to formvar/carbon-coated nickel grids, floated on water and then negative-stained with 2% uranyl acetate (rat cardiac), 4% uranyl acetate with 0.1% trehalose32 (mouse cardiac cMyBP-C-ko) or 2% ammonium molybdate (mouse cardiac wt). Species used were as follows: rat, Sprague–Dawley; wt mouse, cba/b16; cMyBP-ko mouse, SV/129.
For preparing plastic sections of muscle for electron microscopy, the best preservation is obtained by fast freezing/freeze substitution with subsequent embedding in resin.23,33,34 Frog muscle skinned in relaxing solution was fast-frozen by slamming onto a liquid-helium-cooled copper block, freeze-substituted in acetone/1% tannic acid and embedded in Epon.33–35 For each sample, thin ~ 100-nm sections were cut with a Reichert Ultracut-E and stained with 2% uranyl acetate and Reynolds lead citrate. At this thickness, there are about 2.5 unit cells of the ~ 40-nm myosin filament lattice within the depth of the section.
Rabbit psoas muscle was labelled with antibody against MyBP-C as described earlier12 and briefly here. Thin fibre bundles were excised from demembranated muscle, soaked in antibody solution for 72 h, fixed in 2.5% glutaraldehyde followed by 1% osmium, dehydrated in ethanol series and embedded in Araldite. Thin sections were cut as above.
Adult rat cardiomyocytes were a kind gift from Friederike Cuello. To fresh cells in Hepes-buffered Tyrode solution, 20 mM BDM was added to relax the cells for 5 min before they were spun into a soft pellet and taken up in phosphate-buffered saline (PBS) plus 20 mM BDM. They were then spun again and taken up in 4% paraformaldehyde in PBS for 10 min. Cells were washed, permeabilised in 0.2% triton in PBS for 5 min, washed and then transferred to antibody solution. Rabbit anti-rat cardiac MyBP-C polyclonal antibody was used.17 Myocytes were incubated in a small volume of concentrated antibody (1:10) overnight while rotating at 4 °C. The sample was washed, then part of the sample was subsequently incubated in goat anti-rabbit immunoglobulins at 0.1 mg/ml for 2 h. Samples were pelleted in 2% gelatine in PBS, the excess gelatine was removed and pellets were prepared for electron microscopy by standard methods. Thin sections were viewed in a Hitachi 7600 electron microscope.
Suitable regions were recorded on electron microscope film (Kodak SO-163) at 20,000× nominal magnification and the films scanned with a Nikon Coolscan 8000ED 4000 dpi scanner and binned four times to give a pixel resolution of about 1.3 nm. Images for anti-cMyBP-C-labelled cardiac muscle and mouse wt cardiac muscle were recorded directly onto a CCD camera (Tietz Fastscan F114 camera).
Longitudinal sections of striated muscle show transverse stripes from the different components such as the M-band and the Z-band. The appearance of the transverse stripes can be enhanced by integrating the density in scanned images along the stripes and then displaying as a plot profile. For this study, comparisons were made between plot profiles of half A-bands of the different samples. The plot profile of a single image can be quite noisy. We have used a novel method for averaging several sarcomeres. Using ImageJ§, selected regions were first rotated to make the cross bridge striations precisely vertical. For this, the Fourier transform was calculated and the angle for rotation found from the meridional spots corresponding to the cross bridge periodicity (43 nm and/or orders thereof). Then, half A-bands that included regions with preferably straight and prominent M-bands and cross bridge striations and strong A-band edges were selected, band-pass filtered for very low and high periodicities and their profile plots calculated and the values saved. Suitable leftward half A-bands were first flipped about a vertical axis and then treated as above. The saved plot 1D values were then cross-correlated with the use of a custom-written program and summed after weighting with the width of the selected region. Two rounds of averaging were done. For the second round, the mean profile plot found from the first round was used as the reference for cross-correlation.
For each plot shown, we estimate the total number of thick filaments that have contributed to it. An electron microscope image is formed as a projection of the density through the depth of a sample. Hence, for the ~ 100-nm-thick sections used in this study, there are about 2.5 filaments in the depth of the section as the unit cell size is ~ 40 nm. For each image, we can calculate the total number of filaments included in the 3D volume from the width of the selected regions, lattice size (40 nm) and the section thickness (~ 100 nm thick, hence, 2.5 unit cells in depth). Typically, several regions totalling at least 1.5-μm width, hence ~ 94 filaments, were used to produce the final plot profile (e.g., 1500 × 100/(40 × 40) = ~ 94).
We are grateful to John Squire, Steven Marston and Jon Kentish for helpful discussions. We thank the following: Theo Floyd for expert analysis of the frog data during his undergraduate project, Cathy Timson for expert technical help, Dan Fitzsimmons for help with the experiments conducted in Madison, Leon de Windt for the provision of a wild strain mouse. We are grateful to the referees for constructive comments. This work was supported by the British Heart Foundation grant PG/06/010 (to P.K.L.) and in part by NIH grant AR34711 (to R.C.) and HL82900 (to R.M.).