Abstract. The architecture of the junctional sarcoplasmic reticulum (SR) and transverse tubule (T tubule) membranes and the morphology of the two major proteins isolated from these membranes, the ryanodine receptor (or foot protein) and the dihydropyridine receptor, have been examined in detail. Evidence for a direct interaction between the foot protein and a protein component of the junctional T tubule membrane is presented. Comparisons between freeze-fracture images of the junctional SR and rotary-shadowed images of isolated triads and of the isolated foot protein, show that the foot protein has two domains. One is the large hydrophilic foot which spans the junctional gap and is composed of four subunits. The other is a hydrophobic domain which presumably forms the SR Ca2+-release channel and which also has a fourfold symmetry. Freeze-fracture images of the junctional T tubule membranes demonstrate the presence of diamondshaped clusters of particles that correspond exactly in position to the subunits of the feet protein. These results suggest the presence of a large junctional complex spanning the two junctional membranes and intervening gap. This junctional complex is an ideal candidate for a mechanical coupling hypothesis of excitation-contraction coupling at the triadic junction.

No abstract

To probe the physiological role of calsequestrin in excitation-contraction coupling, transgenic mice overexpressing cardiac calsequestrin were developed. Transgenic mice exhibited 10-fold higher levels of calsequestrin in myocardium and survived into adulthood, but had severe cardiac hypertrophy, with a twofold increase in heart mass and cell size. In whole cell-clamped transgenic myocytes, Ca2+ channel- gated Ca2+ release from the sarcoplasmic reticulum was strongly suppressed, the frequency of occurrence of spontaneous or Ca2+ current-triggered "Ca2+ sparks" was reduced, and the spark perimeter was less defined. In sharp contrast, caffeine-induced Ca2+ transients and the resultant Na+-Ca2+ exchanger currents were increased 10-fold in transgenic myocytes, directly implicating calsequestrin as the source of the contractile-dependent pool of Ca2+. Interestingly, the proteins involved in the Ca2+-release cascade (ryanodine receptor, junctin, and triadin) were downregulated, whereas Ca2+-uptake proteins (Ca2+-ATPase and phospholamban) were unchanged or slightly increased. The parallel increase in the pool of releasable Ca2+ with overexpression of calsequestrin and subsequent impairment of physiological Ca2+ release mechanism show for the first time that calsequestrin is both a storage and a regulatory protein in the cardiac muscle Ca2+-signaling cascade. Cardiac hypertrophy in these mice may provide a novel model to investigate the molecular determinants of heart failure.

We have examined the structure of calsequestrin in three-dimensional images from deep-etched rotary-replicated freeze fractures of skeletal muscle fibers. We selected a fast-acting muscle because the sarcoplasmic reticulum has an orderly disposition and is rich in internal membranes. Calsequestrin forms a network in the center of the terminal cisternae and is anchored to the sarcoplasmic reticulum membrane, with preference for the junctional portion. The anchorage is responsible for maintaining calsequestrin in the region of the sarcoplasmic reticulum close to the calcium-release channels, and it corroborates the finding that calsequestrin and the spanning protein of the junctional feet may interact with each other in the junctional membrane. Anchoring filaments may be composed of a protein other than calsequestrin.

Isolated heavy sarcoplasmic reticulum vesicles retain junctional specializations (feet) on their outer surface. We have obtained en face three-dimensional views of the feet by shadowing and replicating the surfaces of freeze-dried isolated vesicles. Feet are clearly visible as large structures located on raised platforms. New details of foot structure include a four subunit structure and the fact that adjacent feet do not abut directly corner to corner but are offset by half a subunit. Feet aligned within rows were observed to be rotated at a slight angle off the long axis of the row creating a center-to-center spacing (32.5 nm) slightly less than the average diagonal of the feet (35.3 nm). Comparison with previous information from thin sections and freeze-fracture showed that this approach to the study of membranes faithfully preserves structure and allows better visualization of surface details than either thinsectioning or negative-staining.