Abnormal formation of sarcoplasmic reticulum networks and triads during early development of skeletal muscle cells in mitsugumin29-deficient mice

Recently, we detected a novel membrane protein, mitsugumin29 (MG29), in the triads in rabbit skeletal muscle cells and suggested important roles for this membrane protein in the formation of the sarcoplasmic reticulum (SR) networks and triads in muscle cells. In the present study, we examined the development of skeletal muscle cells in MG29-deficient mice to try to determine the roles played by MG29 in the formation of the SR networks and triads. Ultrastructural observations revealed some morphological abnormalities in these mice, such as incomplete formation of the SR networks, an irregular running of the transverse tubule and a partial defect in the triads at the A-I junctional region. These ultrastructural abnormalities occurred during early myogenesis and were preserved until the adult stage. The possible roles for MG29 in the formation of SR networks and triads in skeletal muscle cells are discussed in the light of these observations.     

Identification and characterization of the high affinity [3H]ryanodine receptor of the junctional sarcoplasmic reticulum Ca2+ release channel

The high affinity ryanodine receptor of the Ca2+ release channel from junctional sarcoplasmic reticulum of rabbit skeletal muscle has been identified and characterized using monoclonal antibodies. Anti-ryanodine receptor monoclonal antibody XA7 specifically immunoprecipitated [3H]ryanodine-labeled receptor from digitonin-solubilized triads in a dose-dependent manner. [3H]Ryanodine binding to the immunoprecipitated receptor from unlabeled digitonin-solubilized triads was specific, Ca2+-dependent, stimulated by millimolar ATP, and inhibited by micromolar ruthenium red. Indirect immunoperoxidase staining of nitrocellulose blots of various skeletal muscle membrane fractions has demonstrated that anti-ryanodine receptor monoclonal antibody XA7 recognizes a high molecular weight protein (approximately 350,000 Da) which is enriched in isolated triads but absent from light sarcoplasmic reticulum vesicles and transverse tubular membrane vesicles. Thus, our results demonstrate that monoclonal antibodies to the approximately 350,000-Da junctional sarcoplasmic reticulum protein immunoprecipitated the ryanodine receptor with properties identical to those expected for the ryanodine receptor of the Ca2+ release channel.     

Acetylcholinesterase in membrane fractions derived from sarcotubular system of skeletal muscle: presence of monomeric acetylcholinesterase in sarcoplasmic reticulum and transverse tubule membranes

Microsomes were isolated from white rabbit muscle and separated into several fractions by centrifugation in a discontinuous sucrose density gradient. Four membrane fractions were obtained namely surface membrane, light, intermediate and heavy sarcoplasmic reticulum. The origin of these microsomal vesicles was investigated by studying biochemical markers of sarcoplasmic reticulum and surface and T-tubular membranes. The transverse tubule derived membranes were further purified by using a discontinuous sucrose density gradient after loading contaminating light sarcoplasmic reticulum vesicles with calcium phosphate in the presence of ATP. All membrane preparations displayed acetylcholinesterase activity (AChE, EC 3.1.1.7), this being relatively more concentrated in T-tubule membranes than in those derived from sarcoplasmic reticulum. The membrane-bound AChE of unfractioned microsomes notably increased its activity by aging, treatment with detergents and low trypsin concentrations indicating that the enzyme is probably attached to the membrane in an occluded form, the unconstrained enzyme displaying higher activity than the vesicular acetylcholinesterase.Sedimentation analysis of Triton-solubilized AChE from different membrane fractions revealed enzymic multiple forms of 13.5S, 9–10S and 4.5–4.8S, the lightest form being the predominant one in all membrane preparations. Therefore, in both sarcoplasmic reticulum and T-tubule membrane the major component of AChE appears to be a membrane-bound component, probably a G₁ form.     

Evidence for the association of dystrophin with the transverse tubular system in skeletal muscle

Polyclonal antibodies to dystrophin (the protein product of the human Duchenne muscular dystrophy gene) were used to identify and characterize dystrophin in isolated triads from rabbit skeletal muscle. Anti-dystrophin antibodies recognize an approximately 400,000-Da protein in isolated triads or heavy microsomes from skeletal muscle. Treatment of heavy microsomes with buffers containing high salt or EDTA to remove peripheral or extrinsic membrane proteins does not remove dystrophin; however, treatment of intact triads with trypsin shows that dystrophin is extremely sensitive to mild proteolytic digestion. Isolation of junctional complexes from skeletal muscle triads indicates that dystrophin is tightly associated with the triadic junction. Fractionation of the triadic junction into junctional transverse tubular membranes and junctional sarcoplasmic reticulum membranes has shown that dystrophin is enriched in junctional transverse tubular membranes. Thus, our results suggest that dystrophin is a component of the triad junction which is exposed to the cytoplasm and embedded in or attached to the transverse tubular membrane.     

T-tubule depolarization-induced SR Ca2+ release is controlled by dihydropyridine receptor- and Ca(2+)-dependent mechanisms in cell homogenates from rabbit skeletal muscle

In vertebrate skeletal muscle, the voltage-dependent mechanism of rapid sarcoplasmic reticulum (SR) Ca2+ release, commonly referred to as excitation-contraction (EC) coupling, is believed to be mediated by physical interaction between the transverse (T)-tubule voltage-sensing dihydropyridine receptor (DHPR) and the SR ryanodine receptor (RyR)/Ca2+ release channel. In this study, differential T-tubule and SR membrane monovalent ion permeabilities were exploited with the use of an ion-replacement protocol to study T-tubule depolarization-induced SR 45Ca2+ release from rabbit skeletal muscle whole-cell homogenates. Specificity of Ca2+ release was ascertained with the use of the DHPR antagonists D888, nifedipine and PN200-110. In the presence of the "slow" complexing Ca2+ buffer EGTA, homogenates exhibited T-tubule depolarization-induced Ca2+ release comprised of an initial rapid phase followed by a slower release phase. During the rapid phase, approximately 20% of the total sequestered Ca2+ (approximately 30 nmol 45Ca2+/mg protein), corresponding to 100% of the caffeine-sensitive Ca2+ pool, was released within 50 ms. Rapid release could be inhibited fourfold by D888. Addition to release media of the "fast" complexing Ca2+ buffer BAPTA, at concentrations > or = 4 mM, nearly abolished rapid Ca2+ release, suggesting that most was Ca2+ dependent. Addition of millimolar concentrations of either Ca2+ or Mg2+ also greatly reduced rapid Ca2+ release. These results show that T-tubule depolarization-induced SR Ca2+ release from rabbit skeletal muscle homogenates is controlled by T-tubule membrane potential- and by Ca(2+)- dependent mechanisms.     

Separation of cardiac plasmalemma into cell surface and T-tubular components. Distribution of saxitoxin- and nitrendipine-binding sites

To compare surface sarcolemmal with T-tubular distributions of [3H]saxitoxin (STX)- and [3H]nitrendipine (NTD)-binding sites, we centrifuged membrane vesicles from sheep and bovine ventricles on a 10-40% linear sucrose gradient from which fractions were assayed for STX and NTD binding; for markers of surface sarcolemma (ouabain-sensitive Na,K-ATPase activity, [3H]quinuclidinyl benzilate binding); and for markers of junctional sarcoplasmic reticulum known to be preferentially associated with T-tubules (ryanodine-sensitive Ca2+ uptake, calsequestrin, an Mr 300,000 putative phosphorylatable "foot" protein, and electron microscopically visible junctional sarcoplasmic reticulum-plasmalemma complexes). We identified three distinct peaks in the sucrose gradient, each characterized by significant high and low affinity STX- and high affinity NTD-binding: Peak I (approximately 19% sucrose), highly enriched in surface sarcolemma; Peak III (approximately 36% sucrose), enriched in junctional sarcoplasmic reticulum markers and hence in junctional sarcoplasmic reticulum complexes with T-tubule; and Peak II (approximately 27% sucrose), showing greatest specific STX binding and only moderate NTD binding, enriched in T-tubular membrane, unassociated with junctional sarcoplasmic reticulum. For ventricular myocytes, the ratio NTD sites/STX sites was 2.5 for surface sarcolemma, but only approximately 1.0 for T-tubules. Unlike data published for mammalian skeletal muscle, sheep and beef cardiac NTD receptors were not significantly more concentrated in T-tubular than in surface plasmalemma.     

Rapid calcium release from the isolated sarcoplasmic reticulum is triggered via the attached transverse tubular system

Rapid replacement of 0.15 M K gluconate with 0.15 M choline Cl led to multiphasic Ca2+ release from a heavy fraction of rabbit skeletal muscle microsomes. Following the initial lag period (0-50 ms), about 15 nmol of Ca2+/mg of protein was rapidly released with first-order rate constants k = 60-140 s-1. Subsequently, a larger amount of Ca2+ (up to 56 nmol/mg) was released at a slower rate (k = 0.8-1.5 s-1). The Ca2+ released in both rapid and slow phases was reaccumulated within 60 s. In agreement with a previous report (Caswell, A. H., Lau, Y. H., Garcia, M., and Brunschwig, J-P. (1979) J. Biol. Chem. 254, 202-208), French press treatment of the tubule/sarcoplasmic reticulum (SR) complex results in dissociation of transverse tubular membrane (T-tubules) from SR. Subsequent incubation with 0.4 M potassium cacodylate results in the reassociation of the complex, as shown by sucrose density-gradient sedimentation. Upon T-tubule dissociation, both rapid and slow Ca2+ release was inhibited. Upon reassociation, the rapid Ca2+ release was completely restored and the slow phase partially restored. The results indicate that the T-tubule associated with SR plays a crucial role in triggering rapid Ca2+ release induced by ionic replacement. Other types of Ca2+ release, e.g. those induced by Ca2+ alone or with drugs such as caffeine and quercetin, are unaffected by T-tubule dissociation, and hence produced by direct stimulation of the SR membrane.     

Characterization of "depolarization"-induced calcium release from sarcoplasmic reticulum in vitro with the use of membrane potential probe.

In the triad, the complex of transverse (T) tubule and sarcoplasmic reticulum (SR) Ca2+ release is induced from SR by mediation of the T-tubule. We report here evidence that this Ca2+ release is produced by depolarization of the T-tubule moiety. Thus, we found that the amount of [14C]SCN- taken up by T-tubules and triads (but not that by SR) increased upon incubation with (K, Na) gluconate, Mg ATP, indicating that the T-tubule was polarized making the lumenal side (equivalent to the extracellular side of an intact muscle fiber) more positive. Upon mixing with choline chloride, the procedure to induce Ca2+ release, [14C]SCN- uptake decreased, indicating that the T-tubule became depolarized. Activation of the T-tubule polarization by Na+ and prevention of it by digoxin [inhibitor of the (Na+, K+) pump], respectively, led to activation and inhibition of choline chloride-induced SR Ca2+ release.     

Molecular organization of transverse tubule/sarcoplasmic reticulum junctions during development of excitation-contraction coupling in skeletal muscle.

The relationship between the molecular composition and organization of the triad junction and the development of excitation-contraction (E-C) coupling was investigated in cultured skeletal muscle. Action potential-induced calcium transients develop concomitantly with the first expression of the dihydropyridine receptor (DHPR) and the ryanodine receptor (RyR), which are colocalized in clusters from the time of their earliest appearance. These DHPR/RyR clusters correspond to junctional domains of the transverse tubules (T-tubules) and sarcoplasmic reticulum (SR), respectively. Thus, at first contact T-tubules and SR form molecularly and structurally specialized membrane domains that support E-C coupling. The earliest T-tubule/SR junctions show structural characteristics of mature triads but are diverse in conformation and typically are formed before the extensive development of myofibrils. Whereas the initial formation of T-tubule/SR junctions is independent of association with myofibrils, the reorganization into proper triads occurs as junctions become associated with the border between the A band and the I band of the sarcomere. This final step in triad formation manifests itself in an increased density and uniformity of junctions in the cytoplasm, which in turn results in increased calcium release and reuptake rates.     

A concentration-dependent localization of octopamine-sensitive adenylate cyclase activity in locust skeletal muscle

A concentration-dependent localization of octopamine-sensitive adenylate cyclase activity has been demonstrated in skeletal muscle of the locust, Schistocerca gregaria, using an histochemical technique. In the intermediate speed contracting muscle fibres from the fan region of the extensor-tibiae muscle of the locust hindleg, low concentrations of DL-octopamine (10(-8) M) induce reaction product preferentially in the sarcoplasmic reticular component of the dyads. At slightly higher concentrations (10(-7) and 10(-6) M) lower amounts of diffuse reaction product are also found in the non-dyad sarcoplasmic reticulum and at the sarcolemmal membrane, with occasional amounts of a less diffuse, punctuate product in the transverse tubule (T-tubule) component of the dyads. At higher concentrations (10(-5) and 10(-3) M) the predominant product is the dense, plaque-like accumulations of reaction product in the T-tubule component of the dyads. The results are discussed in terms of the likely physiological significance of the accumulation of reaction product in these different locations.     

Purification of morphologically intact triad structures from skeletal muscle

A procedure has been devised for isolation of triads (t-tubule/sarcoplasmic reticulum (SR) junctional complexes) from rabbit skeletal muscle. The procedure consists of preparation of a heavy microsomal fraction followed by two sequential 90-min sucrose gradient centrifugations to enrich the triads. A pyrophosphate/phosphate/magnesium buffer system was introduced to decrease aggregation in order to achieve effective separation. The preparation time is 12 h. Some differences between purified triads isolated by two variants of this method are noted. The purity of the triad fractions has been estimated by particle counting to be in the vicinity of 50%. There is good retention of morphology and Ca++-loading activity and enrichment in Na+,K+-ATPase and adenylate cyclase. The triads are practically devoid of contractile elements, mitochondria, and free plasmalemma, and low in content of light SR. The method for obtaining enriched triads is reproducible, and sufficient yields are obtained for structural, biochemical, and functional characterization.     

A genetic model for the study of abnormal nerve-muscle interactions at the level of excitation-contraction coupling: the mutation muscular dysgenesis

Excitation-contraction in muscle fibers are coupled through a complex mechanism involving multiproteic components located at a specialized cellular site, the triadic junction. Triads in normal muscle fiber result from the apposition of sarcoplasmic reticulum citernae and T-tubule and possess strikingly organized ultrastructural elements, bridging both types of membranes, the "junctional feet". Muscular dysgenesis in the mouse is characterized by total muscle inactivity in the developing skeletal muscles due to excitation-contraction uncoupling. Triads have been found to be disorganized with no "junctional feet" and dihydropyridine (DHP) binding sites are decreased with no slow Ca2+ currents, suggesting a basic defect in the excitation-contraction coupling machinery itself. We may hypothesize that muscular dysgenesis results in a marked defect in a functional protein involved in the morphogenesis of the triad and/or directly involved in Ca2+ release for contraction.     

The process of new plasmalemma formation in focally injured skeletal muscle fibers

The major ultrastructural events in murine skeletal muscle fibers were examined 3 to 24 hr after segmental injury induced by painting with aldehyde fixative. At 3 hr the viable stump of injured myofibers was separated from the necrotic segment by a zone of supercontracted myofibrils. No demarcating membrane was evident at this time, although occasionally collapsed segments of plasmalemma partially covered the viable stump. By 12 hr after injury myonuclei near the viable stump were centrally placed and numerous whorls of membrane material appeared in the vicinity of the Golgi apparatus. At this stage a convoluted, tortuous membrane exhibiting extensive interdigitations has sealed the structurally normal part of the injured fiber. The myoplasm immediately within this demarcating membrane possessed few myofilaments but numerous vesicles and tubules, several of which were continuous with the demarcating membrane; most degraded sarcoplasmic organelles remained external to the demarcating membrane and leukocytes were observed internalizing the debris. It appears that after segmental injury to skeletal muscle fibers, active production of new sarcoplasmic membranes occurs, which contributes to the formation of the part of the plasmalemma that demarcates the viable portion of the muscle fiber from the injured area.     

Sulfhydryl oxidation overrides Mg(2+) inhibition of calcium-induced calcium release in skeletal muscle triads

We studied the effect of oxidation of sulfhydryl (SH) residues on the inhibition by Mg(2+) of calcium-induced calcium release (CICR) in triad-enriched sarcoplasmic reticulum vesicles isolated from rabbit skeletal muscle. Vesicles were either passively or actively loaded with calcium before eliciting CICR by dilution at pCa 4.6-4.4 in the presence of 1.2 mM free [ATP] and variable free [Mg(2+)]. Native triads exhibited a significant inhibition of CICR by Mg(2+), with a K(0.5) approximately 50 microM. Partial oxidation of vesicles with thimerosal produced a significant increase of release rate constants and initial release rates at all [Mg(2+)] tested (up to 1 mM), and shifted the K(0.5) value for Mg(2+) inhibition to 101 or 137 microM in triads actively or passively loaded with calcium, respectively. Further oxidation of vesicles with thimerosal completely suppressed the inhibitory effect of [Mg(2+)] on CICR, yielding initial rates of CICR of 2 micromol/(mg x s) in the presence of 1 mM free [Mg(2+)]. These effects of oxidation on CICR were fully reversed by SH reducing agents. We propose that oxidation of calcium release channels, by decreasing markedly the affinity of the channel inhibitory site for Mg(2+), makes CICR possible in skeletal muscle.     

Dihydropyridine receptor alpha subunits in normal and dysgenic muscle in vitro: expression of alpha 1 is required for proper targeting and distribution of alpha 2

We have studied the subcellular distribution of the alpha 1 and alpha 2 subunits of the skeletal muscle dihydropyridine (DHP) receptor with immunofluorescence labeling of normal and dysgenic (mdg) muscle in culture. In normal myotubes both alpha subunits were localized in clusters associated with the T-tubule membranes of longitudinally as well as transversely oriented T-tubules. The DHP receptor-rich domains may represent the sites where triad junctions with the sarcoplasmic reticulum are being formed. In cultures from dysgenic muscle the alpha 1 subunit was undetectable and the distribution patterns of the alpha 2 subunit were abnormal. The alpha subunit did not form clusters nor was it discretely localized in the T-tubule system. Instead, alpha 2 was found diffusely distributed in parts of the T-system, in structures in the perinuclear region and in the plasma membrane. These results suggest that an interaction between the two alpha subunits is required for the normal distribution of the alpha 2 subunit in the T-tubule membranes. Spontaneous fusion of normal non-muscle cells with dysgenic myotubes resulted in a regional expression of the alpha 1 polypeptide near the foreign nuclei, thus defining the nuclear domain of a T-tubule membrane protein in multi-nucleated muscle cells. Furthermore, the normal intracellular distribution of the alpha 2 polypeptide was restored in domains containing a foreign "rescue" nucleus; this supports the idea that direct interactions between the DHP receptor alpha 1 and alpha 2 subunits are involved in the organization of the junctional T-tubule membranes.     

Morphology of isolated triads

The triad is the junctional association of transverse tubule with sarcoplasmic reticulum terminal cisternae. A procedure for the isolation of highly enriched triads from skeletal muscle has been described in the previous paper. In the present study, the structural features of isolated triads have been examined by thin-section, negative-staining, and freeze-fracture electron microscopy. In isolated triads, key features of the structure observed in situ have been retained, including the osmiophilic "feet," junctional structures between the transverse tubule and terminal cisternae. New insight into triad structure is obtained by negative staining, which also enables visualization of feet at the junctional face of the terminal cisternae, whereas smaller surface particles, characteristic of calcium pump protein, are not visualized there. Therefore, the junctional face is different from the remainder of the sarcoplasmic reticulum membrane. Junctional feet as viewed by thin section or negative staining have similar periodicity and extend approximately 100 A from the surface of the membrane. Freeze-fracture of isolated triads reveals blocklike structures associated with the membrane of the terminal cisternae at the junctional face, interjunctional connections between the terminal cisternae and t-tubule, and intragap particles. The intragap particles can be observed to be closely associated with the t-tubule. The structure of isolated triads is susceptible to osmotic and salt perturbation, and examples are given regarding differential effects on transverse tubules and terminal cisternae. Conditions that adversely affect morphology must be considered in experimentation with triads as well as in their preparation and handling.     

Sequential docking, molecular differentiation, and positioning of T-Tubule/SR junctions in developing mouse skeletal muscle.

Skeletal muscle Ca(2+) release units (CRUs) are junctions of the surface membrane/T-tubule system and the sarcoplasmic reticulum (SR) that function in excitation-contraction coupling. They contain high concentrations of dihydropyridine receptors (DHPRs) in the T-tubules and of ryanodine receptors (RyR) in the SR and they are positioned at specific locations in the sarcomere. In order to characterize the sequence of developmental steps leading to the specific molecular and structural organization of CRUs, we applied a range of imaging techniques that allowed us to follow the differentiation of the membrane compartments and the expression of junctional proteins in developing mouse diaphragm muscle. We find that docking of the two membrane systems precedes the incorporation of the RyRs into the junctions, and that T-tubule/SR junctions are formed and positioned at the I-A interface at a stage when the orientation of T-tubule is predominantly longitudinal. Thus, the sequence of developmental events is first the docking of T-tubules and SR, secondly the incorporation of RyR in the junctions, thirdly the positioning of the junctions in the sarcomere, and only much later the transverse orientation of the T-tubules. These sequential stages suggests an order of inductive processes for the molecular differentiation and structural organization of the CRUs in skeletal muscle development.     

Occurrence of unusual tubular invaginations of the plasma membrane in smooth muscle cells of the lamprey,Lampetra japonica

Numerous tubular structures were observed in the surface region of smooth muscle cells making up the vascular walls in the lamprey, Lampetra japonica; they were designated as surface tubules. The limiting membrane of the surface tubules was connected to the plasma membrane, allowing communication of the lumen of the tubule with the extracellular space. Tannic acid reacted with osmium, serving as an extracellular marker, penetrated into the tubules but not into the intracellular organelles, such as the endoplasmic reticulum and the Golgi complex. The surface tubules were grouped in longitudinal parallel rows, separated from each other by tubule-free areas where dense plaques were present. Each tubule was fairly cylindrical (approximately 60 nm in diameter) and often ramified into two or three branches with a blind end. Occasionally, these tubules were encircled by the sarcoplasmic reticulum which was located immediately beneath the plasma membrane. Similar tubules were also observed in the surface region of vascular endothelial cells and fibroblasts in the adventitial connective tissue. The possibility that the surface tubules in the present observations are analogous to the smooth muscle caveolae or the striated muscle T-tubule is discussed.     

G-protein dependent potentiation of calcium release from sarcoplasmic reticulum of skeletal muscle

Skinned fibre experiments were conducted to determine if guanine nucleotide-binding proteins play a role in excitation-contraction coupling of skeletal muscle. By itself, the GTP-gamma S, a non hydrolysable GTP analogue was unable to induce calcium release from the sarcoplasmic reticulum, even at concentrations as high as 500 microM. However, calcium- or caffeine-induced calcium releases were enhanced by GTP-gamma S in micromolar concentrations. This response was blocked by GDP-beta S or Pertussis toxin. 32P-ADP-ribosylation catalysed by Pertussis toxin, radiolabelled G-protein alpha subunits in the range of 40 kDa on membrane subcellular fractions of rat skeletal muscle. Using Western blot analysis with antibodies raised against the bovine transducin, G-proteins were identified in frog and rat skeletal muscle subcellular fractions. In most of the muscle fractions (plasma membrane, T-tubules, triads, sarcoplasmic reticulum), the anti-beta subunit antibodies recognized a 36 kDa protein which comigrated with transducin beta subunit. It appears therefore that some of the G-proteins identified by ADP-ribosylation or immunostaining in several subcellular fractions from skeletal muscle, are implicated in the modulation of calcium release from sarcoplasmic reticulum. These results suggest that a Pertussis toxin sensitive G-protein is present at the loci of E-C coupling, and that it serves to regulate the calcium release.     

The epitope for monoclonal antibody A20 (amino acids 870-890) is located on the luminal surface of the Ca2(+)-ATPase of sarcoplasmic reticulum.

The epitope for monoclonal antibody A20 was mapped to amino acids 870-890 of the Ca2(+)-ATPase of rabbit fast-twitch skeletal muscle sarcoplasmic reticulum. The antibody did not react with the epitope in intact sarcoplasmic reticulum vesicles but reacted with the epitope when the vesicles were solubilized with the detergent C12E8 or made permeable by incubation in a hypotonic medium. By contrast, antibody A52, which binds to a cytoplasmic epitope consisting of amino acids 657-672, reacted with the Ca2(+)-ATPase in vesicular, permeabilized vesicular, and C12E8-solubilized states. These results clearly demonstrate that antibody A20 binds to a luminal epitope and provide the first demonstration that a specific segment of the Ca2(+)-ATPase is located on the luminal surface of the sarcoplasmic reticulum. These results are consistent with, and support, our model for folding of the Ca2(+)-ATPase (Brandl, C. J., Korczak, B., Green, N. M., and MacLennan, D. H. (1986) Cell 44, 597-607) in which residues 657-672 were proposed to form part of the cytoplasmic nucleotide binding domain, while residues 870-890 were proposed to form a luminal loop between proposed transmembrane sequences M7 and M8.