General and specific lipid-protein interactions in Na,K-ATPase.

The molecular activity of Na,K-ATPase and other P2 ATPases like Ca(2+)-ATPase is influenced by the lipid environment via both general (physical) and specific (chemical) interactions. Whereas the general effects of bilayer structure on membrane protein function are fairly well described and understood, the importance of the specific interactions has only been realized within the last decade due particularly to the growing field of membrane protein crystallization, which has shed new light on the molecular details of specific lipid-protein interactions. It is a remarkable observation that specific lipid-protein interactions seem to be evolutionarily conserved, and conformations of specifically bound lipids at the lipid-protein surface within the membrane are similar in crystal structures determined with different techniques and sources of the protein, despite the rather weak lipid-protein interaction energy. Studies of purified detergent-soluble recombinant αß or αßFXYD Na,K-ATPase complexes reveal three separate functional effects of phospholipids and cholesterol with characteristic structural selectivity. The observations suggest that these three effects are exerted at separate binding sites for phophatidylserine/cholesterol (stabilizing), polyunsaturated phosphatidylethanolamine (stimulatory), and saturated PC or sphingomyelin/cholesterol (inhibitory), which may be located within three lipid-binding pockets identified in recent crystal structures of Na,K-ATPase. The findings point to a central role of direct and specific interactions of different phospholipids and cholesterol in determining both stability and molecular activity of Na,K-ATPase and possible implications for physiological regulation by membrane lipid composition. This article is part of a special issue titled "Lipid-Protein Interactions."

Three-dimensional structure of the acetylcholine receptor by cryoelectron microscopy and helical image reconstruction.

Long tubular vesicles have been grown from isolated Torpedo postsynaptic membranes, in which the receptors are arranged helically on the vesicle surface. The structures of these tubes have been analyzed by cryoelectron microscopy of specimens embedded in thin films of ice, combined with helical image reconstruction. Complete data sets from tubes belonging to several helical families have been obtained to a resolution of 17 A in all directions. Confirming a preliminary study (Toyoshima, C., and N. Unwin. 1988. Nature (Lond.). 336:247-250), the central ion channel has an almost constant diameter throughout the molecule except for the portion extending through the hydrophobic part of the lipid bilayer, where the pore is too small to be resolved. However, the density on the pseudo fivefold axis running through the pore is consistently highest in the cytoplasmic half of the bilayer, suggesting the gate is located in that region. The path followed by each subunit has been identified throughout the length of the receptor. The two alpha subunits follow equivalent paths. All subunits have similar features which change in character at the same level relative to the membrane.

Arrangement of the acetylcholine receptor subunits in the resting and desensitized states, determined by cryoelectron microscopy of crystallized Torpedo postsynaptic membranes.

Two conformational states of the nicotinic acetylcholine receptor have been investigated by cryoelectron microscopy of flattened vesicular crystals grown from Torpedo marmorata postsynaptic membranes. One was obtained from the vesicles without acetylcholine present, and is presumed to correspond to the native, or resting state; the other was obtained from the vesicles after exposure to 100 microM to 5 mM carbamylcholine (an acetylcholine analogue) and is presumed to correspond to a desensitized state. Both conformations were determined in three-dimensions to a resolution of 18 A, sufficient to reveal the configurations of the five subunits around the central ion channel over most of their length. The subunits of either structure have a similar appearance, consistent with their amino acid homology. They are each aligned almost parallel to the axis of the receptor, conferring a high degree of pentagonal symmetry to the bilayer portion and a contiguous region on the synaptic side. Their external surfaces form a pronounced ridge in the bilayer portion, which broadens toward the synaptic end. Comparison of features in the two three-dimensional maps reveals that carbamylcholine induces a quaternary rearrangement, involving predominantly the delta-subunit. The densities corresponding to this subunit are tilted by approximately 10 degrees tangential to the axis of the receptor over a large fraction of its length, and become misaligned relative to the densities corresponding to the other four subunits. The gamma-subunit is also affected, being displaced slightly away from the axis of the receptor. The alpha- and beta-subunits may be affected on a more localized scale. The overall changes are most pronounced in the synaptic region, where the ligand-binding site is located, and in the cytoplasmic region, which may be closer to the gate of the channel. The physiological process of desensitization appears to be associated with a structural transition in which the subunits switch to a less symmetrical configuration.

Two-dimensional crystallization of catalase on a monolayer film of poly(1-benzyl-L-histidine) spread at the air/water interface.

Two-dimensional (2D) crystals of beef liver catalase were prepared by adsorption to a film of synthetic polypeptide, poly(1-benzyl-L-histidine) (PBLH), spread at the air/water interface. The crystallization experiments were carried out in the pH range of 4.8-6.4 for catalase solutions at low concentration (10 micrograms/ml). The pH-dependence suggested an electrostatic interaction in the binding of catalase to the PBLH film. At lower pH, small crystals were formed at a low binding rate, and at higher pH the binding was rapid and densely-packed 2D arrays with poor crystallinity were formed. To stimulate crystal growth, a thermal treatment was applied. One-shot heating of the interfacial catalase-PBLH film to 35-40 degrees C was remarkably effective to form larger 2D crystals. The structure of catalase 2D crystals has been analyzed by Fourier filtering of the transmission electron micrographs. The crystal form is a new one, containing four catalase molecules in the unit cell with lattice parameters of alpha = 187 A, b = 225 A and gamma = 92.8 degrees.

Two-dimensional crystallization and cryo-electron microscopy of photosystem II.

Two dimensional (2D) crystals of photosystem II (PSII) were obtained from n-heptyl-beta-D-thioglucoside-solubilized monomers of spinach PSII complex by conventional detergent dialysis. The 2D crystals were either large cylindrical vesicles (1 to 2 micrometer by 4 to 6 micrometer as flattened vesicles) or large monolayer sheets (ca. 1 micrometer X 1 micrometer), both suitable for cryo-electron microscopy. Images of unstained crystals embedded in ice were recorded using low-dose microscopy and analyzed by digital image processing. Both types of crystals had the same unit cell size and the same packing arrangement of PSII particles. The plane group was p22(1)2(1) and the unit cell was rectangular with dimensions of 16.7 nm X 15.3 nm containing four monomers (two face-up and two face-down). SDS-PAGE and immunoblot analyses of the 2D crystal indicated that the constituent subunits of particles in the 2D were CP47, D1, D2, cytochrome b-559 and psbI protein. A projection map of 20 A resolution revealed that each monomer has asymmetrical shape with a length of 8.1 nm and a maximal width of 7.5 nm consisting of four areas of density. Two high-density areas with similar sizes were located close to each other to form a roughly rectangular core of 4.0 nm X 6.5 nm. From its size similarity to the size of the L/M heterodimer of the bacterial reaction center, this high density core area was tentatively assigned to the D1/D2 heterodimer. The remaining large and small areas were also tentatively assigned to CP47 and cytochrome b-559 plus psbI protein respectively.

The structure of the R-type straight flagellar filament of Salmonella at 9 A resolution by electron cryomicroscopy.

The supercoiled forms of the flagellar filaments are thought to be constructed from a mixture of two distinct subunit conformations arranged in a regular manner. We analyzed the structure of one of the two straight flagellar filaments, each of which is built up with all its subunits in one of the two conformations. The filament we studied was isolated from the strain SJW1655 of Salmonella typhimurium and had a right-handed helical symmetry. With recent advancements in electron cryomicroscopy, such as a liquid helium temperature stage for frozen hydrated specimens and a stable field emission source, and also by averaging high resolution data with a proper correction of the contrast transfer function, the density distribution map of this straight flagellar filament was generated in far more detail than before by including data up to 9 A resolution. The structure shows a densely packed core region from about 15 to 55 A in radius, where a pair of concentric tubular features of high density is present without well-defined subunit boundaries, and an outer part from 55 to 115 A, where the subunits are mostly well separated from each other. The outer tube in the core region, from 35 to 55 A in radius, contains many rod-like features with near-axial orientation and closest lateral distances of around 10 A, which are most likely to represent the alpha-helical bundles that were predicted in our previous report. In the inner tube, from 15 to 30 A in radius, the rod-like features are less clear. Between the inner and outer tubes are the short spoke-like densities, which are radially tilted and are connecting the two tubes. The outer part, from 55 to 115 A, contains an axially elongated column density and a slewed projection with a narrow neck region. When compared with the other straight filament having left-handed helical symmetry, this outer part does not show any significant changes in orientation, suggesting that the switch in the subunit conformation and packing involved in the polymorphic transitions is quite subtle and only occurs within the core region. Reassignment of each structural domain to the amino acid sequence is suggested, based on the volume of each domain, which was determined rather precisely by a proper correction of the contrast transfer function for both amplitudes and phases.

Soluble P-type ATPase from an archaeon, Methanococcus jannaschii.

MJ0968 has been proposed to be an ancestor of P-type ATPase, because its primary structure is highly homologous to that of the core catalytic domain of P-type ATPase. However it completely lacks amino acid sequences that possibly constitute transmembrane domains. To examine if MJ0968 is indeed a P-type ATPase, it was overexpressed in Escherichia coli and purified. It did show ATPase activity, autophosphorylation and inhibition by vanadate. All these properties support the idea that MJ0968 is indeed a soluble P-type ATPase.

Organization of cytoplasmic domains of sarcoplasmic reticulum Ca(2+)-ATPase in E(1)P and E(1)ATP states: a limited proteolysis study.

In order to characterize the domain organization of sarcoplasmic reticulum Ca(2+)-ATPase in different physiological states, limited proteolysis using three proteases (proteinase K (prtK), V8 and trypsin) was conducted systematically and quantitatively. The differences between E(2) and E(2)P were examined in our previous study and E(2)P was characterized by the complete resistance to all three proteases (except for trypsin attack at the very top of the molecule (T1 site)). The same strategies were employed in this study for E(1)ATP, E(1)PADP and E(1)P states. Because of the transient nature of these states, they were either stabilized by non-hydrolyzable analogues or made predominant by adjusting buffer conditions. Aluminum fluoride (without ADP) was found to stabilize E(1)P. All these states were characterized by strong (E(1)ATP) to complete (E(1)PADP and E(1)P) resistance to prtK and to V8 but only weak resistance to trypsin at the T2 site. Because prtK and V8 primarily attack the loops connecting the A domain to the transmembrane helices whereas the trypsin T2 site (Arg(198)) is located on the outermost loop in the A domain, these results lead us to propose that the A domain undergoes a large amount of rotation between E(1)P and E(2)P. Combined with previous results, we demonstrated that four states can be clearly distinguished by the susceptibility to three proteases, which will be very useful for establishing the conditions for structural studies.

ADP-insensitive phosphoenzyme intermediate of sarcoplasmic reticulum Ca(2+)-ATPase has a compact conformation resistant to proteinase K, V8 protease and trypsin.

Sarcoplasmic reticulum Ca(2+)-ATPase was digested with proteinase K, V8 protease and trypsin in the absence of Ca(2+). Unphosphorylated enzyme was rapidly degraded. In contrast, ADP-insensitive phosphoenzyme formed with P(i) and phosphorylated state analogues produced by the binding of F(-) or orthovanadate, were almost completely resistant to the proteolysis except for tryptic cleavage at the T1 site (Arg(505)). The results indicate that the phosphoenzyme and its analogues have a very compact form in the cytoplasmic region, being consistent with large domain motions (gathering of three cytoplasmic domains). Results further show that the structure of the enzyme with bound decavanadate is very similar to ADP-insensitive phosphoenzyme. Thapsigargin did not affect the changes in digestion time course induced by the formation of the phosphorylated state analogues.

Three-dimensional image analysis of the complex of thin filaments and myosin molecules from skeletal muscle. I. Tilt angle of myosin subfragment-1 in the rigor complex.

A three-dimensional image of the "rigor" complex of actin and chymotryptic myosin subfragment-1 was reconstituted from electron micrographs of negatively stained specimens. Data went out to 20 A radially and 26 A axially. The reconstituted images allowed us to deduce the angle between the major axis of the main part of myosin subfragment-1 and the axis of the actin helix. The subfragment-1 molecules were attached to the actin filament in a configuration in which they were tilted by only about 15 degrees from the plane perpendicular to the axis of the actin helix. The implication of the smaller tilt angle than the commonly accepted value is discussed.

Three-dimensional image analysis of the complex of thin filaments and myosin molecules from skeletal muscle. IV. Reconstitution from minimal- and high-dose images of the actin-tropomyosin-myosin subfragment-1 complex.

Three-dimensional images of the actin-tropomyosin-myosin subfragment-1 (S1) complex were reconstituted from both minimal- and high-dose electron micrographs by using a conventional reconstruction technique. Higher resolution (1/15 A-1) than those of the previous reconstructions was attained. A multi-domain structure similar to that of the actin-S1 complex described in the previous paper (1) was observed and a ne diagram of the multi-domain structure of the actin-tropomyosin-S1 complex is presented. The shape of S1 molecules in the rigor complex was clearly resolved. In a view perpendicular to the filament axis, S1 had an axially bent profile; only the tail portion, which was thin but was not small in diameter, was steeply inclined. These features were more prominent in the model from minimal-dose images than that from high-dose images.

Three-dimensional image analysis of the complex of thin filaments and myosin molecules from skeletal muscle. V. Assignment of actin in the actin-tropomyosin-myosin subfragment-1 complex.

To assign the actin molecule in the three-dimensional image of the actin-tropomyosin-myosin subfragment-1 (actin-TM-S1) complex, the three-dimensional image of the actin-tropomyosin complex was correlated to that of actin-TM-S1. To assess the similarity of two structures in a quantitative manner, we used a normalized cross-correlation function ("similarity function"). The calculation of similarity indicated that domain A and domain B defined in (1, 2) correspond to actin-tropomyosin. This assignment indicates that one S1 molecule strongly interacts with only one actin molecule, but at least two regions of S1 contribute to the binding. Comparison of the reconstituted models of thin filaments with those of decorated thin filaments suggested a change in the shape of the actin molecule.

Three-dimensional image analysis of the complex of thin filaments and myosin molecules from skeletal muscle. II. The multi-domain structure of actin-myosin S1 complex.

A three-dimensional image of the "rigor" complex of actin and chymotryptic myosin subfragment-1 (S1) was reconstituted from electron micrographs of specimens embedded in unbroken and unbacked stain sheets of uranyl acetate over the holes of perforated carbon films to an effective resolution of 20 A radially and 26 A axially. The morphological unit of actin-S1 complex consists of at least three domains and myosin S1 shows a multi-domain submolecular structure. Possible ways to assign actin to one or more of these three domains are discussed. Two candidates for the shape of S1 molecule are also shown. Both candidates have a complex "embryo"-like shape. The solid model of the actin-S1 complex appears to be far less polar than that shown in the original electron micrographs or the projected density map of the reconstituted image shown by Toyoshima and Wakabayashi (1). The conspicuous polarity of the arrowhead pattern is related to the projected image of the spiral shape of the main part of the S1 molecule, which is almost at right angles to the helix axis. As one way to reconcile the non-tilted configuration of S1 in the rigor complex with sliding theory, the possibility of a pivoting mechanism (rotation of S1 head in the horizontal plane normal to the helix axis rather than in the vertical plane parallel to the helix axis) is discussed.

Electron microscopic study on the location of 23 kDa and 50 kDa fragments in skeletal myosin head.

The functional activities of myosin head are located in a 95 kilodalton (kDa) heavy chain which can be divided into three fragments of 23 kDa, 50 kDa, and 20 kDa. ATP hydrolysis sites were suggested to be located in the 23 kDa and 50 kDa fragments, and actin binding sites were in the 50 kDa and 20 kDa fragments. In this study, we obtained electron microscopic images of the myosin molecule bound with antibodies directed to the 23 kDa and 50 kDa fragments. We determined that the antigenic sites for 23 kDa fragment are located at 140-180 A from the head-rod junction of myosin, and those for 50 kDa fragment at 160 A from the junction and at the tip of the head itself. The relationship between the spatial locations and the primary structures is discussed.

Structure determination of tubular crystals of membrane proteins. I. Indexing of diffraction patterns.

Indexing of diffraction patterns is the starting point of every crystallographic analysis. The diffraction patterns from helical particles consist of a series of layer-planes, the indexing of which involves the assignment of the start numbers of the helices that contribute to the layer-planes. The indexing could be challenging if the diameter of the helical particle is large and if the conventional indexing methods based on selection rule is used. Presented here is a tutorial on how to index them using (h, k; n) indexing method, which is particularly useful for tubular crystals of membrane proteins.

Contrast transfer for frozen-hydrated specimens: determination from pairs of defocused images.

Electron imaging of frozen-hydrated biological molecules allows density maps to be obtained directly, without the need for fixatives or stains. The appearance of such maps may, however, be strongly influenced by the contrast transfer properties, which have not previously been evaluated by quantitative experiments. Here we determine the contribution due to amplitude contrast in a typical (approximately 300 A thick) frozen specimen, consisting of arrays of acetylcholine receptor, by comparing pairs of images recorded with different defocuses. We find that this specimen is imaged as a "weak-phase-weak-amplitude" object and that the contribution due to amplitude contrast is 7%.

Structure determination of tubular crystals of membrane proteins. II. Averaging of tubular crystals of different helical classes.

A set of programs has been developed for averaging the data from tubular crystals belonging to different helical classes. This was done either by (i) cutting out molecules constituting a unit cell from density maps, and aligning and averaging them in real space; (ii) transforming the densities in a unit cell to layer-line data according to a (possibly artificial) helical symmetry, aligning and averaging them in reciprocal space. These methods were applied to tubular crystals of Ca2+-ATPase. Either method worked well and substantially improved the data quality. Transforming the reconstructed images to the layer-line data has many advantages and is essential for fully exploiting the power of averaging.

Structure determination of tubular crystals of membrane proteins. III. Solvent flattening.

Solvent flattening is considered to be a principal means for improving the data quality in X-ray crystallography. It could be equally effective for tubular crystals of membrane proteins imaged by electron microscopy because of the large empty space inside the tubes. However, tubular crystals are difficult objects for solvent flattening due to lack of electron diffraction amplitudes. Therefore, solvent flattening was used to align images more accurately and to improve the completeness of the data by reducing contributions of noise in the solvent (+ lipid) region. The methods developed were tested with the tubular crystals of Ca2+-ATPase embedded in amorphous ice. The improvement of the data quality was remarkable when solvent flattening was applied to many individual images before averaging. In this way, noises contaminated in the protein region by contrast transfer function were removed effectively. Solvent flattening was far more powerful than simple averaging described in Part II of this series (K. Yonekura, C. Toyoshima, Ultramicroscopy 84 (2000) 15).

Image analysis of the complex of actin-tropomyosin and myosin subfragment 1.

A three-dimensional image of the "rigor" complex of actin-tropomyosin-S1 was reconstituted from both low dose (10 electrons/A2) and high dose (greater than 500 electrons/A2) electron microscopic images of specimens embedded in unbroken and unbacked stain sheets of uranyl acetate over the holes of perforated carbon films. Myosin S1 shows multi-domain submolecular structure as has been earlier observed in actin-S1 ( Wakabayshi & Toyoshima, 1981) and actin-heavy meromyosin ( Katayama & Wakabayashi , 1981). The morphological unit of the actin-tropomyosin-S1 was found to be composed of at least three domains (domains A, B and D) and three regions (C, E and H). A myosin S1 molecule has a complex shape, which cannot be represented by a simple rod with one major axis. The shape of S1 should be approximated by at least two rods. The domain D is identified as the main part of S1. The angle between the major axis of this domain and the axis of actin helix was about 72 degrees, which is almost right angle. The angle between the axis of actin helix and major axis of the region E, which is less bulky than the domain D and makes no contact with actin, is much smaller than the value for the domain D. The resolution of reconstituted images from both high and low dose micrographs was improved so that the radial resolution became about 15 A and the axial one became about 25 A.(ABSTRACT TRUNCATED AT 250 WORDS)

Temperature-induced change of thick filament and location of the functional sites of myosin.

The combination of the small-angle X-ray camera to use the synchrotron radiation (National Lab. of High Energy Physics, Tsukuba) and a sensitive X-ray detecting system (FCR System, Fuji Medical System Inc.) using the imaging plate enabled us to record a small angle X-ray diagram of liver rabbit muscle within few minutes. Live relaxed rabbit muscle kept at 25 degrees C gave clear relaxed pattern, in which myosin layer line can be observed up to 13th order. When it was cooled down to 5 degrees C, it gave the small-angle X-ray pattern which looks like that obtained from contracting frog muscle, whereas cooled muscle produced no tension. The 8th meridional reflection almost vanished. The pattern is different from that obtained from rigor muscle: actin layer line at 72 A spacing can not be observed. The "stiffness" and (1,1) reflection on the equator increased. This indicates that more crossbridges are at the vicinity of the thin filaments without developing tension at low temperature. This might be related to the marked decrease in the rate of the step from M*ADP to M.ADP at low temperature. The position of the ATP binding site, SH1, actin binding site(s) was determined by three-dimensional image reconstruction method. Actin-binding site was determined by comparing the three-dimensional image of actin-tropomyosin-S1 and that of actin-tropomyosin-troponin-Ca. The position of ATP binding site and SH1 was determined by three-dimensional reconstruction of the complexes of actin-tropomyosin and S1 of which the ATP binding site or SH1 was labelled with avidin. It was found that SH1 locates near the actin-binding site at the same side of S1. The distance from head-rod junction to ATP binding site and SH1 was similar. But they locate at the different side of S1 and the distance between two sites is about 5 nm, which is consistent with that obtained by energy transfer method.