Electron Microscope Observations on Ca2+-ATPase Microcrystals in ...

Vol. 263, No. 11, Issue of April 15, pp. 5267-5294.1988 Printed in U.S. A.

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Electron Microscope Observations on Ca2+-ATPase Microcrystals in Detergent-solubilized Sarcoplasmic Reticulum* (Received for publication, July 28, 1987)

Kenneth A. TaylorS5, Nandor Mullnerll, Slawomir Pikulall, Laszlo Duxll, Camillo Peracchiall, Sandor Vargall, and Anthony Martonosill** From the lDepartment of Biochemistry and Molecular Biology, State University of New York,Health Science Center, Syracuse, New York 13210, the IlDepartment of Physiology, University of Rochester School of Medicine and Dentistry, Rochester, New York 14642, and the $Department of Anatomy, Duke University Medical Center, Durham, North Carolina 27710

membrane crystals (Taylor et aZ.,l984,1986a, 1986b), and xCrystallinearraysofCa2+-ATPasemoleculesdevelop in detergent-solubilized sarcoplasmic reticulum ray andneutron diffraction studies on oriented layers of duringincubationfor several weeks at 2 "C under sarcoplasmic reticulum membranes (Blasie et al. 1985) pronitrogen in a medium of 0.1 M KCl, 10 mM K-3-(N- vided a low resolution profile of the overall shape of ATPase morpholino)propanesulfonate, pH 6.0,3 mM MgC12,20 molecules and their interaction in the native membrane. mM CaC12,20%glycerol, 3 mM NaN3, 5 mM dithiothreThe production of microcrystals of Ca*+-ATPasein deteritol, 25 IU/mlTrasylol, 2 pg/ml 1,6-di-tert-butyl-p- gent-solubilized sarcoplasmic reticulum and purified ATPase cresol, 2 mg/ml protein, and 2-4 mg of detergent/mg preparations (Pikula et al., 1988) is the first step in extending ofprotein.Electronmicroscopyofsectioned,negathe structural analysis to a level of resolution where functiontively stained, freeze-fractured, and frozen-hydrated ally relevant details of the Ca2+-ATPase structure willbe Ca2+-ATPasecrystals indicatesthattheyconsist of recognizable. In this report electron microscope observations stackedlamellar arrays ofCaZ+-ATPasemolecules. are presented on the Ca2+-inducedcrystals of Ca2+-ATPase Prominent periodicities of ATPase molecules within the lamellae arise from a*enteredrectangular lattice in detergent-solubilized and native sarcoplasmic reticulum. of dimensions164 X 55.5 A. The association of lamellae EXPERIMENTAL PROCEDURES into three-dimensional stacks is assumedto involve The methods for the isolation of sarcoplasmic reticulum and puriinteractionsbetweentheexposedhydrophilicheadgroups of ATPase molecules, that is promoted byglyc- fied ATPase preparations and for the characterization of their enzyerol and 20 mM Ca2+. Similar Ca2+-inducedcrystals matic activity, and protein and phospholipid composition are described in the preceding report (Pikula et al., 1988). Delipidation of were observed with purified or purified and delipisarcoplasmic reticulum was carried out according to Dean and Tandated Ca2'-ATPase preparations at lower detergent/ ford (1977, 1978). The Caz+-ATPasecrystals were produced by incuprotein ratios. Cross-linking of Ca2+-ATPase crystals bation of sarcoplasmic reticulum or purified Ca*+-ATPase(2 mgof with glutaraldehyde protects the structure against protein/ml) conat 2 "C under nitrogen atmosphere for several weeks in ditions such as low Ca2+, highpH, elevated tempera- a medium of 0.1 M KCl, 20 mM K-MOPS,' pH 6.0, 3 mM MgC12,20 ture, SH group reagents, high concentration of deter- mM CaC12, 20%glycerol, 3 mM NaN3, 5 mM dithiothreitol, 25 IU/ml and 2-4 mg of detergent/ gents, and removal of phospholipids by extraction with Trasylol, 2 pg/ml 1,6-di-tert-butyl-p-cresol, mg of protein. The four detergents employed in these studies were organic solvents that disrupt unfixed preparations.

ClzE8,Brij 36T, Brij 56, and Brij 96. Detailed information about the selection of detergents and theeffects of medium components on the stability of Caz+-ATPaseis presented by Pikula et al. (1988). The microcrystals of Caz+-ATPasewere analyzed using negatively The kinetic description of the ATP-dependent transport of stained, freeze-fractured, and thin sectioned material as described Ca2+by the Ca2+-ATPaseof sarcoplasmic reticulum is now earlier (Dux and Martonosi, 1983a, 1983b, 1983c, 1983d, 1984; Dux al., 1985, 1987; Peracchia et al., 1984; Taylor et al., 1984, 1986a, refined toits probable limit (de Meis and Vianna, 1979; et 1986b; Pikula et a t , 1988). Electron micrographs of frozen hydrated Tanford, 1984; Inesi and de Meis, 1985); further progress in microcrystals were obtained from specimens prepared as described the understanding of the transport process requires detailed by Dubochet (1982) and examined with low electlon dose techniques knowledge of the structure of the Ca2+pump. The establish- on a Philips EM420 using a Gatan model 626 cryotransfe. Jystem ment of the amino acid sequence of Ca2+-ATPaseisoenzymes and cooling holder. There was no internal calibration of magnifica(MacLennan et al., 1985; Brandl et al., 1986,1987) with tion. Images of frozen hydrated membranes were screened initially optical diffraction; images the* diffracted well were digitizedusing predictions about their secondary and tertiary structure rep- by 2 Perkin-Elmer PDSlOlOM microdensitometer a t a step size of 6.95 resents a major step in that direction. In a parallel line of A with respect to theoriginal object. Digitized images werecomputer development, electron microscope analysis of Ca*+-ATPase processed as described earlier (Taylor et al., 1984). Sigma supplied Brij 36T, Brij 56, Brij 96, phospholipase C (Clos* This work was supported by Research Grants AM 26545 (to A. tridium welchii), and bovine serum albumin. C12Ee wasa product of M.), GM20113 (to C. P.), and GM 30598 (to K. A. T.) from the Behring Diagnostics. Trasylol (aprotinin, 10,000IU/ml) was obtained National Institutes of Health, United States Public Health Service, from Mobay Chemical Corp., New York, asolectin (95% purified soy Grant PCM 84-03679 (to A. M.) from the National Science Founda- phosphatides) from Associated Concentrates, Woodside,NY, and tion, and by a grant from the Muscular Dystrophy Association. The phospholipase AZ (Crotalus durissus) from Boehringer Mannheim, costs of publication of this article were defrayed in part by the GMBH, Mannheirn, Federal Republic of Germany. Salyrgan was a payment of page charges. This article must therefore be hereby product of Mann Research Laboratories, NY, uranyl acetate of Fisher, marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The abbreviations used are: MOPS, 3-(N-morpholino) propanesulfonic acid; ClzE8,octaethylene glycol dodecyl ether SDS, I Established Investigator of the American Heart Association. sodium dodecyl sulfate. ** To whom correspondence should be addressed.

5287

5288

Microcrystals of Ca2+-ATPase

FIG. 1. Electron microscopy of Ca"+-ATPasecrystals in thin sections. Sarcoplasmic reticulum (2 mg of protein/ml) was solubilized in the standard crystallization medium with ClzER(2 mg/mg protein) and incubated under nitrogen a t 2 "Cfor 15 days. The crystalline sediment was embedded in Epon-Araldite mixture andprocessed for electron microscopy. Depending on conditions duringfixation, embedding, sectioning, and viewing the observed periodicities in different specimens varied between 103 and 147 A. Magnification, X 207,000.

5289

Microcrystals of Ca2+-ATPase and glutaraldehyde (8%aqueous) of Polysciences, Inc., Warrington, PA. RESULTS

Microcrystals of Ca2’-ATPase in Detergent-solubilizedSarcoplasmicReticulum-Sarcoplasmic reticulum vesicles suspended in the standard crystallization medium of 0.1 M KCl, 10 mM K-MOPS, pH 6.0, 3 mM MgC12,20 mM CaCl,, 20% glycerol, 3 mM NaN3, 5 mM dithiothreitol, 25 IU/ml Trasylol, and 2 pg/ml 1,6-di-tert-butyl-p-cresol at 2 “C readily dissolve upon addition of 2 mgof CI2E8or 4 mg of Brij 36T/mg of protein and form an opalescent solution. After centrifugation at 54,000 X g for 1 h more than 75% of the total protein remains inthe supernatantindicating extensive solubilization of the Ca2’-ATPase. This was confirmed by SDS-polyacrylamide gel electrophoresis of the nonsedimentable protein fraction that identified the 109,000-Da band of the Ca2+ATPase as the principal protein component. After 1 day of incubation in the standard crystallization medium at 2 “C under nitrogen, electron microscopy of specimens negatively stained with uranylacetate shows Ca2+ATPase molecules in various stages of aggregation and the absence of intact vesicles; occasionally small crystalline arrays begin to appear interspersed with apparently random Ca2+ATPase aggregates. During incubation for 1-2 weeks under nitrogen Ca2’-ATPase microcrystals develop that increase in number and size during the next several months. Electron microscopy of sectioned (Fig. l ) , negatively stained (Figs. 2 and 3), frozen-hydrated (Fig. 4), and freeze-fractured crystals (Fig. 5) reveals multilamellar arrays of Ca2+-ATPasemolecules in different projections; the crystals vary in size between 0.2 and 1.0 pm. The three-dimensional character of the crystals is evident from electron microscopy of thin sectioned (Fig. 1) and frozen-hydrated (Fig. 4) material that shows the ordered arrays in various orientations within the specimen, from freeze-etch electron microscopy that reveals the stacked lamellae (Fig. 5), and from serial sections that resolve the structure of the crystals in the thirddimension (not shown). Two distinct patterns are observed in sectioned or in negatively stained materials. In eo ! view, layers of densities are seen that repeat at -103-147 A in sectioned specimens (Fig. l), at 13O-laO A in negatively stained material (Fig. 2), and at 170-180 A in images of frozen-hydrated specimens (Fig. 4). Theserepresent side views of stacked lamellar arrays of ATPase molecules. The smaller spacings obtained in sectioned material are presumably due to shrinkage of the specimen during dehydration, embedding, sectioning, and viewing. We assume that the cores of the lamellae contain a lipiddetergent phase into which the hydrophobic tail portions of the ATPase molecules are inserted symmetrically on both sides. The periodicity of the lamellae is presumably defined by contacts between the hydrophilic headgroups of ATPase molecules. The nearly ubiquitous presence of stacked lamellae suggests that under the conditions used either a three-dimensional crystallization occurs or that the Ca2+-ATPasemolecules first form two-dimensional crystalline sheets which aggregate to form the stacked lamellae. The variable size of the

lamellae within the stacks as seen in negatively stained as well as in sectioned material suggests that the stacks arise through aggregation of two-dimensional crystalline sheets. In areas of favorable orientation, the negative stain exposes yiews of the interacting surfaces studded with rows of 40-50A diameter particles (Fig. 2). The particles are presumed to represent the head portions of Ca2’-ATPase molecules. The variable size and the broken ends of the arrays suggest that the crystals are fragile and undergo fragmentation during processing in spite of the care exercised to avoid shear stress. The thickness of crystalline sheets is much greater than that observed earlier in images and three-dimensional reconstructions of frozen hydrated vanadate-induced crystals of Ca2’-ATPase in native sarcoplasmic reticulum membranes (Taylor et al., 1986a, 1986b). Crystals formed in native sarcoplasmic reticulum membranes have a thickness of 100-110 A (Taylor etal., 1986b),and all ATPase molecules project out of the cytoplasmic side of the bilayer. The observed thickness of this new crystal form is compatible with the dimensions of ATPase molecules in the native membranes if it is assumed that theATPase molecules e@endout of both sides of a lipiddetergent phase of about 40-A thickness. There is no suggestion in these edge-on viewsof a regular pattern to the stacking of the lamellae. In the second view (Fig. 3) the projected image normal to the plane of the lamella shows ordered arrays of stain-excluding particles. The particles are presumed to represent views of the cytoplasmic domains of Ca2+-ATPasemolecules. Under the solvent conditions where crystals are formed andare stable, images of the negatively stained crystals viewed from thisorientation usually show superlattice periodicities or moire fringes. Attempts to disrupt the stacked layers either mechanically or by lowering the ionic strength have not so far produced convincing images of a single lamella, at least in specimens preserved for electron microscopy using negative stain. In-plane projections of single sheets have been difficult to obtain due to the strong tendency of the sheets to aggregate. The presence of moire fringes is strong evidence of a slight rotational misalignment between the successive stacks of twodimensional sheets that prevents the separation of the low resolution projection of layers by Fourier filtering. Computed diffraction patterns obtained from electron micrographs of stacks of frozen hydrated two-dimensional sheets (qg. 6) gave unit cell dimensions of 164.2 & 2.2 and 55.5 f 1.5 A (average of 17 crystals) with an included angle of 90 The best Teesolution that has been obtained so far is limited to about 30 A. The diffraction pattern, when indexed in this fashion, shows a set of systematic absences with observed spots obeying the selection rule h + k = 2n (where n is any integer). Such a selection rule is indicative of a lattice centered on the “c” face of the crystal. Under this indexing scheme the 80-A periodicities observed in images of negatively stained and frozen-hydrated membranes represent the haif-period of the “a” cell dimensions, while the two sets of 50-A striations are the (1,1)and (-1,l) spacings. O.

FIG.2. Negatively stained crystalline arrays of sarcoplasmic reticulum Ca2+-ATPasesolubilized with C,,E, (2 mg/mg protein) in the standard crystallization medium. Magnification, X 308,000, FIG.3. Negatively stained Ca2+-ATPasecrystals in sarcoplasmic reticulum solubilized with CI2E, (2 mg/mg protein) in the standard crystallization medium. The prominent large spacing is the half-period of the “a”cell dimension. Striations oblique to thisdirection are the (1,l)and the(-1,l) periodicities. Magnification, X

308,000. FIG.4. Electron micrograph of ATPase crystals preserved frozen-hydrated in amorphous ice. This

view is parallel to theplane of the crystalline sheet and illustrates thestacking of the membranes and theirspacing in images of membranes preserved in amorphous ice. Magnification, X 108,000.

5290

Microcrystals of ea2+-ATPase

FIG.6. Electron micrographs and computed diffraction patt e r n s of ATPase crystals preserved frozen-hydrated in amorphous ice. Theseare views normal to the membrane plane and exhibit some evidence of moiri! patterns, suggesting stacking of several membrane sheets. The striationsrepresent the half-period of the 164-A celldimension. Magnification, X 216,000.

Among the 17 two-sided plane groups allowed for biological molecules, only two are centered, c12 and c222 (nomenclature of Holser, 1958). Both of these two-sided plane groups would show diffraction patterns with mm symmetry (two perpendicular mirror lines in the hkO zone). The irregular stacking of the sheets often breaks the mm symmetry of the diffraction pattern but does not produce intensity in the “forbidden” FIG. 5. Freeze-fractured images of Ca2’-ATPase crystals. spots (Fig. 6). The Ca*+-ATPasecrystals were formed in sarcoplasmic reticulum (2 Freeze-fracture electron microscopy of crystals formed in mg of protein/ml) solubilized with Brij 36T ( a , 4 mg/mg protein) or the presence of Brij 36T (4 mg/mg protein, Fig. 5a) or CI2Ea with CI2ER(b, 2 mg/mg protein) in thestandard crystallization (2 mg/mg protein, Fig. 5b) shows poorly resolved clusters of medium. After incubation for 7 days a t 2 “C under nitrogen the particles projecting from the membrane plane and generally samples were fixed with 1%glutaraldehyde for 24 h and processed for freeze-etch electron microscopy as described under “Experimental confirms thestructure derived from frozen-hydrated, secProcedures.” Magnification, X 124,000. tioned, or negatively stained material. After 10-20 days of incubation in suspensions containing2 mg of sarcoplasmic reticulum protein/ml, the crystalline ar-

Microcrystals of Ca2+-ATPase rays account for more than two-thirds of the Ca*+-ATPase content and constitute the dominant structural feature of the specimens. The crystalline aggregates spontaneously settle on the bottom of the tubes leaving a slightly opalescent supernatant that contains not more than 10-30% of the protein content of the startingmaterial. Electron microscope analysis of the negatively stained supernatant shows groups of Ca2+ATPase particles that are occasionally arranged in filamentous structures, but there is no evidence for crystalline order. Based on SDS-polyacrylamide gel electrophoresis, the principal protein component of thesupernatant is the Ca2+ATPase. Sarcoplasmic reticulum solubilized with C12E, (2 mg/mg and protein) in a storage medium containing 0.1 mMCa" 20% glycerol or 20 mM Ca2+but no glycerol loses its ATPase activity after storage for 3 weeks under nitrogen, and crystals cannot be detected. Filamentous aggregates of Ca2+-ATPase molecules are particularly frequent in material stored in the absence of glycerol and may indicate a tendency of the denatured Ca2'-ATPase to associate into long filaments. The Ca2+-ATPase crystalswere routinely formed at 2 "C. Exposure of preformed crystals to 25 "C for periods as short as 2-5 h caused significant disruption, andat 37 "Cno crystals were left after 5 h of incubation. Formation of crystals was never observed at 25 "C. The instability of the three-dimensional crystals a t even moderately high temperatures may be due in part tothermal-mechanical fluctuations that enhance repulsions between the lamellae of the crystalline stacks (Evans andParsegian, 1986). Microcrystals of Ca2+-ATPase in Detergent-solubilized Purified Ca2+-ATPaseor Purified-Delipidated Sarcoplasmic Reticulum Preparations-Negatively stained Ca2+-ATPasecrystals prepared from purified ATPase or from purified-delipidated sarcoplasmic reticulum have essentially identical lamellar repeat distances and unit cell dimensions as those obtained from solubilized sarcoplasmic reticulum vesicles (not shown). Fixation with 1% glutaraldehyde for 24 h at 2 "C of crystals obtained preserves the essential structural features

5291

in all three systems. These observations support the conclusion that the crystalline arrays consist of Ca2+-ATPasemolecules. Optimal crystallization of Ca2+-ATPasein purified or delipidated preparations was obtained at detergentlprotein ratios significantly lower then those used in the case of sarcoplasmic reticulum (Pikula et al., 1988). The presence of residual deoxycholate and the lower phospholipid content of these preparations may contribute to the differences. Extraction of Phospholipidsfrom Glutaraldehyde-fixed Ca2+ATPase Crystuls-Ca2+-ATPase crystals formed in the standard crystallization medium after solubilization with Brij 36T (4 mg/mg protein) were fixed with 1%glutaraldehyde at 2 "C for 24 h with preservation of the crystalline structure. The glutaraldehyde-fixed crystals were subsequently treated in various ways to remove phospholipids, and the effect of lipid removal on the appearance of the crystals was tested by electron microscopy. Exposure of the glutaraldehyde-fixed crystals to CIzEs(20 mglmgprotein) did not cause perceptible change in the crystal structure after 24 h at 2 "C, although under similar conditions the unfixed crystals disintegrated. Phospholipase A from C. durissus (0.01 mg/mg protein in thepresence of 1-4% bovine serum albumin) or phospholipase C from C. welchii (0.04 and 0.3 mg/mg protein) was also without effect on the glutaraldehyde-fixed crystals, although they reduced the phospholipid content of the preparations to 3.4 and 4.8 mol of phospholipid/ mol of ATPase, respectively (Table I). The glutaraldehyde-fixed preparations were extracted with 90% acetone or with chloroform/methanol ( 2 1 and 1:1, v/v) for 24 h at 25 "C, decreasing their phospholipid content to 5.6 and 4.0 mol of phospholipid/mol of ATPase, respectively (Table I). Considerable aggregation occurred during the solventextraction that made the observation of crystals by negative staining difficult, but crystalline Ca2+-ATPasearrays still could be recognized in thin sectioned material. These observations establish that thecontribution of phospholipids to thecrystal structure, in fixed specimens, is minor, but they do not rule out the possible requirement for phospholipids in

TABLEI The lipid content of sarcaplasmie reticulum and purified ATPase preparations The phospholipid content of the preparations was determined as described under "Experimental Procedures," assuming a molecular weight of 109,000 for the ATPase and an ATPase content of 80%.The data are presented as mean S.E. for rz independent assays. The crystals wereformed by incubation of sarcoplasmicreticulum or purified ATPase (2 mg/ml) in the standard incubation medium containing Brij 36T (4 mg/mg protein) for several weeks at 2 "Cunder nitrogen (Preparation 1). The crystalline aggregates (Preparation 2) were separated from the supernatant (Preparation 3) by brief centrifugation at 500 X g. After fixation with 1% glutaraldehyde for 24h the crystalline sediment was extracted three times with large volume of 90% acetone (Preparation 4) or chloroform/ methanol, 1:1 (v/v) (Preparation 5) at 25 "C. Prior to phospholipase treatment the cross-linked aggregates were washed with 0.1 M KCl, 10 mM imidazole, pH 7.0, to remove glutaraldehyde.The treatment with phospholipase AP (C. durissus) was at phospholipase Az concentration of 0.01 mg/mg protein with or without 1-4.6% serum albumin for 1 h at 25 "C (Preparation 6). Phospholipase C (C. welchii) digestion wasperformed at phospholipase concentrations of 0.04-0.3 mg/mg protein for 1-5 h at 25 "C (Preparation 7), essentially as describedearlier (Martonosi et al., 1971). The deoxycholate-extracted samples (Preparation 8) were prepared according to Dean and Tanford (1977). Lipid content Preparation

Sarcoplasmic reticulum ATPase

n

mollmol

1. Standard preparations 2. Crystalline aggregates 3. Supernatant (mother liquor) 4. Crystalline aggregates extracted with 90% acetone 5. Crystalline aggregates extracted with chloroform/methanol 6. Crystalline aggregates after digestion with phospholipase A? 7. Crystalline aggregates after digestion with phospholipase C 8. Deoxvcholate-extracted DreDaration

86.4 + 3.9 34.1 +. 4.1 124.4 f 4.4 5.6 +- 0.6 4.0 & 0.6 3.4 f 0.6 4.8 + 2.5

Purified ATPase ATPase

n

mollmol

6 5 5 3 3 2 2

28.9 & 1.2 17.5 + 0.1 43.0 f 0.1 2.2 & 0.3 2.2 f 0.1

2 2 2 2 2

+ 0.7

3

10.3


5292

the formation andstability of crystals in the absence of fixative. Lucy and Glauert (1964) and others (de Kruijff et al., 1985) observed the formation of elaborate structures in phospholipid-cholesterol-detergent systems by negative staining electron microscopy. Therefore, several control experiments were performed to assess the possible contribution of lipid-detergent structures to the three-dimensional crystals of Ca2+ATPase. Sarcoplasmic reticulum lipids were extracted according to Folch et al. (1957) and suspended in the crystallization media containing 8 mg of C12Es/mlat a final lipid concentration of 2 mg/ml. The pH of the systems was adjusted to pH 6.0, 6.2, 7.0, 7.4, and 7.9, and following incubation at 2 "C for 1, 2, and 3 weeks the samples were analyzed for electron microscopy after negative staining. No periodic structures similar to the CaZ'-ATPase crystals were seen. As a further control, asolectin (10 mg/ml) was dispersed in the standard crystallization medium in the absence of detergents by sonication for 20 min at 25 "C. Single walled and multilayered liposomes of various sizes were observed by negative staining without the typical periodicities seen in the Ca2+-ATPase arrays. Therefore, lipid or lipid-detergent systems of the kind that may have been encountered in the solubilized sarcoplasmic reticulum did not produce the type of ordered arrays that may be mistaken for protein crystals.

of ATPase molecules, as shown schematically in Fig. 7. Contacts between the exposed headgroups may play a role in the association of the lamellae into three-dimensional structures. The high Ca2+(20 mM) and low pH (6.0) required for crystallization are presumed to promote these interactions. The formation of multilamellar arrays is optimal at 2 "C; the crystals rapidly disintegrate at 25 "C without irreversible denaturation of Ca2+-ATPaseand can be reformed again by lowering the temperature. In addition to the effects of temperature on the conformation of Ca2'-ATPase (Pick and Karlish, 1982) and on its interactionswith Ca2+ions (Ikemoto, 1975),the cohesion betweenlamellae in the three-dimensional structures may require some constraint on molecular motions that is achieved at low temperature.The segmental and translational motions of ATPase molecules and the thermal mechanical undulations of the lamellae at 25 "C or higher (Evans and Parsegian, 1986) may impose sufficient stress on

DISCUSSION

The microcrystals formed in detergent-solubilized sarcoplasmic reticulum in the presence of20mM Ca2+ and 20% glycerol represent ordered arrays of Ca2+-ATPasemolecules. This conclusion is supported by the following observations. 1) Identical crystals were obtained using sarcoplasmic reticulum, the purified Ca2+-ATPase, or the purified-delipidated Ca2+-ATPaseas starting materials. 2) The crystalline aggregates formed under standard conditions with ClZE8or Brij 36T as detergents represent more than half of the totalprotein mass. SDS-polyacrylamide gel electrophoresis of the sedimented crystalline aggregates revealed the -109-kDa band of the Ca2'-ATPase as theprincipal protein component. 3) Similar structures were not observed in extracted sarcoplasmic reticulum lipids or in lipid/detergent mixtures. 4) The crystals were preserved in glutaraldehyde-fixed preparations even after theremoval of phospholipids by extraction with organic solvents or by treatment with phospholipases. Therefore, it is unlikely that the arrays would consist of phospholipid-detergent micelles of the types observed by Lucy and Glauert (1964) or de Kruijff et al. (1985). Nevertheless, some contribution by lipids and detergents to the fornation of the crystals cannot be excluded, since even in delipidated ATPase preparations aminimum lipid content of -10 mol of phospholipid/mol of ATPase is usually retained in fact, complete removal of lipids is usually accompanied by denaturation of Ca2+-ATPase(Hidalgo et al., 1986). By electron microscopy of frozen-hydrated, negatively stained, sectioned, or freeze-fractured specimens the crystalline arraysare seen to consist*of superimposed layers of densities that repeat at 130-170 A. In prqjection each of these layers contains regular arrays of -40-A diaqeter particles with two sets of periodicities of -50 and -80 A, respectively. In images of negatively stained or freeze-fractured crystals the lamellae appear as symmetrical structures. The presence of a stain-excluding region in the center of the lamellae suggests that theircore contains a continuous lipid-detergent phase into which the hydrophobic tail portions of Ca2+-ATPase molecules are symmetrically inserted. Both surfaces of the lamellae are covered by the exposed hydrophilic headgroups

ATPase headgroups

Lipid-detergent layer

FIG.7. Schematic representation of the lamellar arrays of Ca"+-ATPasemicrocrystals formed in detergent-solubilized sarcoplasmic reticulum. Shaded areas represent side views of hydrophobic sheets covered with projecting headgroups of Ca2+ATPase molecules represented by open circles.

b)

170

Lo

FIG.8. Diagram illustrating one possible model for the packing of Ca"+-ATPasemolecules in the new crystal form. In this model, ATPase dimers are located at the corners and at the center of an orthogonal unit cell having cell dimensions of a = 164.4 A and b = 55.5 A. A 2-fold axis parallel to the"b" unit cell axis relates the two ATPase molecules within the dimer. Screw diads at (a/4 and 3a/4 arise due to the combination of the 2-fold axis with the face centering. a, view of the model from a direction normal to the plane of the two-dimensional crystalline array. b, view looking down the short or "b" axis of the unit cell. This particular view also has the unique feature of being down the 2-fold rotation axis.

5293


the structure to cause separation between and disorder within the “b” axis. A packing model in which the 2-fold axis runs parallel with the “b”cell axis is illustratedin Fig. 8. The area the lamellae. Cross-linking of Ca”-ATPase crystals with glutaraldehyde of the unit cell in these crystals is 9113 A*. This compares protects the structure against conditions (low Ca’+, high pH, with 7371 A* for the vanadate-induced crystalswhich contain elevated temperatures, Salyrgan,high concentration of deter- two ATPase molecules/unit cell (Taylor et al., 1984) and with 3070-3560 A’ for lanthanum-induced crystals which contain gents, etc.) that disrupt unfixed preparations. The interlamellar distances as well as the organizationof ATPase parti- one ATPase molecule/repeating unit (Dux et al., 1985). Both cles within the lamellae are stabilized by fixation; therefore, of these latter crystal forms can be induced in native sarc9’ glutaraldehyde must form cross-links notonly between adja- plasmic reticulum membranes. The unit cell area of 9113 A cent ATPase molecules in the same lamellae (intralamellar is large enough t o give a reasonable but dense packing to the cross-links) but also between ATPase molecules in adjacent four ATPase molecules required for the c12 plane group, but lamellae (interlamellar cross-links). The formationof inter- too small for the eight molecules that would be required for lamellar cross-links confirms the close approach and probable two-sided plane group c222, if there is be any space allotted interaction between ATPase molecules located in neighboring for a lipid-detergent phase. The 1840 A2/ATPase molecule is less than that observed in the two crystal forms induced in lamellae that was inferred from electronmicrographs. The model outlined in Fig. 7 implies that the multilamellar native membranes. This difference is probablydue to the stacksconstitutingtheCa*+-ATPasecrystalsare held to- molecular packing, which symmetrically distributes thebulky sides of the gether by hydrophobic and polar interactions involving both cytoplasmic domains of the Ca2+-ATPase on both proteins and phospholipids. Similar structures were obtained lipid-detergent phase. For comparison, the cross-sectional previously from purple membranes (Henderson and Shotton, area of the lipid bilayer domain of individual ATPase molethree-dimensionalreconstructions of 1980), mitochondrial cytochrome oxidase (Vanderkooi et al., culesmeasuredfrom 1972; Ozawa et al., 1982), ubiquino1:cytochrome c reductase images of frozen hydrfted vanadate-induced crystallinememof Neurospora crassa (Weiss et al., 1985), and from thelight- branes is about1200 A’ (Taylor et al., 1986b). harvesting chlorophylla / b protein (McDonnel and Staehelin, REFERENCES 1980; Li and Hollingshead,1982). These structuresmay differ from genuine crystals of membrane proteins such as porin Blasie, J. K., Herbette, L., and Pachence, J. (1985) J . Membr. Biol. 86,l-7 (Garavito et al., 1983) or the photosynthetic reaction center Brandl, C. J., Green, N. M., Korczak, B., and MacLennan, D. H. (Deisenhofer et al., 1985) in which the detergents and small (1986) Cell 4 4 , 597-607 amphiphiles serve mainly to coat the hydrophobic protein Brandl, C. J., de Leon, S., Martin, D. R., and MacLennan, D. H. surfaces and the primary force for crystallization is the inter- (1987) J. Biol. Chem. 262,3768-3774 Dean, W. L., and Tanford, C . (1977) J . Biol. Chem. 2 5 2 , 3551-3553 action between the hydrophilicregions of theproteins (Michel, 1983). Experiments are in progress forthe production Dean, W. L., and Tanford, C . (1978) Biochemistry 17, 1683-1690 of stablethree-dimensionalcrystals from fully delipidated Deisenhofer, J., Epp, O., Miki, K., Huber, R., and Michel, H. (1985) 318,618-624 Ca*+-ATPase using detergent-small amphiphile combinationsdeNature Kruijff, B., Cullis, P. R., Verkleij, A. J., Hope, M. J., van Echteld, (Michel, 1983; Deisenhofer et al., 1985). C. J. A., and Taraschi, T. F. (1985) in The Enzymes of Biological The crystallization conditions reportedhere have produced Membranes (Martonosi, A., ed) 2nd Ed.,Vol. 1, pp.131-204, stacks of two-dimensional crystals. Although it is possible Plenum Publishing Corp., New York that these crystals are indeed three-dimensionalcrystals, we de Meis, L., and Vianna, A. L. (1979) Annu. Reu. Biochem. 48, 275can detectso far no evidence of a regular packing in the third 292 J., Lepault, J., Freeman, R., Berriman, J. A,, and Homo, dimension. Disruptionof the large stacks invariablyproduces Dubochet, J.-C. (1982) J . Microsc. ( 0 x f . j 1 2 8 , 219-237 shorter stacks of two-dimensional crystals, which in projec- Dux, L., and Martonosi, A. (1983a) J. Biol. Chem. 258, 2599-2603 tion show evidence of moire patterns indicative of an azi- Dux, L., and Martonosi, A. (1983b) J . Biol. Chem. 258, 10111-10115 muthalrotation betweenopposed membranesheets.This Dux, L., and Martonosi, A. (1983~) J . Biol. Chem. 258, 11896-11902 rotation may occur at discrete angles, which would indicate a Dux, L., and Martonosi, A. (1983d) J. Biol. Chem. 258, 11903-11907 regular patternof interactions between membranes. This type Dux, L., and Martonosi, A. (1984) Eur. J. Biochem. 141,43-49 L., Taylor, K. A., Ting-Beall, H. P., and Martonosi, A. (1985) of interaction would produce three-dimensional crystals hav- Dux, J . Biol. Chem. 2 6 0 , 11730-11743 ing the commonly occurring coincidence lattice defects. The Dux, L., Pikula, S., Mullner, N., and Martonosi, A. (1987) J . Biol. ribosome crystals described by Unwin and Taddei (1977) and Chem. 262,6439-6442 the bacteriorhodopsin crystals reported by Henderson and Evans, E. A., and Parsegian, V. A. (1986) Proc. Natl. Acad. Sci. U. S. A . 83,7132-7136 Shotton (1980) are examples that exhibit this type of disorder. At present the diffraction resolution obtainable from the two- Folch, J., Lees, M., and Sloane Stanley, G. H. (1957) J. Biol. Chem. 226,497-509 dimensional crystalline arraysis insufficient to determine the Garavito, R. M., Jenkins,J., Jansonius, J. N., Karlsson, R., and possible rotations that may occur. Rosenbusch, J. P. (1983) J. Mol. Biol. 1 6 4 , 313-327 The two-sided plane group determined here (c12 or 12222) Henderson, R., and Shotton, D. (1980) J . Mol. Biol. 139,99-109 would require that ATPase dimers be present in the crystals. Hidalgo, C.,De La Fuente, M., and Gonzalez, M. E. (1986) Arch. Biochem. Biophys. 2 4 7 , 365-371 However, unlike the ATPase dimers observed in vanadateinduced crystalsformedinnative sarcoplasmic reticulum Holser, W. T. (1958) 2. Kristallogr. 1 1 0 , 266-281 Ikemoto, N. (1975) J . Biol. Chem. 250, 7219-7224 membranes, which are relatedby a 2-fold rotation axis normal Inesi, G., and de Meis, L. (1985) in The Enzymes of Biological et al., 1984, 1986a,1986b), to the membrane plane (Taylor Membranes (Martonosi, A., ed) 2nd Ed., Vol. 3, pp. 157-191, the dimers in the new crystal form are related by a 2-fold Plenum Publishing Corp., New York rotationaxiswithinthemembraneplane. At present, we Li, J., and Hollingshead, C. (1982) Biophys. J. 37, 363-370 cannot determine with certainty which of the two cell axes Lucy, J. A., and Glauert, A. M. (1964) J . Mol. Biol. 8,727-748 contain the 2-fold rotation axis. One model having the 2-fold MacLennan, D. H., Brandl, C. J., Korczak, B., and Green, N. M. (1985) Nature 3 1 6 , 696-700 rotation axis parallel with the “a” cell axis can be excluded, Martonosi, A., Donley, J. R., Pucell, A. G., and Halpin, R. A. (1971) because the size of the transmembranedomain- of the ATPase Arch. Biochem. Biophys. 1 4 4 , 529-540 molecule is too large in relation to the 55.5-A dimension of McDonnel, A., and Staehelin, L. A. (1980) J. Cell Biol. 8 4 , 40-56

to

5294


Michel, H. (1983) Trends Biochem. Sci. 8,56-59 Ozawa, T., Tanaka, M., and Wakabayashi, T. (1982) Proc. Natl. Acad. Sci. U. S. A . 7 9 , 7175-7179 Peracchia, C., Dux, L., and Martonosi, A. (1984) J. Muscle Res. Cell Motil. 5 , 431-442 Pick, U., and Karlish. S. J. D. (1982) J. Biol. Chem. 257.6120-6126 Pikula, S., Mullner, N., Dux, L., and Martonosi, A. (1988) J. Bwl. Chern. 263,5277-5286 Tanford, C. (1984) CRC Crit. Rev. Biochem. 1 7 , 123-151 Taylor, K. A., Dux, L., and Martonosi, A. (1984)J. Mol. Biol. 1 7 4 , 193-204

Taylor, K., Dux, L., and Martonosi, A. (1986a) J . Mol. Biol. 1 8 7 , 417-427 Taylor, K. A., Ho, M. H., and Martonosi, A. (1986b) Ann. N.Y. Acad. Sci. 483,31-43 Unwin, N., and Taddei, C. (1977) J. Mol. Biol. 114,491-506 Vanderkooi, G., Senior, A. E., Capaldi, R. A., and Hayashi, H. (1972) Bwchim. Biophys. Acta2 7 4 , 38-48 Weiss, M., Perkins, S. J., and Leonard, K. (1985) in The Enzymes of Biological Membranes (Martonosi, A., ed) 2nd Ed., Vol. 4, pp. 333346, Plenum Publishing Corp., New York