Resolution of the V1 ATPase from Manduca sexta into subcomplexes and visualization of an ATPase-active A3B3EG complex by electron microscopy.

The effect of the ATPase activity of Manduca sexta V(1) ATPase by the amphipathic detergent lauryldimethylamine oxide (LDAO) and the relationship of these activities to the subunit composition of V(1) were studied. The V(1) was highly activated in the presence of 0.04-0.06% LDAO combined with release of the subunits H, C, and F from the enzyme. Increase of LDAO concentration to 0.1-0.2% caused the characterized subcomplexes A(3)B(3)HEGF and A(3)B(3)EG with a remaining ATPase activity of 52 and 65%, respectively. The hydrolytic-active A(3)B(3)EG subcomplex has been visualized by electron microscopy showing six major masses of density in a pseudo-hexagonal arrangement surrounding a seventh mass. The compositions of the various subcomplexes and fragments of V(1) provide an organization of the subunits in the enzyme in the framework of the known three-dimensional reconstruction of the V(1) ATPase from M. sexta (Radermacher, M., Ruiz, T., Wieczorek, H., and Grüber, G. (2001) J. Struct. Biol. 135, 26-37).

from the V O part as an in vivo regulatory mechanism (10), is the object of our studies and comprises the eight subunits A, B, H, C, D, E, G, and F with apparent molecular masses of 67, 56, 54, 40,32,28,14, and 16 kDa, respectively (11). Low resolution structural studies of this V 1 complex using small-angle x-ray scattering have shown that the hydrated enzyme is an elongated molecule. The x-ray data define a mushroom-shaped V 1 ATPase, which consists of an ϳ145 Å headpiece, joined by an elongated stalk (8). Image processing of electron micrographs of negatively stained V 1 (9,12,13) has revealed that the headpiece consists of a pseudo-hexagonal arrangement of six masses surrounding a seventh mass. These barrel-shaped masses of approximately 30 Å in diameter and 80 Å in length, which consist of the major subunits A and B, are arranged in an alternating manner (9). The hexagonal barrel of subunits A and B encloses a cavity of ϳ32 Å in which part of the central stalk is asymmetrically located. The stalk protrudes from the bottom side of the headpiece forming an angle of ϳ7°with the vertical axis of the molecule. At the upper end of the hexagonal barrel extensions can be observed, assumed to belong to the N termini of subunit A (9,13). Further insights into the topology of the M. sexta V 1 ATPase were obtained by differential protease sensitivity, release by chaotropic agents (13), and cross-linking studies (13,14). These studies resulted in a model in which the subunits H, C, D, G, and F are exposed in the enzyme, whereas subunit E is shielded in the complex (6,13,14).
Here we report an investigation of the structure-function relationship of the V 1 stalk subunits in M. sexta using the detergent LDAO. 1 We show that the detergent liberates a highly active A 3 B 3 DEG complex and various hydrolytic-active subcomplexes, A 3 B 3 HEGF, A 3 B 3 HDEG, and A 3 B 3 EG. Electron microscopy has been used to visualize directly the two-dimensional structure of the A 3 B 3 EG subcomplex demonstrating a hexagonal modulation surrounding a central cavity with an interior mass.
(ROTH) and Pefabloc SC (final concentration of 8 mM; BIOMOL). (ii) Only the fractions that eluted from the Mono-Q HR 10/10 column (Pharmacia) at 250 -280 mM NaCl and contained the complete V 1 complex, as judged from SDS-PAGE, were pooled, concentrated on a Centricon 100K concentrator (Schleicher & Schuell), and applied on a Sephacryl TM S-300 HR column (10/30, Pharmacia) instead of a Superdex HR200 (10/30) column as described previously (15). The V 1 complex eluted in a single symmetrical peak containing the eight subunits A-H. The purity and homogeneity of the protein sample was analyzed by Native-PAGE (16) and SDS-PAGE (17). SDS gels were stained with Coomassie Brilliant Blue G250, Blue R250 (18), or with silver (19). Protein concentrations were determined by the bicinchonic acid assay (Pierce). ATPase activity, which is stimulated by Ca 2ϩ , was performed in the presence of an ATP-regenerating system as described by Lötscher et al. (20).
Isolation of V 1 Subcomplexes-Dissociation of the V 1 complex was induced by incubation of the enzyme in buffer A (20 mM Tris/HCl (pH 8.1) and 150 mM NaCl) with or without the addition of 9.6 mM 2-mercaptoethanol and different concentrations (0.05 to 0.5% (w/v)) of LDAO. The sample was shaken for 1 h on ice and then applied at a flow rate of 0.5 ml/min to a Sephacryl TM S-300 HR column (10/30, Pharmacia) equilibrated with buffer A with or without the addition of 2-mercaptoethanol and detergent. Fractions containing pure subcomplexes were pooled and reapplied to the same Sephacryl column equilibrated with buffer A and 9.6 mM 2-mercaptoethanol but without LDAO, to remove the detergent.
Electron Microscopy and Two-dimensional Image Analysis-For electron microscopy the protein was diluted in 20 mM Tris/HCl (pH 8.1) and 150 mM NaCl to 20 -40 g/ml. The sample was applied to 400 mesh carbon-coated copper grids and deep stained with uranyl acetate. Micrographs were recorded at a calibrated magnification of 58 300ϫ on a Philips CM 120 electron microscope under low electron dose conditions (10 e Ϫ /Å 2 ). The negatives were scanned on a flat-bed SCAI (Zeiss) microdensitometer with 7 m pixel size. The images were reduced by binning to a final pixel size of 21 m, corresponding to 3.6 Å on the scale of the specimen. Particles were selected from the micrographs using the sole selection criterion that they were far enough apart from their neighbors such that no overlap occurred in the 0°image. For image processing the SPIDER software (21), version 5.0 with extension was used. Images of 2132 particles were windowed from 11 micrographs and normalized in contrast with the outer area of each image, excluding a round masked area around the center (9). The particles were translationally/rotationally aligned using a combination of correlation methods, starting with a first reference created by reference-free alignment (22), followed by reference-based alignments as previously described for the V 1 ATPase (9). After each alignment correspondence analysis was applied (23) and the images were classified using Diday's method of moving centers (24) followed by hierarchical ascendant classification. In each step 6 class-averages were created and used as reference for a multi-reference alignment. For each correspondence analysis classification, 100 iterations were performed. All alignments were based on cross-correlations of two-dimensional Radon transforms (25). The resolutions of the final classes were determined using the Fourier ring correlation (26,27) with a cut-off criterion of five times the noise correlation (FRC 5 ) (28).

Effect of LDAO on the V 1 ATPase Activity-The
Ca 2ϩ -ATPase activity of the isolated V 1 was about 1.7 Ϯ 0.2 mol of ATP hydrolyzed/min/mg when performed in the presence of an ATPregenerating system. The effect of various concentrations of LDAO on the ATPase activity of V 1 is shown in Fig. 1. The presence of 0.04 -0.08% (w/v) of this detergent causes a stimulation of Ca 2ϩ -ATPase activity, with a maximal activity of 3.2 Ϯ 0.3 mol ATP hydrolyzed/min/mg. The activity of V 1 in 0.1 and 0.2% LDAO was 52 and 65%, respectively, compared with the activity of the untreated enzyme. Higher concentrations of LDAO (0.4 and 0.5% (w/v)) inhibited the enzyme to less than 10% of total ATPase activity. When the enzyme was applied onto a Native-PAGE in the presence of 0.4% or 0.5% of detergent two new bands (Fig. 1B, I and II) with higher molecular mass were obtained. Both bands were cut out from the gel, destained, and subjected to SDS-PAGE (Fig. 1C), indicating that both contain a V 1 subcomplex, consisting of the subunits A, B, D, E, and G and forming oligomers. The ratio of the protein band of V 1 relative to the bands I and II was 52:24:25, respectively, based on the quantitation of the staining intensity of the three bands. Therefore, the data imply that the inhibition of ATPase activity is mainly caused by the presence of higher LDAO concentration (see below).
Resolution of V 1 into Subcomplexes-To establish whether the changes of ATPase activity might be caused by the resolution of the enzyme into subcomplexes, size-exclusion chromatography has been performed in the presence of LDAO. V 1 , incubated with 0.05% LDAO and 2-mercaptoethanol, was applied to a Sephacryl S-300 HR column and a major peak eluted at about 18 min (Fig. 2, panel A) and several smaller ones after 44 min. The fractions of the major peak were pooled and subsequently reapplied onto the same column in absence of detergent. The protein eluted as an active V 1 complex (3.0 mol ATP hydrolyzed/min/mg) without the subunits H, C, and F as shown by SDS-PAGE (Fig. 2, panel A).
Application of V 1 in the presence of 0.1% LDAO and 2-mercaptoethanol to a Sephacryl S-300 HR column results in a major peak accompanied by a shoulder (24 min) (Fig. 2, panel  B). Detergent of the pooled fractions of the major peak was removed as described above. The eluted protein with an ATPase activity of 0.9 Ϯ 0.1 mol ATP hydrolyzed/min/mg contains a six-subunit enzyme composed of the polypeptides A, B, H, E, G, and F (Fig. 2, panel B).
Several lines of evidence indicate that redox-modulation of the V-ATPase, proposed as a mechanism of regulation of this complex (29 -31), leads to structural changes in the enzyme (3,29,32). In this connection we incubated V 1 ATPase with 0.1% LDAO in the absence of the reducing agent, 2-mercaptoethanol. Size-exclusion chromatography of this mixture, which was done as described above, resulted in an exclusion diagram with a main peak at 21 min and several minor peaks at 53 min (Fig. 2, panel C). SDS-PAGE of the main peak, which was freed of detergent by chromatography in the presence of 2-mercaptoethanol, yielded an A 3 B 3 EG subcomplex (Fig. 2, panel C) with an ATPase activity of 1.0 Ϯ 0.1 mol ATP hydrolyzed/min/mg.
To study the inhibitory effect of high LDAO concentrations (Ն 0.5%) as described above in more detail, V 1 was incubated with 0.5% detergent plus 2-mercaptoethanol and applied to the column. The protein eluted as a broad peak. Detergent was removed by a subsequent column step (Sephacryl S-300 HR), and three peaks were eluted (Fig. 2, panel D). The first peak contained a V 1 complex without the subunits C and F, followed by a subcomplex consisting of the subunits A, B, and E in a stoichiometry of 1.8:1:1, based on the quantitation of the staining intensity of the three bands of theses subunits, respectively (Fig. 2, panel D). The ATPase activity of the V 1 (-C,F) complex was 0.8 mol ATP hydrolyzed/min/mg, whereby the A 2 BE subcomplex had lost its enzyme activity. Peak three contained only detergent. Thus, the V 1 ATPase is mainly inhibited by the presence of 0.5% LDAO (Fig. 1B) and dissociates in subcomplexes after removing of the detergent.
Electron Microscopic Characterization of the A 3 B 3 EG Subcomplex-Electron micrographs of the negatively stained A 3 B 3 EG subcomplex show a homogenous distribution of this complex (Fig. 3). A visual representation of map factor 1 versus 2 of the first correspondence analysis of the A 3 B 3 EG subcomplex (Fig. 4) reveals the variation of the subcomplex. Particles with a well pronounced pseudo-hexagonal arrangement and a centrically seventh mass can be observed (indicated by arrows in Fig. 4), as well as particles, which are most probably slightly tilted, with an elongated seventh mass toward the bottom and left side of the map.
Three averages of the A 3 B 3 EG subcomplex calculated from 501, 490, and 456 particles (Fig. 5) show a pseudo-hexagonal arrangement of six protein densities with an overall diameter of ϳ115 Å. This is in agreement with structural data of the V 1 ATPase determined recently by electron microscopy (13), which likewise showed six major masses of density that were grouped in a pseudo-hexagonal arrangement with a diameter of ϳ130 Å. The center of the particle shows a seventh mass, which would be consistent with the presence of a central stalk. Fig. 5 (panel II) shows an average in which an additional density can be observed near one of the outer densities. This may be caused by small tilts of the A 3 B 3 EG subcomplex perpendicular to its hexagonal axis, sufficient to displace the central mass without apparent distortion of the outer hexagon. In the previous threedimensional study of the V 1 ATPase (9) it was observed that many particles exhibited an inclination of about 20°relative to the on-axis view. The resolution of the three averaged images was 29 Å, as determined by Fourier ring correlation FRC 5 (28) (Fig. 5B).  (panel A, B, and D). Afterward, the samples were applied onto a Sephacryl TM S-300 HR column in the presence of detergent. B, the indicated (f) fractions were pooled and reapplied to a Sephacryl TM S-300 HR column in the absence of LDAO but presence of 2-mercaptoethanol (panels A-C). Panel D shows the elution diagram after the second column step in the absence of LDAO. The detergentfree samples were subjected to SDS-PAGE and stained either with Coomassie Blue R250 (panel A, blot B), G250 (panel A and C), or with silver (Ref. 19, panel D).

DISCUSSION
The results presented here show that addition of various amounts of the zwitterionoic detergent LDAO to the V 1 ATPase from M. sexta induces the dissociation of the enzyme into the subcomplexes A 3 B 3 DEG, A 3 B 3 HEGF, and A 3 B 3 EG with altered rates of ATPase activity. Below the critical micelle concentration of 0.05% in a buffer Ն pH 7.0 (33) the monomeric detergent leads to release of the subunits H, C, and F combined with an enhancement of ATP-hydrolytic activity. By assaying the A 3 B 3 DEG subcomplex after chromatography in the absence of LDAO, where activation of ATPase activity was essentially the same, it was possible to conclude that the enhancement is either due to the release of an ensemble of the subunits H, C, and F or to the removal of one of these three polypeptides (see below).
Increasing LDAO concentrations to 0.1% shows that the reduced V 1 can be converted into an A 3 B 3 HEGF complex, whereby under oxidizing conditions the complex dissociates further into a smaller subcomplex composed of the subunits A, B, E, and G. Most recent studies on the oxidized and reduced V 1 ATPase from M. sexta have also shown that the subunits H and F are less accessible to proteolytic digestion in the reduced form (32). This is in line with the data presented in which the presence of 0.1% LDAO leads to the dissociation of these two subunits only under oxidizing conditions. The ATPase activity of the A 3 B 3 HEGF and A 3 B 3 EG subcomplex after detergent removal is about 52 and 58% of that of V 1 ATPase, respectively (see above). This observation that both complexes lacking subunit D still retain a high ATPase activity is important, because this subunit has been proposed as a structural and functional homologue of the ␥ subunit of F-ATPases (34,35). Recently, prolonged digestion of the M. sexta V 1 with trypsin resulted in an active enzyme lacking subunit D (13). This is consistent with a previous report of Xie (36), which suggested that subunit D is not essential for ATP hydrolysis in the bovine enzyme. However, the resulting hydrolytic-active A 3 B 3 EG-domain of M. sexta forms what can be called a "core" complex (see Fig. 6). For the H ϩ -ATPase of clathrin-coated vesicles it has been demonstrated that at least the four subunits A, B, C, and E are necessary for ATP hydrolysis (37,38). In these studies recombinant subunits E and C were reconstituted with a biochemically prepared A-B complex, resulting in an A 3 B 3 E subcomplex with reproducible reconstitution of ATPase activity (37) and a 6-fold stimulation of hydrolytic activity when recombinant C was added to the A 3 B 3 E complex. Addition of subunit C to the A-B subcomplex resulted in slight activity (38). Most recent studies reveal that subunit C (39 -41) together with subunit H (42,43) might play a role in bridging the V 1 and V O domains rather than acting as a core subunit (40) (Fig. 6). Together with our data this reinforces the hypothesis that at least the subunits A, B, and E might constitute part of the minimal ATPhydrolytic core of V 1 . The presence of both stalk subunits G and F in the A 3 B 3 E complex is not surprising because of their close proximity in the enzyme (Fig. 6) (9). The model is based on the combination of presented biochemical and structural data and topology studies described recently (13,14). Because a nucleotide-dependent arrangement of subunit E has been observed in V 1 (13) the model presented describes the subunit topology in the presence CaATP, in which E-G and E-F cross-linked products can be formed. The position of subunit C (dark gray) bases on most recent electron micrographs of a V 1 subcomplex of the C. fervidus V-ATPase (41) and the V 1 (-C) complex from M. sexta (9), indicating that subunit C is located at the bottom of the central stalk (9,41) and connected to the center of V O (41). vesicles (44), and M. sexta (13). Furthermore, an arrangement of the subunits A, B, E, and G close to each other is in accordance with the formation of an A 2 BEG subcomplex of M. sexta upon treatment with chaotopic iodide, but without hydrolytic activity (13,45). Notably, from recent studies using a bifunctional cross-link reagent it has been proposed that subunit E (Vma4p), with an apparent molecular mass of 26 kDa, is located at the outer surface of the A 3 B 3 hexamer in the V 1 V O ATPase from yeast (46). Close inspection of the recently determined three-dimensional structure of the V 1 ATPase from M. sexta (9) clearly indicates that no additional mass is located at the outer surface of the A 3 B 3 hexamer of V 1 alone.
However, a direct correlation between the dissociation of the individual subunits H, C, and F forming different subcomplexes and the change of ATPase activity is not clear at the moment. Nevertheless, the increase of hydrolytic activity after removal of the three subunits H, C, and F implies that it needs such an ensemble of subunits, whose simultaneous dissociation might lead to structural and thereby functional alterations in the remaining subcomplex.
To date the best structural model for M. sexta V 1 ATPase is an 18 Å three-dimensional reconstruction obtained from singleparticle images of the molecules embedded in deep stain (9). The V 1 model consists of a headpiece of 130 Å in diameter, with the six major subunits A and B alternating around a cavity and a compact central stalk. Inside the cavity the stalk can be seen connected to only two of the major subunits (9). Zero-length cross-linking studies (13), photo-affinity labeling (14) as well as the formation of an A 2 BE complex (see Fig. 2D) indicated that the catalytic A subunit of V 1 is close to subunit E (Fig. 6), which is proposed to be located in the cavity of the A 3 B 3 hexamer (9,13,14). The presented two-dimensional averages of the A 3 B 3 EG subcomplex from M. sexta determined from negatively stained specimens reveals a hexagonal modulation as seen previously in projection images of V 1 molecules in stain (13,21) and are interpreted as three copies each of the nucleotidebinding subunits A and B (9). The two-dimensional analysis of the A 3 B 3 EG domain shows also that the A 3 B 3 hexamer is occluded by a seventh mass centrally to the hexamer. This hexagonal modulation of the A 3 B 3 hexamer with an internal seventh mass is remarkably similar to the subunit arrangement of the closely related F 1 complex. Cryo-electron micrographs of F 1 from Escherichia coli that were labeled with Fab fragments of monoclonal antibodies (47) or with monomaleimido gold (48 -50) have shown that the additional seventh mass includes the central stalk subunits ␥ and ⑀.
In summary, the amphipathic detergent LDAO is a useful tool for examining the composition of the ATP-hydrolytic core of the M. sexta V 1 ATPase. The data presented show that at least the subunits A, B, E, and G are essential for ATP-hydrolysis in this enzyme. Previous experiments (13,14) on the interactions between the different subunits in the M. sexta V 1 using differential protease sensitivity and cross-linking studies and the results presented here suggested that subunit E, from which no isoform exists in Manduca V-ATPase (51, 52), contributes to the coupling element in the V 1 ATPase of M. sexta.