ARCHIVES

Vol.

OF BIOCHEMISTRY

299, No. 1, November

AND

BIOPHYSICS

15, pp. 1055109,

1992

Monomaleimidogold Labeling of the y Subunit of the Escherichia co/i F, ATPase Examined by Cryoelectron Microscopy Stephan Wilkens’ Institute

Received

and Roderick A. Capald?

of Molecular Biology, University

May

20, 1992, and in revised

form

of Oregon,

July

Eugene, Oregon 97403

7, 1992

A novel approach for locating sites of interest in a protein complex has been developed using monomaleimidonanogold (MMN). The Escherichia coli F, ATPase, when prepared without the 6 subunit, contains only a single reactive cysteine on one of the three copies of the (Ysubunit. This site was reacted with MMN and the gold cluster visualized on the protein complex by cryoelectron microscopy. Additional sites for modification with MMN were added by introducing cysteine residues through sitedirected mutagenesis. Labeling of two mutants, $%3-C and yT106-C, in which Set-8 and ThrlO6, respectively, had been replaced by a cysteine, placed the gold cluster on the central mass that is seen in the hexagonal projection of the ECFl complex. The results establish that the central mass contains the N-terminal part of the y subunit. ‘~1 1992 Academic Press, Inc.

Electron microscopy provides a useful approach for studying the structure of multisubunit complexes (e.g., l-3). We have used this method to examine the ATP synthase (ECFiF,,)” of Escherichia coli (4-7), a complex of eight different subunits, five of which, the cu, p, y, 6, and t subunits, are present in the molar ratio of 3:3:1:1:1 in the ECF, part (8-10). The three other subunits, a, b, ’ Permanent address: Fritz-Haber-Institut der Max-Planck-Gesellschaft, Abteilung Elektronenmikroskopie, Faradayweg 4-6, 1000 Berlin 33, Germany. ’ To whom correspondence should be addressed. Fax: (503) 346-5891. 3 Abbreviations used: ECF,, Escherichia coli F, ATPase; MMN, monomaleimidonanogold; DMAB, borane-dimethylamine; DMSO, dimethyl sulfoxide; Mops, 4-morpholinepropanesulfonic acid; EDTA, ethylenediaminetetraacetic acid; DTT, DL-dithiothreitol; CM, N-[4-[7(diethylamino).4-methylcoumarin-3.yllmaleimide]; NaDodSOl PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis; LDAO, N,Ndimethyldodecylamine-N-oxide; NEM, N-ethylmaleimide; mAbs, monoclonal antibodies; TrisHCl, tris(hydroxymethyl)aminomethane hydrochloride. 0003-9861/92

$5.00

Copyright c> 1992 hy Academic Press, Inc. All rights of reproduction in sny form reserved.

and c, make up the F,, part in the stoichiometry of 1:2:10-12, and the molecular weight of the entire complex is close to 530,000 (reviewed in 8-10). We have obtained a low resolution (25 A), three-dimensional structure of the ECF, part by electron microscopy of negatively stained two-dimensional arrays using image reconstruction methods (4). The ECF, part and the intact ECF,FO complex have also been examined by cryoelectron microscopy (4-7). Fab fragments generated from mAbs specific to the various subunits have been used to immunolabel the enzyme, and the location of epitopes on several of the subunits of ECFi identified within the protein structure in this way (5, 7). The use of antibodies to localize subunits or domains within subunits by electron microscopy depends on the antigenicity of the regions of interest, and is limited by the fact that Fabs are relatively large, which can prevent their entry into cavities in the protein complex of interest. As an alternative approach, we are examining the potential of electron dense gold clusters for labeling sites of interest in proteins. Gold clusters of fixed size (14 A diameter as compared to 25 to 50 A for an Fab fragment) attached to a maleimide are now commercially available (monomaleimidonanogold from Nanoprobes, Inc., MMN reagent). We are conducting studies in which cysteine residues (Cys) are being introduced into subunits of the ECF,F, complex by site-directed mutagenesis (e.g., 11). These introduced Cys residues can be used to link a variety of probes including crosslinking reagents, fluorescent compounds, as well as the gold cluster maleimides. Here we describe experiments in which Cys residues were introduced into the y subunit, these sites then reacted with MMN, and the protein-gold conjugates examined by cryoelectron microscopy. Our results show the usefulness of the labeling approach and establish that the N-terminal part of the y subunit constitutes the central mass seen in the hexagonal projection of the ECF, complex. 105

106 EXPERIMENTAL

WILKENS

AND

PROCEDURES

Materials. Monomaleimidonanogold was obtained from Nanoprobes, Inc., and borane-dimethylamine complex (DMAB) was purchased from Aldrich, Inc. The two mutants used in this study, yS8-C and yTIO6-C, were kindly provided by Dr. R. Aggeler and will be described in detail elsewhere. ECF, was isolated from these mutants according to Gogol et al. (4). The enzyme from both of the mutants had ATPase activity in the same range as wild-type enzyme. Labeling of cyst&e residues with monomaleimidonanogold. Each MMN batch was dissolved in 100 ,ul DMSO and stored in 20-&l aliquots in liquid nitrogen prior to use. The concentration of different MMN batches varied between 16 and 30 nmol/batch. Twenty microliters of the MMN solution was added to ECFl (3-4 PM) in 200 ~150 mM Mops (pH 7.0), 10% glycerol, 1 mM EDTA, 2 mM ATP, and 2 mM DMAB. The final MMN concentration during labeling varied from 15 to 28 PM. After incubation for 45 min at room temperature, 1 mM NEM was added to terminate the reaction. In some experiments, the solution was incubated with 30 ELM N-[4-[7-(diethylamino)4-methylcoumarin+ yl]]maleimide (CM) prior to the NEM addition to estimate the extent of labeling by incorporation of fluorescence into subunits as determined by NaDodSO, PAGE. Excess label was removed by gel permeation chromatography on Sephacryl S200 (30 X 0.7 cm) in 50 mM Tris-HCl (pH 7.5), 20% glycerol, 1 mM EDTA, 2 mM ATP, and 2 mM DMAB at 4°C and a flow rate of 1 ml/h. Fractions of 0.25 ml were collected, and the peak fraction was used for cryoelectronmicroscopy. ECFi (4-S mg) in 900 ~150 mM Tris-HCl Preparation of ECF, f-8). (pH 7.5), 20% glycerol, 1 mM EDTA, 2 mM ATP, 1 mM DTT, and 0.2% NJ-dimethyldodecylamine-N-oxide (LDAO) was chromatographed on Sephacryl S200 (45 X 1 cm) in the same buffer but with 0.1% instead of 0.2% LDAO at 4’C and at a flow rate of 1.6 ml/h. About 90% of the activity was pooled and concentrated to approximately 1 ml with a Centricon 10 microconcentrator (Amicon). The concentrated sample was applied to a second Sephacryl S200 column (45 X 1 cm) in the same buffer as used for the labeling reaction to remove the LDAO. Removal of the 6 subunit was confirmed by NaDodSOl PAGE. Cryoelectron microscopy. MMN-labeled ECF, was diluted to approximately 35540 Kg/ml (0.1 PM) in 50 mM Tris-HCl (pH 7.5), 137 mM NaCl, 1 mM EDTA, 2 mM ATP, and 2 mM DMAB and incubated with 0.5 HIM Fab fragments derived from the olI mAb (5) for l-2 h at room temperature. Freezing, storage, and loading of grids was done essentially as described in Gogol et al. (4,5). A Philips CM12 transmission electron microscope equipped with a Gatan cold stage and a self built anticontaminator device was used. Micrographs were taken in low dose mode at a magnification of X60,000 with an estimated electron dose of 12 e / A’ and at an underfocus of PO.9 pm. Image processing. Selected areas of micrographs were scanned with an Optronics drum densitometer with 25 Frn step-size, corresponding to 4.2 A pixel size at the specimen level. The pixel size was calibrated by calculating the Fourier transform of the l/23 A spacing of negatively stained tobacco mosaic virus at the same magnification. Image processing, including selection of single images, low-pass-filtering, and single reference alignment, was done using the SPIDER software package (12) as described by Gogol et al. (4,5). Eigenvector/eigenvalue data compression was used for successive sorting of data sets for handedness and positions of the gold cluster (12). Other methods. ATPase activity was measured with an ATP regenerating system as described by Lotscher et al. (13). NaDodSO, PAGE was performed according to Laemmli (14) with a 3% stacking gel and a lo-18% separation gel at 4 mA. Instead of DTT, 10 mM DMAB was present in the sample buffer to prevent disulfide bond formation since thiol reagents destroyed the gold cluster rapidly. Immunoblotting was performed as described in Mendel-Hartvig and Capaldi (15).

RESULTS

MMN Labeling of ECF, Our previous studies have established that only two of the 19 Cys residues in the ECF, complex are reactive with

CAPALDI

maleimides (16). One of these is in the 6 subunit (GCys&; the other is an as yet unidentified Cys in one of the three 01subunits (16). Other cysteines in the IXsubunit, one Cys in p, the two Cys in y, along with Cys 64 of the 6 subunit, are buried and not labeled by either [14C]NEM (16) or by the fluorescent reagent CM (17), except in the very few partly denatured protein complexes present in preparations. For our structure-function studies of ECFl, we have constructed a number of mutants in which an extra Cys residue has been incorporated into a subunit by site-directed mutagenesis. The experiments described here use two such mutants, yS8-C and yT106-C, in which either Ser 8 or Thr 106, respectively, of the y subunit was changed to a Cys residue. Both mutants had ATPase activities in the range 17-22 pmol ATP hydrolyzed per minute per milligram protein, values similar to those obtained for wild-type enzyme when assayed under identical conditions (17). To simplify analysis, the 6 subunit was removed prior to reacting the enzyme with MMN so that there should only be two sites of labeling, one in the (Y subunit and the second the introduced Cys residue. We have previously shown that the detergent LDAO releases the 6 subunit from ECFl without effect on activity (16, 18). Removal of the 6 subunit was achieved by gel filtration of the wild-type or mutant enzyme through a Sephacryl S200 column in a buffer containing 0.1% LDAO at 4°C. The LDAO was then removed by a second gel filtration step in buffer without the detergent present. Preparations of ECF, obtained in this way, either from wild-type or mutant enzyme, had ATPase activities in the same range as the untreated enzyme. After reacting ECFl (-6) with MMN, unreacted reagent was removed, the enzyme was reacted with Fab’ fragments derived from a monoclonal antibody to the 01subunit, and then samples frozen rapidly for examination by cryoelectron microscopy. Figure IA shows the labeling of ECF, (-6) from the y mutants by MMN at a molar ratio of 1: 6 enzyme to reagent, as determined by NaDodSO, polyacrylamide gel electrophoresis. Binding of the gold cluster altered the migration of the subunits in the gel, leading to the appearance of two new bands. These two bands could be identified as MMN modified cy and y subunits by silver-enhanced gold visualization (Fig. 1B) combined with Western blotting of companion gels to that in Fig. 1A using mAbs to each of the subunits of ECF, (Fig. 1C). From the gel profile it is clear that under the labeling conditions used, not all of the ECF, molecules have incorporated the gold cluster, even accounting for gold released by the gel electrophoresis conditions (band of gold migrating close to the dye front). Less than complete labeling of reactive Cys residues was evident when the enzyme complexes were examined in cryoelectron microscopy (see below). Several factors affected the efficiency of labeling. MMN was found to inactivate ECF, unless a reducing reagent

y SUBUNIT

1

2

OF THE

123456

FIG. 1. Labeling of ECF, from mutants -5%C and TSlOG-C with MMN. (A) yT106-C reacted with MMN and then subjected to NaDodSO, polyacrylamide gel electrophoresis with protein stained by Coomassie brilliant blue. Lane 1 is unmodified enzyme, lane 2 is enzyme after reaction with MMN. (B) Nitrocellulose blot of MMN reacted ECF, -yTlOG-C. The gold has been visualized by silver enhancement (Nanoprobes Inc.). The arrows show the new bands created by modification of subunits with the reagent. Free gold can be seen migrating near the dye front although the sample was freed of unreacted MMN before gel electrophoresis. Therefore there must be release of gold during the electrophoresis procedure. It appears that this release occurs more readily from the y than from the 01subunit (see text). (C) Western blot of MMN reacted ECF, (-6) from the yS8-C mutant. Lanes 1 and 2, MMN-reacted and unreacted ECFl immunoblotted with anti-y mAb; lanes 3 and 4, MMN-reacted and unreacted ECF, blotted with anti-a mAbs; lanes 5 and 6, same as before but reacted with anti-p mAbs.

was present because of metal ion-induced inter- and intramolecular disulfide bond formation (result not shown). Addition of DMAB prevented this effect but may have reduced the reactivity of the maleimide with the protein. When MMN-treated ECF, was reacted with CM, there was very little incorporation of fluorescence into the cr and y subunit (result not shown), indicating that essentially all of the reactive Cys residues had been modified, even though only a fraction of these sites had incorporated the gold cluster. These results indicate breakdown of the MMN and release of the gold cluster, either during storage, or more likely, during or after reaction of the reagent with the protein. Cryoelectron Microscopy of MMN-Labeled Molecules

ECF,

Figure 2 shows a typical field of images of &depleted ECF, from the mutant yS8-C after labeling with MMN and then reacting with anti-a Fab’s. The binding of the anti-a Fab’s fortuitously orients the enzyme molecule in the ice layer so that the hexagonal view of the enzyme is observed (see also Ref. 5). Superimposition of the density of the anti-a Fab’s on every second mass around the periphery converts the images from a hexagonal to a triangular shape. As evident in Fig. 2, the gold clusters can be clearly seen in the micrograph because of their strong electron density. When wild-type ECFl was used, around 8% of all the images contained the gold cluster. With

E. coli F, ATPase

107

ECFi from the mutants yS8-C and yT106-C, from 25 to 30% of all enzyme particles had gold clusters bound. For analysis, those molecules containing a clearly visible gold cluster, and in one case a control set of unlabeled particles, were digitized and sorted for handedness by correspondence analysis. Images were flipped to create a single handedness, rotationally aligned to superimpose the densities of the three anti-a Fab’s, and then sorted into classes based on their main features. Figure 3A shows the average of unlabeled molecules of ECF, from the $S8-C mutant. This averaged image is similar to ones we have reported earlier for the wild-type enzyme (4,5). The image sorting procedure gave one predominant class of images of gold-labeled particles when wild-type ECF, was labeled with MMN. In t,his class, which included 80% of all of the gold-labeled particles, the gold cluster was located on a peripheral subunit and superimposed on the density of an Fab’, identifying the site of labeling as an cy subunit. As expected, molecules labeled with the gold cluster in the Q subunit were found among images of MMN-labeled ECF, from both mutants yS8-C and yTlO6-C. Figure 3B shows the average of this class of particles of ECF, from the ySS-C mutant. In several experiments, independent of the mutant used, 6-8% of all of the particles (labeled and unlabeled) and 30-35% of gold-labeled particles had the gold bound to an N subunit. Our previous studies have shown that there is one, as yet unidentified, Cys in the CYsubunit which reacts with maleimides but in only one of the three copies of this subunit per ECF, complex (16). The implication that the three (Y subunits can be structurally different in F, ATPases is supported by differences in their reactivity to Lucifer yellow (19) as well as by the different affinities for nucleotides in noncatalytic sites present on this subunit (20). The major class of gold-labeled images in experiments with ECF, from the $S8-C or yT106-C mutants had the gold cluster associated with the central mass. Figure 3C shows the image average of this class from ECF, of the $S8-C mutant. In two different experiments involving this mutant, 14-16% of all images (labeled and unlabeled) and 55-60% of the gold-labeled particles contained a gold cluster associated with the central mass. In wild-type enzyme, particles with the gold cluster bound to the central mass were at most l-2% of all images, and less than 5% of gold-labeled images. The labeling of the central mass by the gold is, therefore, the result of reaction of MMN with the Cys at position 8 of the y subunit. The major class of images obtained in experiments with ECF, from the mutant yT106-C also had the gold cluster bound to the central mass. When the locations of the gold on the y subunit in ECFl from both mutants were compared, they were within 5 A of each other, suggesting that residues S8 and T106 are close in the y subunit. Because of the low reactivity of ECF, with MMN, as described

108

WILKENS

AND

CAPALDI

FIG. 2. Electron micrograph of MMN-modified ECF, (-6) from the yS8-C mutant in a thin layer of amorphous ice. Gold-labeled been decorated with Fab’ fragments derived from a1 monoclonal antibodies.

already, only around O-5% of particles contained gold clusters on both the LYand the y site. Several minor classes of gold-labeled images, each containing 3-5% of the labeled particles, were obtained in all data sets. Some of these are averages of poorly defined (noisy) images; others include images oriented differently because they do not have a full complement of anti-a Fab’s; yet others may be partly denatured complexes with

ATPase has

gold bound to an intrinsic Cys in the /3 subunit or to trace amounts of 6 subunit not removed prior to labeling. DISCUSSION

The results presented here are significant in two respects. They show that MMN can be used to tag specific sites in a protein, in this case Cys residues intrinsic to the ECF, complex (on the cy subunit) and Cys residues

109

y SUBUNIT OF THE E. coli F, ATPase

FIG. 3. Averages of three classes of images from micrographs such as shown in Fig. 2. (A) ECF,-anti-a1 images). (B) ECF,-anti-& Fab complex labeled with gold on an (Y subunit (39 images). (C) ECF,-anti-a1 subunit (99 images). The scale bar represents 20 A.

introduced into the enzyme by site-directed mutagenesis (yS8-C and yT106-C). The strong electron density of the gold clusters makes them easily seen in individual images of the protein-antibody complexes, and these probes should prove particularly useful in identifying the position of the various subunits of ECF, in side views of the intact ECF,Fo complex. Such views can be obtained by examining membrane vesicles containing the entire ATP synthase by cryoelectron microscopy (6). We have generated mutants in which Cys residues are present in the t subunit of ECF, , as well as in the b subunit of FO. Cryoelectron microscopy is now being used to examine the ATP synthase from these mutants after labeling with MMN. The labeling experiments also unambiguously identify the central mass that is observed in hexagonal views of ECF, as the y subunit. Previous evidence that this central mass contains the y subunit was indirect, and was based on the observation that this feature is still present in images of enzyme from which both the 6 and the t subunit had been removed by proteolysis (5). The observation that gold clusters attached to two different sites on the y subunit each superimpose on the central mass is direct evidence that this feature is the y subunit. Both sites labeled with MMN are in the N-terminal part of the polypeptide. MAbs to the C-terminal part of the y subunit were seen to lie near the periphery of ECF, (5). Therefore, the y subunit must extend for at least 40 A across one end of the molecule.

Fab complex without gold bound (111 Fab complex labeled with gold on y

REFERENCES 1. Stoffler, Bioeng.

G., and Stoffler-Meilicke, 13,303-330.

2. Darst, S. A., Kubalek, 340,730-732.

M. (1984) Annu.

E. W., and Kornberg,

3. Amos, L. A., Henderson, R., and Unwin, phys. Mol. Biol. 39, 183-231.

Reu. Biophys.

R. D. (1989) Nature

P. N. T. (1982) Prog. Bio-

4. Gogol, E. P., Lucken, U., Bork, T., and Capaldi, R. A. (1989) Biochemistry 28,4709-4716. 5. Gogol, E. P., Aggeler, R., Sagermann, M., and Capaldi, R. A. (1989) Biochemistry 28, 4717-4724. 6. Lucken,

U., Gogol, E. P., and Capaldi,

R. A. (1990) Biochemistry

29,5339-5343.

7. Gogol, E. P., Johnston, Proc.

Natl.

Acad.

E., Aggeler, R., and Capaldi, R. A. (1990)

Sci. USA

87,

9585-9589.

8. Futai, M., Noumi, T., and Maeda, M. (1989) Annu. 58, 111-136.

Reu. Biochem.

9. Schneider, 497.

Reu. 51, 477-

E., and Altendorf,

10. Senior, A. E. (1988) Physiol.

K. (1987) Microbial. Reu. 68,

177-231.

11. Aggeler, R., Chicas-Cruz, K., Cai, S-Y., Keana, J. F. W., and Capaldi, R. A. (1992) Biochemistry 31, 2956-2961. 12. Frank, J., Shimkin, 343-358.

B., and Dowse, H. (1981) Ultramicroscopy

6,

13. Lotscher, H. R., deJong, C., and Capaldi, R. A. (1984) Biochemistry 23,4134-4140. 14. Laemmli,

U. K. (1970)

15. Mendel-Hartvig, 1278-1284.

Biochemistry

cJ., and Capaldi,

16. Mendel-Hartvig, J., and Capaldi, Acta 1060,115~124.

9, 4620-4626.

R. A, (1991) Biochemistry R. A. (1991)

Biochim.

30,

Biophys.

ACKNOWLEDGMENTS

17. Aggeler, R., and Capaldi, R. A. (1992) J. Biol. Chem. 267,

We are grateful to Dr. Robert Aggeler for providing the mutants and mAbs used in this study. We thank Drs. P. Turina and R. Aggeler for helpful discussions and Erin Johnston for help with image processing. This work was supported by National Institutes of Health Grant HL24536 and a grant from the Lucille P. Markey Charitable Trust.

18. Liitscher, H-R., deJong, C., and Capaldi, R. A. (1984) Biochemistry 23, 4140-4143. 19. Nalin, C. M., Snyder, B., and McCarty, 24, 2318-2324. 20. Cross, R. L. (1988) J. Bioenerg.

Biomembr.

in press.

R. E. (1985) Biochemistry 20,

395-405.