Introduction

Membrane-bound, multisubunit, ion-translocating ATP synthases/ATPases are present in every domain of life and arose from a common ancestor (Gogarten and Taiz 1992). The overall structure of these enzymes is well conserved and consists of two domains connected by (at least) two stalks (Boekema et al. 1997; Böttcher and Gräber 2000; Wilkens and Capaldi 1998). The hydrophilic, cytoplasmic domain catalyzes ATP hydrolysis (Forgac 2000; Futai et al. 1995; Schäfer et al. 1999; Senior et al. 2000), while the membrane-bound domain translocates ions, Na+ or H+, against their electrochemical gradient (Dimroth 1997; Fillingame et al. 2000; Müller et al. 2001). The ATPase in bacteria, mitochondria, and chloroplasts is of the F1FO type; its subunit composition is different in different organisms or organelles but the minimal subunit composition usually found in bacteria is α3β3γδεabc9–12 (Fillingame 2000). A large proportion of the structure of the F1FO ATPase has been elucidated (Abrahams et al. 1994; Menz et al. 2001; Stock et al. 2000). F1FO ATPases are fully reversible machines and their typical cellular function is to synthesize ATP by means of the electrochemical ion gradient across the cytoplasmic membrane. The V1VO ATPase is found in eukaryotes. Its subunit composition also depends on the organism or organelle but the minimal subunit composition as found in yeast is A3B3CDEFGHac 4 c′1 c′′1 d (Forgac 2000). The structure of the V1VO ATPase, and the functions and localizations of their subunits within the enzyme complex are less clear. In vivo, the V1VO ATPase does not catalyze ATP synthesis, its cellular function is to generate steep ion gradients by means of ATP hydrolysis. Another difference F1FO ATPases is the apparent inability of the isolated V1 domain to catalyze ATP hydrolysis (Forgac 1999; Kane 1995).

It has long been known that members of the third domain of life, Archaea, synthesize ATP by means of ion-gradient-driven phosphorylation, and membrane-bound ATP synthases have been demonstrated in membranes of methanogens (Becher and Müller 1994; Mountfort 1978), halobacteria (Michel and Oesterhelt 1980) and thermoacidophiles (Lübben and Schäfer 1989). Enzymes have been purified and characterized (Chien et al. 1993; Hochstein et al. 1987; Ihara et al. 1997; Inatomi 1986; Inatomi et al. 1993; Lübben et al. 1987; Scheel and Schäfer 1990; Steinert and Bickel-Sandkötter 1996; Wilms et al. 1996). Interestingly, the primary sequences of their major subunits A and B were clearly shown to be more related to V1VO than to F1FO (Müller et al. 1999). Furthermore, the genomic sequences available today show that the overall subunit composition of the A1AO ATPase is very similar to the V1VO ATPase. For example, like the V1VO ATPase, the A1AO ATPase contains only two membrane-bound subunits. Like V1VO ATPases, archaeal A1AO ATPases may contain duplicated and even triplicated proteolipids (Müller et al. 1999; Ruppert et al. 1999, 2001). Therefore, the A1AO ATPase can formally be regarded as a chimeric enzyme combining functional features of F1FO such as its cellular function as ATP synthase with structural features of the V1VO ATPase (Mukohata and Ihara 1990; Müller et al. 1999; Schäfer and Meyering-Vos 1992).

Conventional cloning approaches as well as genome sequencing projects have identified at least nine structural A1AO ATPase genes in every archaeon analyzed so far (Fig. 1). The genes atpI and atpK code for subunit I and subunit K (the proteolipid), while the genes atpHECFABD code for hydrophilic proteins. In contrast to the high degree of conservation on the genetic level, the number of polypeptides present in A1AO ATPase preparations reported in the literature was lower than expected from the genetic data and ranged from three to six (Chien et al. 1993; Hochstein et al. 1987; Ihara et al. 1997; Inatomi 1986; Inatomi et al. 1993; Lübben et al. 1987; Scheel and Schäfer 1990; Steinert and Bickel-Sandkötter 1996; Wilms et al. 1996). Only the two major subunits A and B and the proteolipid were assigned, the other polypeptides of these preparations were neither identified nor assigned to specific functions. In addition, subunit I (the homologue of subunit a of V1VO ATPases), which should be essential for ion translocation, was never found in any A1AO ATPase preparation. Therefore, the A1AO ATPase preparations reported to date lack subunits of the hydrophilic and hydrophobic domains and are thus far from being complete. In particular, a complete membrane domain is a prerequisite for identifying the coupling ion in A1AO ATPases, one of the still unresolved questions for methanogenic archaea which generate both primary Na+ and H+ gradients across their cytoplasmic membrane during methane formation (Deppenmeier et al. 1996; Müller et al. 1999). Because it is widely assumed that multienzyme complexes from hyperthermophiles are much more stable than those from mesophiles, we investigated whether the A1AO ATPase from the completely sequenced, hyperthermophilic archaeon Methanococcus jannaschii can be solubilized and purified without loss of subunits. M. jannaschii is of particular interest because it is the only organism known so far to have a proteolipid three times the size of that of most bacteria and archaea and, in addition, this proteolipid has lost one of the proton-translocating residues (Ruppert et al. 1999). To monitor solubilization and stability of the membrane domain, the encoding genes were amplified, fused to malE, the fusions were purified and antibodies were raised against the heterologously produced proteins. We report here a solubilization and purification procedure yielding an apparently homogeneous A1AO ATPase preparation containing nine of the nine subunits deduced from the DNA sequence. The gene–polypeptide correspondence was established by comparing experimentally derived N-terminal sequences of the subunits with sequences deduced from the DNA sequences. Electron microscopy revealed a typical two-domain structure, and inhibitor studies revealed that the A1 and AO domains are functionally coupled.

Fig. 1.
figure 1

Genetic organization of the ATPase genes of Methanococcus jannaschii, Methanosarcina mazei Gö1, Archaeoglobus fulgidus, Pyrococcus horikoshii, and Halobacterium NRC-1. Note that atpD is separated from atpB by 334 kbp in M. jannaschii and 2 kbp in Halobacterium NRC-1. Homologous genes are indicated by identical patterns. It is not clear whether ahaG is an authentic gene/ATPase gene. MJ0614 could be the homologue of ahaG of M. mazei, M. barkeri, and M. acetivorans based on gene arrangements, but the homology is too low for an unequivocal assignment. AhaG homologues were not found in the other archaea

Full size image

Materials and methods

Materials

All chemicals were reagent grade and were purchased from Merck AG (Darmstadt, Germany). N′,N′-dicyclohexylcarbodiimide (DCCD) and Triton X-100 were from Sigma Chemical (Deisenhofen, Germany).

Organism

M. jannaschii (DSM 2661) was obtained from the Deutsche Sammlung für Mikroorganismen und Zellkulturen, Braunschweig, Germany. For purification of the ATPase, M. jannaschii was grown in a 300-l fermentor at 85°C in the medium described at pH 6.0 (Jones et al. 1983) except that 3 g/l NaHCO3, 18 g/l NaCl, 0.5 g/l Na2S but no cysteine-HCl was added. The fermentor was pressurized to 0.3 MPa with H2/CO2 (80:20). The gas flowthrough was adjusted to 1–7 l/min, depending on the growth phase. At an optical density at 600 nm (OD600) of 0.6, the cells were harvested by centrifugation (10,000 g; 20 min; 4°C) in a Sorvall Superspeed RC2-B. The pellets were stored at −80°C.

Purification of the A1AO ATPase

M. jannaschii cells (36–40 g) were lysed by osmotic shock and homogenized in buffer containing 25 mM Tris-HCl (pH 7.0), 5 mM MgCl2, 0.1 mM PMSF (phenylmethylsulfonyl fluoride) and DNase. After cell debris had been removed by centrifugation (11,000 g; 30 min; 4°C), the membranes were pelleted by ultracentrifugation (100,000 g; 2 h; 4°C). The membranes were washed in 100 mM HEPES (pH 7.0), 5 mM MgCl2, 10% glycerol (v/v). The protein concentration was determined as described by Lowry et al. (1951). Membrane proteins were solubilized with Triton X-100 (1 g/g membrane protein) for 30 min at 35°C. After ultracentrifugation (100,000 g; 100 min), the supernatant was applied to a 20%–66% sucrose gradient, and centrifuged for 20 h in a vertical rotor (153,000 g; 4°C). Samples containing the highest ATPase activity were pooled and applied to DEAE-sepharose, equilibrated with 50 mM Tris-HCl (pH 7.5), 5 mM MgCl2, 10% glycerol (v/v), 0.1% Triton X-100. Elution was performed with a salt gradient (0–1 M NaCl in 50 mM Tris-HCl (pH 7.5), 5 mM MgCl2, 10% glycerol (v/v), 0.1% Triton X-100), fractions with ATPase-activity were pooled and applied to a BioPrepSE1000/17 column, equilibrated with 50 mM Tris-HCl (pH 7.0), 5 mM MgCl2, 10% glycerol (v/v), 0.1% Triton X-100. All steps were performed at 4°C.

ATPase activity

ATPase activity was measured in an assay mixture containing 100 mM MES [2-(morpholino)ethanesulfonic acid], 100 mM Tris (pH 6.0), 40 mM NaHSO3, 5 mM MgCl2, and enzyme solution. After preincubation for 3 min at room temperature followed by 3 min at 80°C the reaction was started by the addition of Na2-ATP to a final concentration of 2.5 mM. Activity was measured by the release of inorganic phosphate as described by Heinonen and Lahti (1981). DCCD was dissolved in ethanol and preincubated with the enzyme for at least 20 min at room temperature; controls received the solvent only.

Generation of antibodies and Western blotting

For the expression studies, the transmembrane helices two and three or four and five of atpK, and the hydrophilic part of atpI were amplified by PCR by introducing restriction sites at the 5′end and the 3′end (primer: OatpKTM23.5′(BamHI): 5′-GGTGCAGGATCCACAGGAGCA-3′; OatpKTM23.3′(PstI): 5′-GGAAGGCTGCAGAAAACTATTGC-3′; OatpKTM45.5′ (BamHI): 5′-GGTCAGGGATCCGCTGCTTC-3′; OatpKTM45.3′(PstI): 5′-GCCAAAACTGCAGCCCT ACCC-3′; atpI.5′(NdeI): 5′-GAGACCCGTACATATGAAGTTA-3′; atpIcD.3′(BamHI): 5′-GAGCATGGATCCCTGT GAGCTACTC-3′). PCR fragments were cloned into pMalc2stop, and transformed in E. coli BL21-CodonPlus(DE3)-RIL. Cultures were grown in LB at 37°C, and expression was induced at an OD600 of 0.5 by the addition of IPTG (isopropyl-β-d-thiogalactoside) to a final concentration of 0.3 mM. After 2 h of growth, cells were harvested, washed, and disrupted in a French press. Cell debris was removed by centrifugation and the supernatant was applied to an amylose resin to purify the fusion protein. Since there is no MalE in M. jannaschii and since a MalE antibody does not crossreact with cell-free extract of M. jannaschii, the entire fusion protein was used to immunize rabbits.

Western blotting with SDS polyacrylamide gels was performed as described by Towbin et al. (1979). The nitrocellulose sheets were applied to different antisera and treated with alkaline phosphatase-conjugated goat anti-mouse immunoglobulins in a reaction mixture made up of 0.0075% (w/v) Nitro Blue tetrazolium chloride and 0.03% (w/v) 5-bromo-4-chloro-3-indolyl phosphate in 100 mM Tris, 100 mM NaCl, and 5 mM MgCl2, pH 8.8.

Electron microscopy

Samples containing isolated A1AO ATPase were negatively stained with 4% (w/v) aqueous uranyl acetate, pH 4.7, as described by Valentine and Chignell (1968), and depicted at calibrated magnifications by conventional transmission microscopy.

N-terminal amino acid sequencing

The proteins were separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The proteins were excised and subjected to Edman degradation on a Beckman Proton 3600 sequencer. Identification of the amino acids was performed with Beckman Microbe-HPLC System Gold (Gallagher et al. 1993).

Results

Overproduction of subunits I and K and generation of polyclonal antisera

To monitor the solubilization of the membrane, intrinsic AO subunits K (synonymous with subunit c or proteolipid) and I antibodies were generated. Therefore, the encoding genes were amplified, fused to malE, and the constructs were transformed into E. coli. Upon induction of gene expression by IPTG, growth of the host cells ceased, but neither a MalE-AtpI nor a MalE-AtpK fusion was detectable in Coomassie-stained SDS polyacrylamide gels loaded with whole-cell lysates of the transformants. Therefore, parts of the two genes were amplified and fused to malE. When the gene fragment encoding the hydrophilic domain of subunit I was fused to malE (pHK201), the transformants produced a fusion protein (MalE-AtpICD) of the expected size (data not shown). Subunit K has six predicted transmembrane helices; from the two constructs tested only the transmembrane helices two and three could be produced as MalE fusions (MalE-ATPKTM23; data not shown), but the yield was low. The fusion proteins were purified by affinity chromatography on an amylose matrix and used to immunize rabbits. The specificity of the antisera was tested with cell-free extract, cytoplasmic and membrane fractions of M. jannaschii. As can be seen in Fig. 2A, the antiserum against the cytoplasmic domain of subunit I reacted with a single protein of an apparent molecular mass of 66 kDa, which corresponds well with the deduced size of subunit I of 72 kDa. This protein was exclusively found in the membrane fraction, underlining the prediction that subunit I is part of the membrane domain. In contrast to subunit I from the mesophilic methanogenic archaeon Methanosarcina mazei Gö1, subunit I does not undergo degradation in the time period examined (up to 3 months). The antiserum against subunit K gave one strong signal with a protein of apparent molecular mass 27 kDa (Fig. 2B), which most likely represents a dimer of the proteolipid. Interestingly, the antibody also reacted with a protein of apparent molecular mass 66 kDa, which could represent the K-oligomer (synonymous with c-oligomer); SDS-resistant c-oligomers have been found previously in the Na+-F1FO-ATPase from Acetobacterium woodii (Reidlinger and Müller 1994) and Propionigenium modestum (Laubinger and Dimroth 1988). In addition, the monomer (16 kDa) and degradation products of the proteolipid were seen. The immunoblots verified the membrane localization of subunits K and I, and demonstrate that the antisera are suitable for monitoring the fate of the AO domain during solubilization and purification of the A1AO ATPase.

Fig. 2.
figure 2

Cellular localization of subunit I (A) and subunit K (B). Crude extract (50 μg), the cytoplasm (50 μg) and the cytoplasmic membrane [50 μg (A) or 150 μg (B)] were applied to SDS-PAGE, transferred to a nitrocellulose membrane and then hybridized against an anti-MalE-AtpICD-antiserum (A) or an anti-MalE-AtpKTM23-antiserum (B)

Full size image

Solubilization of the A1AO ATPase from membranes of M. jannaschii

Different detergents were tested under various conditions for their capability to solubilize an ATP-hydrolyzing complex still containing the membrane-bound subunit I. Triton X-100, octylglucosid, or CHAPS (3-3-(cholamidopropyl)dimethylammonio-1-propanesulfonate) were used at a concentration of 1 g/g membrane protein, and incubated with the membranes for 30 min at 35°C. With CHAPS, about 30% of the total activity was solubilized, and 80%–94% of the activity was recovered. With octylglucosid, 32%–46% of the activity was solubilized. Triton X-100 led to a solubilization of 25%–51% of the activity. Higher amounts of detergent, different temperatures or longer solubilization times did not enhance the yield. Subunit I was detected in every solubilisate. For the following experiments we chose Triton X-100 as a detergent for solubilization.

Purification of the A1AO ATPase from membranes of M. jannaschii

The solubilisate was applied to sucrose density gradient fractionation. The highest ATPase activity was recovered in the last third of the gradient. The enrichment factor was 7.6, and the yield was 21.4%. The ATPase-containing fractions were applied to a DEAE-sepharose column, and eluted with a NaCl gradient from 0 to 1 M. The ATPase eluted at 140–260 mM NaCl as a single peak with a specific activity of 3,420 mU/mg and a yield of 13.8%. Because the preparation still contained contaminating proteins, the ATPase-containing fractions were concentrated by ultrafiltration and subjected to a gel filtration on BioPrep SE1000/17. The ATPase eluted as a single sharp peak with a specific activity of 3,450 mU/mg and an overall yield of 4%. The specific activities of various enzyme preparations ranged from 3.5 to 0.8 U/mg protein. The purification is summarized in Table 1.

Table 1. Purification of the A1AO ATPase of M. jannaschii
Full size table

Subunit composition and gene polypeptide correspondence

The ATPase preparation contained 11 polypeptides of apparent molecular masses 76, 70, 66, 54, 45, 28, 27, 25, 16, 13, and 12 kDa. The N-termini of the polypeptides were used to identify the gene products. The 76-, 70-, and 66-kDa polypeptides were only resolved by 10% SDS-PAGE (see Fig. 3). The sequence of the 66-kDa polypeptide matches exactly the sequence deduced from atpI; furthermore, the polypeptide cross reacts with the antiserum against subunit I which is unequivocal evidence that it is subunit I. The 76-, 54-, and 45-kDa proteins were identified as subunits A, B, and C, respectively, by their N-terminal sequences. The 25-kDa protein was identified by the N-terminal sequence as a mixture of subunits D and E. The 16-kDa protein represents the monomeric form of the proteolipid, and the 13- and 12-kDa proteins are subunits F and H, as evident from their N-terminal sequences. These analyses demonstrate that every subunit deduced from the DNA sequence was indeed present in the preparation. However, it should be mentioned that additional polypeptides also appeared in the SDS polyacrylamide gel upon storage of the enzyme. The 70-kDa protein which could only be resolved from subunits A and I in 10% acrylamide reacted with the antibody against the proteolipid, indicating that it represents an SDS-resistant oligomer of the proteolipid; the same is true for the 27-kDa protein (data not shown). SDS-resistant oligomers of proteolipids have been discovered previously in the Na+-F1FO-ATPases from Acetobacterium woodii (Reidlinger and Müller 1994), Ilyobacter tartaricus (Neumann et al. 1998), and Propionigenium modestum (Laubinger and Dimroth 1988). The 28-kDa protein was not identified, but could probably be a degradation/aggregation product of any other subunit. These experiments revealed that the preparation contained all the subunits deduced from the DNA sequence.

Fig. 3.
figure 3

Subunit composition of the A1AO ATPase of M. jannaschii. The purified enzyme (100 μg) was subjected to SDS-PAGE on 10% and 16% gels. Proteins were silver-stained, and the N-terminal sequences were obtained by Edman degradation

Full size image

Catalytic properties of the purified A1AO ATPase

The ATPase hydrolyzed ATP with a constant rate over a period of 12 min (Fig. 4), thereafter the rate declined. Therefore, only initial rates of phosphate release were determined in the following measurements, and the phosphate released from ATP in the absence of enzyme was subtracted from the values used for rate calculations. The ATPase activity was optimal at pH 6.0 (Fig. 5). While pH values lower than 6.0 decreased the activity very strongly, at values higher than 6.0 the decrease was less pronounced (92% activity at pH 7.0). The enzyme was active at temperatures of 60° (21% activity), 70° (28% activity), 90° (89% activity), and 100°C (85% activity); optimal temperature was 80°C. In contrast, when the enzyme was preincubated for 30 min in the absence of ATP at 90° or 100°C, activity was lost completely. Preincubation temperatures ranging from room temperature to 80°C had no effect on ATPase activity (measured at 80°C). Apart from ATP, the enzyme hydrolyzed GTP (86% activity) and even UTP (54% activity) at rather high rates, but CTP (9% activity) at low rates. This result indicates that the ATPase of M. jannaschii is a good GTPase, which was also shown for the Methanococcus voltae ATPase (Chen and Konisky 1993). MnCl2 was superior to MgCl2 (109%), but this is also true for other A1AO ATPases (Inatomi 1986; Inatomi et al. 1993; Steinert et al. 1997). Zn2+ could replace Mg2+ to some extent (64%) but Ca2+, Cu2+, Fe2+, and Ni2+ were less effective. The KM value for Mg-ATP was determined to be 1.2±0.2 mM.

Fig. 4.
figure 4

Kinetics of ATP hydrolysis. After preincubation for 20 min at room temperature and 3 min at 80°C, the reaction was started by addition of Na2-ATP to a final concentration of 2.5 mM. The buffer contained 100 mM MES, 100 mM Tris, 5 mM MgCl2, and 40 mM NaHSO3, pH 6.0, and 16.5 μg of the purified A1AO ATPase of M. jannaschii (diamonds). For the thermal hydrolysis of ATP (circles), no enzyme was added

Full size image
Fig. 5.
figure 5

pH dependence of ATPase activity. After preincubation for 20 min at room temperature and 3 min at 80°C, the reaction was started by addition of Na2-ATP to a final concentration of 2.5 mM. The buffer contained 100 mM MES, 100 mM Tris, 5 mM MgCl2 , and 40 mM NaHSO3, the pH was as indicated in the figure

Full size image

Ultrastructure of the A1AO ATPase from M. jannaschii as revealed by electron microscopy

In electron micrographs the A1AO ATPase appeared to be composed of a base, a stalk and a head part (Fig. 6). Often several A1AO ATPases were aggregated in opposite orientations, brought about by interaction of their AO parts (see large arrow in Fig. 6b). The group of arrows in Fig. 6b points to elongated masses, arranged in a similar way to the slices of an orange. These elongated protein masses are interpreted to represent subunits A and B of the heads of the ATPase complexes seen in side-on views. The arrowhead (Fig. 6b) points to a stalk structure which is interpreted to represent the primary stalk connecting the head with the membrane domain. Paired structures, located between the surface of the cytoplasmic membrane and the enzyme head (Fig. 6c), are structurally distinct from the primary stalk. These are interpreted to represent components of the second stalk, which is part of the stator of the rotatory machine. The A1AO ATPase has an overall length of ~23 nm, with an A1 domain of ~11 nm in width and ~10 nm in length. The length of the AO part is about the width of the A1 particle and the height is ~6.0 nm. The electron micrographs clearly demonstrate the two-domain structure of the enzyme isolated.

Fig. 6.
figure 6

Electron micrograph of the A1AO ATPase of M. jannaschii. Dimensions are given in nm. For explanations, see text

Full size image

Inhibition of ATP hydrolysis by DCCD

To test for a functional coupling between the A1 and AO domains, DCCD, an inhibitor of the A1AO ATPase from methanogens (Becher and Müller 1994), known to block ion flow through the membrane domain, was used. As can be seen from Fig. 7, ATPase activity was inhibited by DCCD. A 50% inhibition was obtained at about 450 μM DCCD. These data demonstrate that most of the A1 and AO domains in this preparation are functionally coupled.

Fig. 7.
figure 7

Inhibition of the A1AO ATPase by DCCD. After preincubation with the inhibitor for 20 min at room temperature and 3 min at 80°C, the reaction was started by the addition of Na2-ATP to a final concentration of 2.5 mM. DCCD was added as ethanolic solution, controls received the solvent only. The protein concentration was 0.3 mg/ml

Full size image

Discussion

To open new avenues to the functional and structural analyses of archaeal A1AO ATPases, we chose methanogenic archaea as model systems and started two approaches, a molecular and a biochemical. The A1 ATPase genes from the mesophile M. mazei Gö1 could be overexpressed in E. coli, and a functional A1 ATPase subcomplex suitable for biochemical and structural analyses was produced (Coskun et al. 2002; Grüber et al. 2001a, 2001b; Lemker et al. 2001). However, up until now, neither the AO nor the A1AO could be overproduced heterologously. As shown here, single subunits of AO could not even be produced as MalE fusions; their overproduction led to growth inhibition of the host. It has been shown previously for the proteolipid from M. mazei Gö1 that it can be produced in E. coli and is targeted to the membrane, but overproduction of the proteolipid apparently has dramatic, growth-inhibiting consequences for the energy status of the host (Ruppert et al. 1998). The same is apparently true for subunit I, which has seven predicted transmembrane spans and which is, most likely, involved in ion transport. Overproduction of the homologue of subunit I of F1FO ATPases, subunit a, also proved to be difficult (Aris et al. 1985; Vik and Antonio 1994).

Attempts to solubilize a complete A1AO ATPase from the mesophilic M. mazei Gö1 or other archaea have failed so far. Because it is widely assumed that multienzyme complexes from hyperthermophiles are much more stable than those from mesophiles, we investigated whether a complete A1AO ATPase could be solubilized from a hyperthermophile, M. jannaschii. As demonstrated here, subunit interactions in the A1AO ATPase from the hyperthermophile are apparently stronger and can withstand detergent treatment. However, this cannot be generalized, since attempts to solubilize complete A1AO ATPases from other thermophiles such as Sulfolobus acidocaldarius (Lübben et al. 1987; Lübben and Schäfer 1989) or Methanothrix thermophila (Inatomi et al. 1993) have failed.

It is apparent from the electron micrographs that the A1AO ATPase consists, like the F1FO and the V1VO ATPase, of a hydrophilic (A1) and a hydrophobic domain (AO) which are connected by two stalks. This two-domain structure has been observed before with membrane-bound enzymes or with the six-subunit-enzyme isolated from M. mazei Gö1 (Wilms et al. 1996). Here, with the nine-subunit-enzyme, similar overall dimensions of the enzyme complex were determined, but in addition to the previous six-subunit preparation additional masses can be seen, which are interpreted to represent the second stalk.

The ATPase operon of M. mazei Gö1 contains an open reading frame at its 3′ end, named ahaG. It is not clear whether this is indeed an authentic gene; a homologue is found in Methanosarcina barkeri (http://genome.ornl.gov/microbial/mbar/) and Methanosarcina acetivorans (Galagan et al. 2002), but not in the genome of M. jannaschii. Since a subunit G was not found in the purified enzyme, its presence seems unlikely, but cannot be excluded with certainty.

From the N-terminal sequences determined, it is now possible to delineate the gene–polypeptide correspondence (Table 2). All except subunit I have an ATG start codon, GTG is used in atpI. The N-terminal methionine was removed from subunits A, B, C, E, H, and K. The annotated start codons of atpA, atpE, and atpI differ from the start codons deduced from the N-terminal sequences. atpA starts 21 base pairs downstream of the annotated start codon, the start codon of atpE is annotated as GTG but translation apparently starts six nucleotides downstream, and atpI starts with a GTG two triplets downstream from the annotated GTG.

Table 2. Gene–polypeptide correspondence of the A1AO ATPase of M. jannaschii
Full size table

In summary, the purification procedure described here enables the isolation of a structurally complete A1AO ATPase comprising nine subunits and will allow a thorough analysis of the structure and function of this unique class of enzymes from archaea.