The Subunit b of the F0F1-type ATPase of the Bacterium Mycoplasma pneumoniae Is a Lipoprotein*

The DNA sequence analysis of the F0F1-ATPase operon of the bacteriumMycoplasma pneumoniae predicted that the subunit b, encoded by the gene atpF, is a lipoprotein of the murein lipoprotein type of Escherichia coli. Here we experimentally verify this prediction by metabolic labeling of subunit b with [14C]palmitic acid and by in vivointerfering with the processing of the prolipoprotein form of subunit b by the antibiotic globomycin, a specific inhibitor of the signal peptidase II. Our results suggest that the subunit b of the F0F1-ATPase of M. pneumoniae is anchored at the cytoplasmic membrane by an N-terminal lipid modification in addition to its transmembrane domain. The lipoprotein nature of subunit b and its proposed membrane topology seems to be characteristic for mycoplasmas, since among all sequenced bacterialatpF genes, only those from Mycoplasma gallisepticum and Mycoplasma genitalium code for a conserved lipoprotein consensus sequence.

The DNA sequence analysis of the F 0 F 1 -ATPase operon of the bacterium Mycoplasma pneumoniae predicted that the subunit b, encoded by the gene atpF, is a lipoprotein of the murein lipoprotein type of Escherichia coli. Here we experimentally verify this prediction by metabolic labeling of subunit b with [ 14 C]palmitic acid and by in vivo interfering with the processing of the prolipoprotein form of subunit b by the antibiotic globomycin, a specific inhibitor of the signal peptidase II. Our results suggest that the subunit b of the F 0 F 1 -ATPase of M. pneumoniae is anchored at the cytoplasmic membrane by an N-terminal lipid modification in addition to its transmembrane domain. The lipoprotein nature of subunit b and its proposed membrane topology seems to be characteristic for mycoplasmas, since among all sequenced bacterial atpF genes, only those from Mycoplasma gallisepticum and Mycoplasma genitalium code for a conserved lipoprotein consensus sequence.
The bacterial membrane-bound F 0 F 1 -type ATPase serves two purposes. The enzyme complex catalyzes the synthesis of ATP in response to an electrochemical proton gradient and generates a transmembrane proton gradient by hydrolyzing ATP (1). Mycoplasmas differ from the majority of the bacteria by the lack of a cytochrome-containing electron transport chain; therefore their F 0 F 1 -ATPase function seems to be restricted to maintaining a proton gradient (2). DNA sequence analyses of the complete F 0 F 1 -ATPase operons of the three mycoplasma species Mycoplasma gallisepticum (3), Mycoplasma genitalium (4), and Mycoplasma pneumoniae (5) indicated the presence of the same eight homologous subunits as in the corresponding operons of Escherichia coli (6) and Bacillus subtilis (7). Therefore, by analogy, we assume that in mycoplasmas the F 1 complex is formed by the five subunits ␣, ␤, ␥, ␦, and ⑀ and the F 0 complex by the subunits a, b, and c. Thus, the F 0 complex would be inserted in the cytoplasmic membrane of mycoplasmas and interact with the F 1 complex, which would be oriented toward the cytosol. Comparative sequence analyses show that of the orthologs in the three mycoplasma species, E. coli and B. subtilis, the genes atpA (subunit ␣) and atpD (subunit ␤) share the highest similarities, about 50 -70% identity at the amino acid level, whereas the other genes are less well conserved and differ in length (5). The most prominent example for a structural difference is the subunit b encoded by the gene atpF. The DNA sequence-based analysis predicts that the subunit b of the three different mycoplasma species has the specific features of a lipoprotein of the murein lipoprotein type of E. coli (5). These include a signal peptide with positively charged amino acids close to the N-terminal end and an accumulation of hydrophobic residues within the signal peptide followed by a cysteine at position 20 -35 of the putative prolipoprotein. This cysteine is part of the conserved sequence of the prolipoprotein modification/processing site and will become the N-terminal amino acid in the mature protein after the signal peptide has been cleaved off by signal peptidase II (8). The processed protein is associated with the membrane by the attachment of a diacyl-glycerol moiety to the SH 2 -group of the cysteine prior to cleavage (9,10). Searching the data bases for subunits b with a lipoprotein signature revealed that this motif was absent not only in E. coli and B. subtilis but in all other available bacterial atpF sequences. This suggests that the lipoprotein character of the subunit b is a specific trait of mycoplasmas. Their most differentiating characteristic compared with other bacteria is the complete lack of a cell wall (11). Being surrounded only by a cytoplasmic membrane might therefore require additional means to anchor certain proteins to the membrane. A relatively high number of lipoproteins have been identified experimentally in several mycoplasmas (12), and the large number predicted from analyses of the DNA sequences of complete mycoplasma genomes support this hypothesis (4,13).
As the DNA sequence-based prediction is not full proof, we decided to examine whether the subunit b of the F 0 F 1 -ATPase of M. pneumoniae is indeed a lipoprotein. The following experimental approaches were taken to prove the lipoprotein character (8) of the F 0 F 1 -ATPase subunit b of M. pneumoniae: (i) metabolic labeling of the proposed lipoprotein with [ 14 C] palmitic acid and identification of the 14 C-labeled subunit b by a specific antibody; (ii) inhibition of the signal peptidase II by globomycin (18) and identification of the uncleaved prolipoprotein.

EXPERIMENTAL PROCEDURES
Organisms and Growth Conditions-M. pneumoniae M129 (broth passage 21) (ATCC 29342) cultures were grown at 37°C in 50 ml Falcon tissue flasks containing modified Hayflick medium (14) supplemented with 20% horse serum (Boehringer Mannheim). After 48 h, attached cells were washed twice with phosphate-buffered saline (0.15 M NaCl, 10 mM sodium phosphate, pH 7.4), scraped off, collected by centrifugation at 6000 ϫ g for 10 min at 4°C, and stored at Ϫ70°C. E. coli strain JM101 was used for propagation of plasmids, and E. coli M15 transformed with the plasmid pSuPMP was used for expression of fusion proteins in E. coli (15).
Preparation of Antiserum against Subunit b-The region of the atpF gene coding for a peptide extending from amino acid residue 123 to 207 ( Fig. 1B) was amplified by polymerase chain reaction and ligated inframe to the mouse dihydrofolate reductase gene in the expression vector pQE40 (Qiagen). This vector allows the regulated expression in E. coli of a fusion protein with six N-terminal histidine residues. The fusion protein was purified by immobilized metal chelate affinity chroma-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. tography and used for immunization of rabbits by standard immunization procedures. The methods have been described in detail previously (15).
Triton X-114 Extraction-M. pneumoniae cells were suspended to a concentration of 2 mg of protein/ml in 10 mM Tris (pH 7.5), 150 mM NaCl, and 10 units ml Ϫ1 aprotinin (Sigma). The Triton X-114 extraction was carried out as originally described by Bordier (16) with some modifications (15,17). The fractions were analyzed by SDS-PAGE 1 and either stained with Coomassie Brilliant Blue or used for Western blotting (14).
Metabolic Labeling of M. pneumoniae-M. pneumoniae cells were grown adhering to plastic in a modified Hayflick medium containing only 10% horse serum till the color of the medium changed from red to yellow. The attached cells were scraped off, suspended in the same volume of growth medium, and dispersed by passing the suspension several times through a needle (0.9 ϫ 70 mm). Fresh medium (7 ml) containing 2 Ci/ml [U-14 C] palmitic acid (Amersham Pharmacia Biotech) was inoculated with 0.7 ml of the M. pneumoniae suspension, and the culture was incubated for 4 -5 days at 37°C. The adherent cells were washed twice with phosphate-buffered saline, scrapped off, and used either for immunoprecipitation or SDS-PAGE. Cells were stored either as pellets or in suspension in phosphate-buffered saline at Ϫ20°C. Methods used for SDS-PAGE, Western blotting and immunoprecipitations, have been described in detail previously (14). All antisera used in these studies were produced in rabbits.
Globomycin Treatment-M. pneumoniae cells were grown in the same medium as described above, but instead of using culture flasks, 2 ml of a 1:10-diluted culture from the log phase were incubated in an Eppendorf tube at 37°C. After 16 h of growth, globomycin (Sankyo Co.) in a final concentration of 50 g/ml was added, and the cells were incubated for 12 h. Cells were collected by centrifugation and either used immediately for SDS-PAGE and Western blotting or stored at Ϫ20°C.

RESULTS AND DISCUSSION
The sequence alignments of three mycoplasma species and E. coli as a representative of a Gram-negative and B. subtilis of a Gram-positive bacterium reveal that only the subunit b proteins of M. gallisepticum, M. genitalium, and M. pneumoniae are N-terminally extended by a signal peptide with a cleavage site analogous to the consensus sequence for prolipoprotein modification and processing (Fig. 1A). In addition hydrophobicity plots predict two closely spaced transmembrane segments for the mycoplasmal subunits as compared with only one in B. subtilis and other bacteria (Fig. 1B). In accordance, the uncleaved subunit b of M. pneumoniae is 37 amino acids, but the processed form is only 9 amino acids longer than the ortholog in B. subtilis.
Mycoplasmas are not able to synthesize long-chain fatty acids by themselves (2,12,19). When grown under laboratory conditions they depend on a supplement of 10 -20% serum, which contains also the fatty acids required for acylation of proteins. Since palmitic acid has been frequently found in mycoplasma lipoproteins and lipids (20), we labeled M. pneumoniae by adding [ 14 C]palmitic acid to the culture medium of growing bacteria. Protein extracts from the labeled cells were analyzed by SDS-PAGE and autoradiography. At least 25 labeled proteins were detected under these experimental conditions (data not shown). The deduced amino acid sequence of subunit b predicts a molecular weight of 24,020 for the subunit b prolipoprotein and 21,000 for the assumed processed form. Only a weak band could be seen on the gel within this size range by autoradiography (Fig. 2B, lane 1). This indicates either that subunit b is synthesized in low amounts or that [ 14 C]palmitic acid is not efficiently incorporated. To identify subunit b unambiguously, the complete 14 C-labeled protein extract was immunoprecipitated with subunit b specific antibodies (see "Experimental Procedure"), and the precipitate was analyzed by SDS-PAGE and Western blotting. A protein with an apparent molecular weight of 21,000 could be detected after immunostaining with the anti-subunit b antibody in the protein extract (Fig. 2A, lane 1) and in the immunoprecipitated sample ( Fig. 2A, lane 2). By subsequent autoradiography of the membrane, a signal could be detected at exact the same position (Fig. 2B, lane 2), providing convincing evidence that the subunit b is acylated with palmitic acid.
The biosynthetic pathway of the murein lipoprotein of E. coli, also known as Braun's lipoprotein, is considered to be repre- 1 The abbreviation used is: PAGE, polyacrylamide gel electrophoresis. sentative for the bacterial lipoproteins (8,21). The penultimate step is the cleavage of the signal peptide from the diacylglycerol-prolipoprotein by the signal peptidase II. This enzyme can be specifically inhibited by the antibiotic globomycin (18). Preliminary growth inhibition tests showed that globomycin at a concentration of 150 g/ml greatly inhibits growth of M. pneumoniae. A concentration of 50 g/ml was used as it permitted growth of the bacteria but inhibited the signal peptidase II. After growth of M. pneumoniae in the presence of globomycin for 16 h, an additional band with an apparent molecular weight of 23,000 could be detected by SDS-PAGE and immunoblotting with anti-subunit b antibodies (Fig. 3). Within the limits of SDS-PAGE for molecular weight estimation, this increase in molecular weight could be attributed to the additional 27 amino acids of the prolipoprotein, which was not further processed to the mature lipoprotein due to inhibition of the signal peptidase II by globomycin.
Inhibition of signal peptidase II action by globomycin and the specific labeling of subunit b with [ 14 C]palmitic acid proved that the subunit b of the F 0 F 1 -ATPase of M. pneumoniae is a lipoprotein. Based on the presence of the lipoprotein motif and the almost identical hydrophobicity plots, which indicate two transmembrane segments near the N terminus, we concluded that the subunit b of the F 0 F 1 -ATPase of M. genitalium and M. gallisepticum is also a lipoprotein.
The structural predictions for subunit b of M. pneumoniae and its proposed function as a protein that interconnects F 0 and F 1 in the enzyme complex strongly suggest it to be an integral membrane protein oriented toward the cytosol of the cell. According to the present model, the N-terminal part of subunit b interacts with subunit a in the membrane, and its C-terminal region interacts with subunit ␦, which is in contact with the F 1 headpiece consisting of the subunits ␣ and ␤ (22). Here we show by partition of proteins during phase separation in solutions of the nonionic detergent Triton X-114 that subunit b is recovered exclusively in the detergent phase (Fig. 4D). This phase is supposed to contain only integral proteins (16). The cytosolic protein elongation factor G (G07_orf688) was partitioned in the aqueous phase (Fig. 4B), and the Triton X-114insoluble protein P65 (F10_orf405) (Fig. 4A) was recovered in the Triton X-114-insoluble fraction (Fig. 4); both of these proteins lack a transmembrane segment. Finally, FtsH (K05_orf 705), with two predicted transmembrane segments partitioned in the detergent phase (Fig. 4C), like subunit b, supporting the assumption that subunit b is an integral membrane protein (13,23). Treatment of intact M. pneumoniae cells with trypsin and proteinase K also showed that subunit b was not accessible to these proteases whereas the surface-exposed protein P1 (24) was cleaved, and the cytosolic protein elongation factor G was not affected under the same experimental conditions (data not shown). From these data we conclude that the C-terminal part of subunit b is oriented toward the cytosol.
Based on the structural prediction for the subunit b and the experimental data provided in this paper, we propose a model for the processing and arrangement of subunit b in the cell membrane of M. pneumoniae (Fig. 5), which should also be valid for M. genitalium and M. gallisepticum. This model improves a previous proposal (3) for the membrane topology of the subunit b of M. gallisepticum, which had not taken into account the lipoprotein character of the protein.
What could be the possible advantage for the unusual lipoprotein structure of subunit b in M. pneumoniae? The most obvious effect is the stronger anchoring of this protein in the membrane through the additional acyl chains. This might be important, since mycoplasmas do not have the rigid murein layer forming a network around the cytoplasmic membrane, which protects bacterial cells against osmotic and mechanical stress (11). Among the 46 predicted lipoproteins in M. pneumoniae (13), we found 9 lipoproteins that are predicted to have at least one more transmembrane segment in addition to the signal peptide. However, in all these instances, the distance between the lipoprotein processing site and the additional transmembrane segment is greater than in subunit b, allowing formation of larger surface-exposed protein loops.
The number of acyl chains attached to the N-terminal cysteine of the subunit b remains unclear. The maximal number observed in the murein lipoprotein of E. coli is three, two by an O-ester formation with the glycerol moiety and one by an amide linkage to the free NH 2 moiety. The amide linkage is catalyzed by a N-acyltransferase (25), but the gene for this enzyme has not been found in the completed genome sequences either of M. genitalium (4) or M. pneumoniae (13,26). Thus it is uncertain whether an N-acyltransferase and its end product, N-acyl cysteine, are present in these species. The few experimental data from other mycoplasmas on this topic indicate that in one species an N-acyltransferase is present but absent in another.
A lipopeptide from Mycoplasma fermentans was found to have a free N terminus (27), supporting the lack of an N-acyl transferase in this bacterium, whereas evidence was provided that lipoproteins from M. gallisepticum have indeed O-ester-bound and amide-linked acyl chains (28). However, the degree of acylation of the lipoproteins remains uncertain in M. gallisepticum as in other mycoplasmas.
Another unresolved question is the origin of the lipoprotein motif in subunits b of mycoplasmas. According to a widely accepted theory, these bacteria originated from Gram-positive bacteria by genome reduction (29). Since Gram-positive and all other bacteria so far analyzed do not carry the lipoprotein motif in the subunit b of the F 0 F 1 -ATPase, one can assume that mycoplasmas adopted this motif during the process of genome reduction either from another lipoprotein gene of the same cell or they received it by horizontal gene transfer. Ultimately, the importance of the lipid modification for a functional subunit b of the F 0 F 1 -ATPase can only decided experimentally by transforming M. pneumoniae with an atpF gene coding for a subunit b devoid of the lipid anchor and analyzing the biological activity of the resulting F 0 F 1 -ATPase. The signal peptide will be cleaved off by signal peptidase II after the cysteine C in position 28 has been modified by the attachment of a diacyl-glycerol moiety. The second transmembrane segment ensures that the C-terminal part of subunit b is oriented toward the cytosol showing the same orientation as in E. coli (1,22).