Subunits of the H’-ATPase of Escherichia coli OVERPRODUCTION OF AN EIGHT-SUBUNIT F,Fo-ATPase FOLLOWING INDUCTION OF A A-TRANSDUCING PHAGE CARRYING THE unc OPERON*

The proton-translocating ATPase complex (F1Fo) of Escherichia coli was purified after induction of a X- transducing phage (kasn5) carrying the ATPase genes of the urn operon. ATPase activity of membranes pre- pared from the induced X-unc lysogen was 6-fold greater than the activity of membranes prepared from strains lacking the unc-transducing phage, confirming the report of Kanazawa et al. ((1979) Proc. Natl. Acad Sci. U. S. A. 76, 1126-1130). The FIFo-ATPase complex was purified in comparable yield from either enriched membranes or control membranes using a modification of the procedure reported by Foster and Fillingame ((1979) J. Biol. Chem 254,8230-8236). Each of the eight subunits that had been reported as components of the FIFo complex from wild type E. coli was overproduced in the A-unc lysogen. All eight subunits co-purified in the same stoichiometric proportion as in the complex purified from wild type E. coli. We conclude that all eight subunits are likely coded by the small segment of chromosomal DNA carried by the X-transducing phage. These experiments provide the first evidence that all eight polypeptides are authentic subunits of the ATP- ase complex rather than contaminants that fortuitously co-purify.

location of H' across the membrane to the synthesis or hydrolysis of ATP (2, 4). The F,-ATPase has been purified from several species of bacteria, mitochondria, and chloroplasts (1, 3). It is an extremely complex enzyme, composed of at least five nonidentical subunits in most species and perhaps six subunits in mammalian mitochondria (1,3). Fragmentary information on the function of different subunits of F1 has been obtained by biochemical reconstitution experiments with purified subunits from two species of bacteria (2, 5). The Fo sector of the complex has been less thoroughly analyzed than F, and the subunits only tentatively identified. The F,Fo-ATPase complexes purified from the thermophilic bacterium PS3 and Escherichia coli contained three polypeptides in addition to the five which compose F, (6,7). On the other hand, the FIFO complex purified from chloroplasts contained four polypeptides in addition to those of F1 (8). Mitochondrial FIFO preparations are considerably more complicated (9-12). A common subunit found in the Fo of all species studied to date is a hydrophobic "proteolipid" protein which is the site of covalent reaction with dicyclohexylcarbodiiiide (DCCD), an inhibitor of the proton-translocating activity of Fa (4). A second subunit from the FO of the thermophilic bacterium PS3 has been implicated in the binding of the F1-ATPase (13). It remains uncertain whether there are other subunits that are authentic components of Fa and what function these subunits may perform.
The FIFO-ATPase of E. coli has been studied extensively in recent years because questions of function are subject to genetic as well as biochemical analysis in this organism (14,15). All mutations affecting the ATPase complex have mapped a t a single locus termed unc, and these genes seem to be organized in an operon (15). Biochemical analysis of mutants altered in different subunits should provide definitive information on the function of each subunit. For example, it has been shown by this approach that both the a and p subunit of F, play some role in catalytic activity, since genetic alteration of either of these subunits abolishes ATPase activity (5, 15, [16][17][18][19][20]. Similarly, it was through the use of DCCD-resistant mutants that the proteolipid protein of Fa was most clearly shown to be the site of specific DCCD reactivity, the reaction leading to inhibition of both proton translocation and ATPase activity (4, 21, 22). Genetic techniques also provide a means of amplifying the genes coding for the complex. Overproduction of the complex would facilitate its preparation in large quantity. This approach was used by Young et al. (23) to obtain a substantial increase in the level of membrane-associated NADH dehydrogenase.
Miki et ad. (24) have described the isolation of a specialized transducing phage, Aasn5, which carries chromosomal DNA including the unc operon. Induction of this phage was shown by Kanazawa et al. (25) to result in increased levels of mem-

Overproduction of Eight-subunit H'-ATPase
brane-associated ATPase activity and evidence was presented indicating that the five subunits of F, were overproduced. Since the ATPase activity was membrane-bound and DCCDsensitive, it seems likely that at least some components of Fo were overproduced as well as F,. We have extended this work and demonstrate here that the eight subunits found in the purified FIFo-ATPase of E. coli are overproduced during induction of AasnS.

EXPERIMENTAL PROCEDURES
Bacterial and Viral Strains-The following derivatives of strain KH716 [asn-31, thi, and rifl (24) were used. Strain "95 was derived by lysogenizing strain KH716 with with XcI857S7. This phage is thermoinducible due to the cI857 mutation and is unable to lyse cells due to the S7 mutation (24). Strain KY7485 (24) was derived from strain KH716 by lysogeny with XcI857S7 and the transducing phage hasn5 (XcI857S7[bgIR-C+, glmS', uncA', asn']), which carries a segment of DNA including the unc operon. In the text, strain KY7485 will be referred to as the X-unc lysogen and strain "95 as the Xlysogen control. Strains AN180 and ML308-225 are nonlysogenic uric+ strains which were utilized as sources of F,Fo and FI, respectively, as described previously (7). Growth of Cells and Induction of X Phages-Cells were grown on minimal medium containing 0.1 M potassium phosphate (pH 7.5), 93 mM NH,Cl, 0.8 mM Na2S04, 16 mM MgC12, and 3.6 pM FeS04 supplemented with 78 mM glucose, 14.8 WM thiamin, and 3.8 mM L-asparagine. Asparagine was omitted for growth of strain KY7485. T o avoid formation of a precipitate, the concentration of minerals in the medium was initially one-half that stated; the remainder was added 2 h before phage induction. Cells were grown in 10 liters of medium in a 14-liter New Brunswick fermenter, with aeration at 8.5 liters/min and stirring at 400 rpm. Under these conditions, saturation was reached at an A of 9 units a t 550 nm (2.4 X 10"' cells/ml). Cells were grown a t 32°C to an optical density of 3 units (8 X 10' cells/ml). X phage production was then induced by raising the temperature of the medium to 42°C over a period of 17 min. After 30 min, the temperature was reduced to 37°C (over a period of 3 min) and aeration and stirring were continued for 3 h. Membrane-associated ATPase activity was maximal a t this time. Cells were harvested, washed, and stored as described previously (7).
Preparation of the ATPase Complex-Membranes were prepared as described (7), except that 6 mMp-aminobenzamidine was included in all buffers. In order to purify the ATPase complex in high yield from membranes of the induced X-unc lysogen, the procedure of Foster and Fillingame (7) had to be modified. The modified procedure also proved superior in purifying the ATPase complex from nonlysogenic strains grown on glucose. Membrane a t 20 mg of protein/ml in Buffer A (50 mM Tris-HC1 (pH 7.5). 5 mM MgS04, 1 mM dithiothreitol, 10% (v/v) glycerol, 6 mM p-aminobenzamidine, and 1 mM phenylmethylsulfonyl fluoride) was adjusted to 1 M KC1 by the addition of solid KC1 and to 0.5% in both sodium deoxycholate and sodium cholate by the addition of 10% (w/v) detergent solutions. The suspension was intermittently mixed for 10 min a t 0°C and then centrifuged at 40,000 rpm (193,000 X gmaJ for 80 min in a Beckman type 50.2 Ti rotor a t 4%. The clear supernatant solution was immediately diluted by the addition of 1 volume of Buffer A and dialyzed against 50 volumes of Buffer A for 16 h a t 4'C with one change of external buffer at 6 h. The resultant turbid suspension was centrifuged as above. The pellet was homogenized in a volume of 100 mM Tris-HC1 (pH 7.5), 5 mM MgSO,, 2 mM dithiothreitol, 10% (v/v) methanol, and 12 mM paminobenzamidine equal to one-fifth that used in the extraction of the membrane. This suspension was diluted to 11 mg of protein/ml, adjusted to 0.6% in sodium deoxycholate, and fractionated with ammonium sulfate as described (7), but taking a 25 to 35% cut. The use of methanol rather than glycerol in this step eliminated sporadic problems with floating precipitates and improved the yield. The ammonium sulfate fractionated material was purified by sucrose density gradient centrifugation as described (7) except 6 mM p-aminobenzamidine and 0.28% deoxycholate (w/v) were included in the gradients.
When the complex was prepared from the A-lysogens, the ammonium su1fat.e fractionation described above was omitted. The particulate fraction of the dialyzed membrane extract was resuspended at 20 mg of protein/ml in 50 mM Tris-HCI (pH 7.5), 5 mM MgC12, 1 mM dithiothreitol, 10% (v/v) methanol, and 6 mM p-aminobenzamidine and resolubilized by the addition of 10% (w/v) sodium deoxycholate to 1.25% (w/v), 10% (w/v) sodium cholate to 0.5% (w/v), and solid KC1 to 0.25 M. After intermittent stirring for 5 min a t O"C, the suspension was centrifuged for 20 min a t 50,000 rpm (227,000 X gmnr) in a Beckman type 50 Ti rotor a t 4°C to remove residual particulate material. The clear supernatant solution was immediately layered onto 10 to 40% (w/v) sucrose gradients and centrifuged as described above for the purification of the complex from nonlysogenic cells. Following centrifugation, the gradients were fractionated into 26 fractions of 0.19 ml following puncture of the tube bottom. The leading portion of the peak of ATPase activity was pooled and reconstituted as described (7).
Analytical Procedures and Assays-The procedures described (7) were used without modification. The polyacrylamide slab gels used contained 13% (w/v) acrylamide, 0.41% (w/v) N,N'-methylenebisacrylamide, and 0.2% (w/v) sodium dodecyl sulfate and had dimensions of 1.5 mm X 9 cm X 14 cm. Electrophoresis was carried out with the buffer described previously (21) for 6 h a t 20 mA/slab gel.
Chemicals-Sodium deoxycholate was obtained from Calbiochem-Behring (La Jolla, CA). Phenylmethylsulfonyl fluoride was obtained from Sigma (St. Louis, MO). Cholic acid (Sigma) was recrystallized twice from 70% ethanol and neutralized with NaOH. p-Aminobenzamidine dihydrochloride was obtained from Aldrich (Milwaukee, WI). " Approximately 20% of the total ATPase activity was inactivated during the extraction step, i.e. it could not be accounted for when the ' The yield at this point was more typically 60%.
' Material applied to sucrose gradient.
, I Includes resolubilization of particulate dialyzed extract, ammonium sulfate fractionation, and resolubilization for application to Sucrose e Material is not pure. gradients.

RESULTS
Gene Dosage-dependent Increase in Membrane ATPase-Kanazawa et al. (25) indicated that lysogeny with the X-unctransducing phage resulted in an increase in membrane ATPase activity consistent with that expected due to gene dosage. Under the conditions used here, lysogeny with A-unc resulted in a 2-fold increase in ATPase activity relative to either a nonlysogenic or nontransducing X-lysogen control ( Table I).
Thermal induction of the X-unc lysogen resulted in a further 3-fold increase in ATPase activity, whereas no change in membrane ATPase was observed on induction of the nontransducing X-lysogen control (Table I). These results are consistent with those reported previously (25), but apply to cells grown on glucose minimal medium and on a large scale ( Table 11). The level of membrane ATPase activity seems to correlate well with the number of copies of the unc operon present per cell.
Purification of FIFO from Induced X-unc Membranes-The F,Fo-ATPase was purified from the induced X-unc membranes by a modification of the procedure of Foster and Fillingame (7). In order to minimize any effects arising from phage induction alone, the complex was also purified by this procedure from strain "95, which contained a heat-inducible X prophage identical with the helper phage in strain KY7485. The complex was purified several times from both types of cells and the results of a typical purification are summarized in Table 11. In order to efficiently solubilize the ATPase complex from induced X-unc membranes, the detergent concentration used for extraction was increased relative to that described (7). Similarly, the detergent concentration had to be increased to resolubilize the particulate ATPase formed after dialysis in reasonable yield. Methanol rather than glycerol was used to stabilize the ATPase during resolubilization since it diminished formation of a precipitate that occasionally floated during centrifugation. The ammonium sulfate fractionation procedure was not used in the modified procedure since it proved possible to obtain high purity F,Fo from membranes of induced X-unc without it, and the yield was poor due to problems in resolubilization. Inclusion of the ammonium sulfate fractionation step was necessary in order to obtain high purity FIFo from membranes of the induced, nontransducing X control. This is indicated by the lower specific activity in Table I1 and by analysis on acrylamide gels as discussed below.
Subunits Overproduced in Induced X-unc-The subunit compositions of the purified ATPase preparations described in Table 11 were compared by SDS-polyacrylamide gel electrophoresis (Fig. 1). The FIFO complex prepared from induced X-unc by the abbreviated procedure contained eight subunits which migrated identically with those found in the complex purified from a nonlysogenic strain by the original procedure of Foster and Fillingame (7). Furthermore, the relative proportions of each subunit, as judged by the staining intensity, were very nearly equal in the X-unc and wild type preparations, suggesting a constant stoichiometric relationship.2 When the complex was prepared from the induced, nontransducing X control by the abbreviated procedure (Table 11), these eight subunits were found as major componen@ in the same relative proportion (Fig. 1). However, other contaminants were also observed, the most prominent being polypeptides with apparent molecular weights of 76,000, 26,000, and 15,000. The relative amount of these contaminants in the induced X-unc preparation was negligible because of the 6-fold  19,OOO. and a true molecular weight of 8,400, respectively; a, d, and e indicate contaminants with apparent molecular weights of 76,000,26,000, and 15,000. increase in F1Fo subunits relative to general membrane proteins.
The results described above indicate that the eight subunits found in FIFo preparations of high purity are all overproduced in equal proportion in the induced X-unc strain. This was directly demonstrated for most of the subunits by analysis of membranes and partially purified fractions (Fig. 2) 1 and 10, FI, 3 pg; lanes 2 and 9, FIFo purified by complete procedure, 5 pg; lanes 3 and 4, membranes of X-unc or control, 40 pg; lanes 5 and 6, particulate dialyzed extract of detergent-solubilized ATPase from A-unc or conintensity of subunits in the region of the y, 8, and e subunits are also apparent, but their identification is more tenuous due to other intensely staining polypeptides in these regions.:' However, the overproduction of all eight polypeptides is apparent in the detergent-solubilized fraction from the membrane. The induced X-unc membrane shows an enrichment for several polypeptides other than the eight cited above, the most obvious having apparent molecular weights of 31,000, 20,000, 15,000, and 11,000. These are probably membrane proteins coded for by the segment of transducing DNA carried on the phage, which corresponds to the region between bglB and asn on the E. coli chromosome. The M , = 31,000 protein was extracted from the membrane by the procedure used to solubilize the DCCD-sensitive ATPase complex, but did not co-purify with the FIFO complex during sucrose gradient centrifugation (Fig. 2).
In our earlier report (7), a polypeptide with an apparent molecular weight of 14,000 co-purified with the FIFO-ATPase complex prepared from cells grown on succinate/acetate/malate and could not be ruled out as a possible component of the FIFO complex in cells grown under these conditions. When The overproduction of the y subunit in the membrane fraction was masked by an intensely staining band of outer membrane protein which migrated to this position. This outer membrane protein migrates anomalously in SDS gels, occasionally to a position of higher apparent molecular weight (26). On the several occasions in which this outer membrane protein migrated with decreased mobility, the overproduction of the y subunit in the membrane fraction has also been obvious. trol, 25 pg; lanes 7 and 8, sucrose gradient pool of FIFO from X-unc or control, 5 pg. Greek letters refer to Fl subunits; 24K. 19K. and 8K refer to Fo subunits; arrocus point to induced proteins of unknown identity in X-unc membranes with apparent molecular weights of 31,000,20,000, 15,000, and 11,OOO.
FIFO was purified from the X-unc lysogen grown on these carbon sources, overproduction of a polypeptide of this molecular weight was not detected.
Energy-transducing Activities of the Purified F1 Fo Preparations-The F'Fo-ATPase purified from the induced X-unc strain exhibited all of the energy-transducing properties of the complex purified by our original procedure (7). The ATPase extracted from induced X-unc membranes exhibited equivalent sensitivity to DCCD, i.e. >70% inhibition of ATPase activity at all stages of purification. At least 80% inhibition of ATPase activity by DCCD was observed with the purified, reconstituted complex. The FIFO from the induced X-unc lysogen exhibited ATP-dependent quenching of quinacrine fluorescence, which is a qualitative index of ATP-coupled proton translocation. The Fn sector prepared from the induced X-unc lysogen by the method of Negrin et al. (22) demonstrated equivalent proton translocation activity (per micrograms of protein) as the Fo prepared from a nonlysogenic wild type strain.

DISCUSSION
Purification of an FIFo-ATPase from E. coli was previously reported to yield a complex composed of eight nonidentical polypeptides. The question remained as to whether all eight components were authentic subunits of the complex. Here we have shown that all eight components co-purify in constant stoichiometric proportion from membranes enriched 6-fold for the ATPase complex. This result by itself strongly suggests that each of the eight subunits is an unc gene product. However, one could envision such a result if one of these polypeptides was a contaminant that was coded for elsewhere in the chromosome if this contaminant was normally produced in a 6-fold or greater excess over the ATPase complex. This possibility can be dismissed since on comparing membranes from the induced X-unc strain to control membranes we observed an obvious increase in intensity of polypeptides corresponding in molecular weight to a, p, y, 24,000, 19,000, and 8,400 (DCCD-binding protein). Subunits S and also appeared to be overproduced in the membrane although identification was less certain. An unlikely possibility that cannot be ruled out is that the 18-megadalton segment of DNA (0.65% of the E. coli chromosome) carried on the X phage codes not only for the ATPase subunits but also for a contaminant that fortuitously co-purifies with the complex.
These experiments suggest but do not prove that all eight subunits are coded by genes at the unc locus. Conceivably, a regulator produced at the unc locus could promote overproduction by genes at another chromosomal location. However, in considering this possibility, it should be noted that mutations affecting the ATPase complex have never been mapped in chromosomal locations other than unc (15).
The eight-subunit ATPase complex has been demonstrated to be active in several properties indicative of its function in oxidative phosphorylation. These include ?'PPi-ATP exchange (7), ATP-driven proton pumping as judged by quinacrine quenching (7), and Fo-mediated proton translocation (22). Despite these demonstrations, it is possible that the complex is composed of more than eight subunits in vivo, some of which are lost during purification. Several other membrane proteins were induced during X-unc prophage induction. The M , = 31,000 protein, which was solubilized with the ATPase complex, likely corresponds to a DNA-binding protein of the outer membrane that has been shown to be coded for by this segment of DNA (27). One of the others may be a P-glucoside transport protein (Enzyme 11) coded for by bgZC which is also carried on the X-unc DNA (24,28).

Other workers have recently reported DCCD-sensitive
ATPase preparations of E. coli. The preparation of Friedl et al. (29) is very similar to that discussed here, but does contain components which we think correspond to the M , = 76,000 and 26,000 contaminants discussed in Fig. 1 and the original purification paper (7). This preparation exhibited energytransducing activities similar to those described above (29). Friedl et al. (29) have cited preliminary findings indicating that the major subunits in their preparation can be synthesized in vitro from the DNA of an independently constructed X-unc-transducing phage, but the data supporting this claim have yet to be published. The composition of the preparation reported by Rosen and Hasan (30) differs significantly from that discussed here and that reported by Friedl et al. (29). Their preparation seemed to lack not only the M , = 24,000 and 19,000 subunits of Fo, but also the 6 subunit of F,. The sensitivity of this ATPase preparation to inhibition by DCCD was greater than that of F1 but less than that of membranes. However, due to its wide reactivity, DCCD is not an entirely specific inhibitor (41, and it wiU under appropriate conditions inhibit the activity of the F,-ATPase (31). Energy-transducing activities were not demonstrated with this unusual preparation.
Prior to this work, there was little question that the a-E subunits of F1 and the M , = 8,400 subunit (DCCD-reactive proteolipid) were true components of FIFO. The results pre-sented here provide strong prima facia evidence that the M, = 24,000 and 19,000 polypeptides are also true subunits. It is of interest that Kagawa and co-workers (6,32) observed two subunits other than the proteolipid and those of F1 in their FIFO and Fo preparations from thermophilic bacterium PS3.
However, one of these subunits was not required in what appeared to be a fully reconstituted FIFo complex (13). Unequivocal proof that both the M , = 24,000 and 19,000 subunits are essential to the function of Fo will require more refined genetic and/or reconstitution experiments analogous to those reported for the a and P subunits (uncA and uncD genes) (5, 15, [16][17][18][19][20].