Genetic Evidence for Interaction between the a and b Subunits of the Fo Portion of the Escherichia coli Proton Translocating ATPase*

A mutation of the b subunit of the Escherichia coli proton translocating ATPase was previously described This mutation, which causes substitution of aspartic acid for glycine at position 9 (b,,g), results in loss of function of the ATPase complex. In this paper we describe the isolation and characterization of two mutations that partially suppress the effects of the bMpg alteration. The suppressor mutations cause amino acid substitutions at position 240 of the a subunit. Mem- branes derived from strains carrying a suppressor mutation and the bMpg mutation exhibited ATP-dependent proton translocating activity.

The proton translocating ATPase of Escherichia coli is a membrane bound enzyme complex that is similar in structure and function to enzymes of mitochondria and chloroplasts (for recent reviews, see Refs. 1,2). The E. coli enzyme is composed of eight subunits with a stoichiometry of alb2c6-12a3P3Y161~1. The genes that encode the ATPase subunits form an operon, the unc operon, located at minute 84 on the E. coli genetic map (3; Fig. 1).
The activities of the ATPase complex, proton translocation, and ATP synthesis/hydrolysis, are tightly coupled and are carried out by two separable domains of the complex, called Fo and F,. Fo, the proton-translocating domain, is composed of the three cytoplasmic membrane-embedded subunits a, b, and c. F1, composed of the remaining subunits, is located on the inner surface of the membrane and functions to either synthesize or hydrolyze ATP, depending upon the needs of the cell.
We have been investigating the structure and function of the ATPase through the isolation of mutants in which individual subunits have been inactivated. Previously, we described a mutation that alters the b subunit, causing a substitution of aspartic acid for glycine at position 9 (bmpg) (4). As a result of this mutation, the mutant strain was defective in ATP driven proton translocation. However, Fo was still able to bind a reduced amount of F,.
In order to investigate the interactions between the b subunit and other subunits of the ATPase, we have isolated mutations that suppress the effects of the bupg mutation. In * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement'' in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$Supported by a National Institutes of Health training grant. Present address: Dept. of Physiology, Tufts University School of Medicine, Boston, MA 02111. § Recipient of U. S. Public Health Service Grant GM18539 from the National Institutes of Health. this paper, we report the characterization of two mutations that alter the a subunit. Both mutations affect the same residue of a, proline 240, causing different changes.

EXPERIMENTAL PROCEDURES
Bacteriological Techniques-Bacterial strains are listed in Table I; all strains are E. coli K12 derivatives. For techniques such as growth of bacteria and phage Pluir, P1 transduction, and bacterial conjugation, standard procedures were used (5). Rich medium was L broth and minimal medium was A salts supplemented with sugars at 0.2%. For measurement of saturation density, single colonies were inoculated into L broth containing ampicillin, the cultures were incubated overnight, and the A m of the stationary phase cultures was measured.
Suppressor Mutant Isohtion-Strains CK1880/pRPG51.8l (AuncFH/bw&) and CK1880/pCK48 (AuncFH/b,&) are unable to grow on succinate minimal medium (Suc-)' because the mutation of the b subunit inactivates the ATPase complex. Suc+ revertants were selected on this medium, pooled, grown in succinate minimal liquid medium and used to prepare lysates of Pluir. The lysates were used to transduce strain CK1804/pRPG51.8 (i1u::TnlO AuncB-C/b,,A) to Ilv+. Ilv+ transductants were screened for Suc+, and cotransductants were candidates for strains carrying suppressor mutations on their chromosomes. Transductants derived from independent pools of revertants were used for further studies. Pluir lysates were prepared from each transductant and used to transduce CK1804 (i1u::TnlO AuncB-C) to Ilv+(Suc-). Recombinants were screened by introduction of pCK47 (b,8), a cosmid packaged in lambda heads. If the strain became Suc+, the chromosome from the Pluir donor had been introduced. The recombinants were then transformed with pRPG51. 8 (b,&) and tested for Suc+ phenotype to verify that the suppressing mutation had been introduced into the strain. To construct the strains used for further biochemical characterization, the recAl mutation was introduced by cotransduction with srl and screening for UV sensitivity.
Recombination of unc Operon onto a Plasmid-The suppressing mutations were transduced into strain CK1871 (Hfr Cavalli i1u::TnlO AuncB-C). The resultant strains were transformed with pWSB18, a plasmid that carries 2.5 kilobases of DNA upstream and downstream of the unc operon but is deleted for the operon. To select for recombination between the chromosome and plasmid, these strains were mated with CK180l/pRPG51 (AuncB-C/b,a). If the plasmid recombined with the chromosome of the donor, it would be transferred to the recipient as a cointegrate (6). Once transferred, a second recombination event, occurring in the recipient cell, could allow the chromosome of the Hfr donor to replace the unc deletion on the plasmid.
As a result of these two crosses, the suppressing mutations would be recombined onto the plasmid. Recombinants from the above mating were pooled, plasmids were isolated, and the plasmids were transformed into strain CK1803/pRPG51 (AuncB-C recAl/b,d). Plasmids that carried the genes for the unc operon were identified by screening; such plasmids were designated pCK41, pCK41.1, or pCK41.12.
Recombinant DNA Techniques-Procedures for preparation of When a strain is referred to in the text, we also present the relevant genotype; the chromosomal genes are described first and the ATPase subunits encoded on plasmids are listed after the slash.
The abbreviations used are: Suc-, unable to grow on succinate minimal medium; Ilv+, able to grow on minimal medium without addition of isoleucine and valine; DCCD, dicyclohexylcarbodiimide; MOPS, 4-morpholinepropanesulfonic acid. plasmid DNA, digestion with restriction enzymes, ligation with T4 DNA ligase, agarose, and acrylamide gel electrophoresis, and transformation of competent cells have been described (7). pCK47 and pCK48 contain the EcoRI fragment from pRPG51 and pRPG51.8, respectively, cloned into pREG152 (8). pCK43 was reconstructed from pCK41 by digestion with BamHI and BglII and ligation. The HindIII-Sal1 fragment containing the uncB,E and most of uncA genes from pCK41 was cloned into pBR322 to generate pCK44. pCK45 was derived from pCK44 by deletion of the AuaI fragments; pCK46 was derived from pCK44 by deletion of the BamHI fragment.
DNA Sequencing-The BamHI-XhoI fragments from pCK44, pCK44.1, and pCK44.12 were cloned into M13mp19 RF DNA using standard techniques (9). Determination of the DNA sequence was by the Sanger dideoxynucleotide method as described (9). Two isolates were sequenced for each M13 cloning.
Preparation and Assay of Membranes-Membranes were isolated essentially as described (10). Briefly, cells from midlog cultures were harvested, suspended at 4 ml/gm wet weight, and disrupted in a French pressure cell. The lysate was clarified by low speed centrifugation, and membranes were pelleted, resuspended, pelleted, and then resuspended at 2 ml/gm wet weight of cells. The fluorescence quenching assay was performed as previously described (11)  ATPase (12) and Lowry (13) assays were performed as described. For DCCD treatment, membranes at a final concentration of 2.5 mg of protein/ml were incubated with 50 p~ DCCD in 50 mM MOPS KOH, 10 mM MgC12, pH 7.3 at 37 "C for 15 min. To strip the membranes of F,, the procedure of Fillingame et al. (14) was used. The stripped membranes were reconstituted with purified F, (11) by incubation at 30 "C for 5 min. isolated by selecting for growth on succinate as the sole carbon and energy source. Several independent mutants that carried suppressor mutations on their chromosomes were isolated as described under "Experimental Procedures." Two independently isolated chromosomal mutations that allowed growth of the bmPg mutant on succinate, designated uncB2031 and uncB2032, were chosen for detailed analysis.

Isolation of Suppressing Mutations-Starting with strains carrying the bwpg mutation, spontaneous Suc+ revertants were
The parental strains, carrying the bmpg mutation alone, exhibited no growth on succinate medium. Strains that carried both a suppressor mutation and the bwpg mutation grew well on succinate and were indistinguishable from the wild type strain. Another phenotype of unc mutant strains is that they do not grow to the high saturation density of a wild type strain, even in rich medium. As shown in Table 11, the suppressor mutations also reverse this defect.
To test the effects of the suppressor mutations with different uncF (b subunit) alleles, strains CK1858 (uncB2031) and CK1861 (uncB2032) were transformed with pRPG51 (bJ) or pRPG51.7 (basplJ16). Strains CK1858/pRPG51 and CK1861/ pRPG51 grew well on succinate minimal medium (data not shown); therefore, the subunits that were altered by the suppressor mutations were functional in the presence of a wild type b subunit. However, the suppressor mutations did not allow growth on succinate in the presence of a different mutation of the b subunit, bwp131 (data not shown). These results showed that the suppressor mutations exhibited allele specificity and were not bypass mutations that eliminated the need for the b subunit.
Amino Acid Changes Caused by the Suppressing Mutations-To determine the nature of the suppressor mutations, the unc operon was recombined from the chromosome of the mutants onto a plasmid as described under "Experimental Procedures." The resultant plasmids, pCK41, pCK41.1, pCK41.12, carried all the genes of the unc operon, except the genes for uncF and H (Fig. 1). When these plasmids were transformed into strain CK1804/pCK48 (AuncB-C/b,&), the plasmids derived from the suppressor mutants rendered the cells capable of growth on succinate, while the plasmid derived from the parental (wild type) strain did not allow growth on succinate. This demonstrated that the suppressor mutations were located in the unc operon and had been recombined onto the plasmid.
We used merodiploids to show that the suppressor mutations were dominant. The merodiploids carried uncB2031 or uncB2032 on their chromosomes and wild type copies of two or three unc genes on plasmids. Despite the presence of the wild type genes, the suppression effect was still observed. Therefore, the dominant effect was used to map the suppressor mutations on plasmids pCK41.1 and pCK41.12.
The uncB,E and most of the uncA genes were subcloned from pCK41, pCK41.1, and pCK41.12 to generate pCK44, pCK44.1, and pCK44.12 ( Fig. 1) as described under "Experimental Procedures." The plasmids were transformed into strain CK1881 (AuncFH) carrying the compatible plasmid pCK48 (bmp&) in order to test the effects of the suppressor mutations. As shown in Table  111, diploids containing   Genotype of chromosomal unc operon of strain CK1881.
pCK44.1 or pCK44.12 exhibited the suppressor phenotype. The suppression is weaker in this case, presumably due to the presence of a wild type copy of the mutant subunit. The diploid containing pCK44 (a,c) did not show such improved growth in the saturation density test. Deletion of the genes for c and the fragment of 01, generating plasmids pCK45 (a), pCK45.1, pCK45.12 ( Fig. l), had no effect on these phenotypes (Table 111). Therefore, the suppressor mutations affected the a subunit. In confirmation of this result, disruption of the uncB gene by deletion of the BamHI fragment from pCK44 (a,c), pCK44.1, and pCK44.12, to generate pCK46 (c), pCK46.1, and pCK46.12, respectively (Fig. l), eliminated the suppression phenotype (Table 111). Although pCK44 had no effect on saturation density, we found that the diploid strain CK1881/pCK48/pCK44 (AuncFH/b,,&/a,c) grew slightly on succinate minimal medium. This effect was also observed with a diploid carrying pRPG56 (a,c) and with a diploid carrying pCK45 (a). Apparently, overproduction of the a subunit caused a slight suppression of the defect caused by the bWpg subunit.
The DNA sequence of the 3' end of the uncB gene and the intergenic region between uncB and E was determined as described under "Experimental Procedures." A single base substitution was observed in both mutants. For mutant uncB2031, a C to T transition occurred at base pair 1742 (data not shown; base pairs are numbered as in Ref. l ) , resulting in conversion of proline at residue 240 to leucine (Fig. 2). For mutant uncB2032, a C to G transversion a t base pair 1741 occurred (data not shown), causing alanine to be substituted for proline 240 (Fig. 2). No other base changes were observed either in the uncB coding region or in the uncB-E intergenic region. For the wild type and mutant uncB2032, the sequence GGCC at base pair 1746 could not be read unambiguously from the sequencing gel; this sequence was confirmed to be identical to the published sequence (1,3) by digestion of pCK45 ( a ) , pCK45.1, and pCK45.12 with HueIII, which digested this sequence. Therefore, the results showed that alteration of proline 240 of a to either leucine or alanine suppressed the effects of the bWpg mutation.
Effects of the Suppressing Mutations-To determine the effects of the a suppressor mutations, membranes were prepared from various strains and tested for their ability to be energized by NADH and ATP, using 9-amino-6-chloro-2methoxyacridine fluorescence quenching as the assay. As shown in Fig. 3A, trace 1, membranes from the bmPg strain were inactive with ATP, although when the same membranes were retested with higher concentrations of ATP a small amount of fluorescence quenching was sometimes observed (data not shown). Both a suppressor mutants (Fig. 3A, traces 2 and 3) retored ATP driven quenching to about 40% of the wild type value (Fig. 3A, trace 4 ) . This quenching was sensitive to DCCD, demonstrating that proton translocation occurred through Fo (data not shown). When the uncB2031 (aku24o) mutant was tested with the bospIdI mutation, no ATP driven fluorescence quenching was observed (Fig. 3B, trace 5 ) ; the bWpIdI mutant alone was also inactive (Fig. 3B, trace 6). In the presence of the wild type b subunit, the aIeu240 strain (Fig. 3B, trace 8) was indistinguishable from wild type (Fig. 3B, trace  7). Identical results were obtained with the uncB2032 (a.h24o) strain (data not shown). These results are consistent with the succinate growth experiments described above.
To demonstrate that the a suppressor mutations affected Fo, the membranes were stripped of Fl and reconstituted with purified wild type F1. When the stripped membranes from the Ala Leu T 7  0.56 29 0.14 33

TGG ATC CTG AAT GTG CCG TGG GCC ATT TTC CAC ATC CTG ATC Trp Ile Leu Asn Val Pro Tru Ala Ile Phe His Ile Leu Ile
Only the types of a and b subunits are listed; the superscript + is used to denote wild type subunits. All the other ATPase subunits are wild type.
pmol Pi/min/mg of protein.
Membranes were incubated at 37 "C for 15 min with and without 50 p~ DCCD as described under "Experimental Procedures." DCCD sensitive specific activity was the difference in the activity of the sample incubated without DCCD (which declined from the original activity) and with DCCD. wild type strain were assayed, a reduction in NADH driven fluorescence quenching was observed (Fig. 3C, truce 9). This result showed that Fo was active and partially dissipated the proton gradient generated by respiration. Reduction in NADH driven fluorescence quenching was not seen in stripped membranes from either the u~u240b,p9 strain ( Fig. 3 0 , truce 1 I), or the bUp3 mutant (Fig. 3E, truce 13). When the stripped membranes were reconstituted with wild type F,, the original level of ATP-stimulated proton translocation was observed. Fig.  3C, truce 10 shows reconstituted membranes from the wild type strain. Fig. 3 0 , truce 12 shows the akuz40b,pg reconstituted membranes. Fig. 3E, truce 14 shows the reconstituted membranes from the bwpg mutant alone; a high concentration of ATP was used, and some activity can be seen in the mutant membranes. Results identical to those shown in Fig. 3 0 were obtained when membranes from the uah240b,pg strain were stripped and reconstituted (data not shown).
The interactions between F, and Fo were not affected by the u suppressor mutations. As shown in Table IV, the bwp9 mutation reduced F, binding but did not abolish it. The same amount of F, was bound to the membrane regardless of whether the suppressor mutations were present. In membranes from the bWp3 mutant strain, some of the membrane bound ATPase activity was sensitive to DCCD. The DCCD sensitive ATPase specific activity was also not increased by the suppressor mutations. Therefore, the interactions between F, and Fo were not changed by the suppressor mutations.
Also, the DCCD sensitivity of the complex was not correlated with the amount of coupled proton translocating activity. independently isolated suppressor mutants were characterized and both carried alterations of the same residue of a, proline 240. Although the amino acid substitutions were different, an aliphatic amino acid replaced proline in both cases, and the properties of the two mutants appeared to be identical. Membranes prepared from cells that carried the b, g mutation alone were partially inactive in ATP driven H' translocation. When the strain carried a suppressor mutation and the b, g mutation, membranes exhibited about 40% of wild type activity. F1 binding and DCCD sensitivity were not affected by the presence of the suppressor mutations.

DISCUSSION
The suppressor mutations probably affected the function of Fo, since membranes that were stripped of F1 and reconstituted with purified wild type F, recovered the activity that they exhibited without this treatment. However, we were not able to demonstrate passive proton permeability, leading to inhibition of NADH driven fluorescence quenching, in stripped membranes from the suppressor mutant strains. The rate of proton translocation through the suppressed mutant Fo may not be sufficiently fast to dissipate the gradient generated by respiration.
The suppression by the a mutations could be a result of direct effects on H' permeability or of effects on Fo formation. In the bmpg mutant strain, the subunits of FO are probably associated with one another since F1 binds to the membrane and since at least some of this membrane-bound ATPase is inhibited by DCCD. In addition, a small amount of ATPdependent proton translocation could occasionally be observed in membranes derived from the mutant (e.g. Fig. 3E). These observations suggest that Fo is formed in the bWpg mutant strain. The amount of membrane-bound F, and its sensitivity to DCCD are unchanged by the suppressor mutations. These observations are inconsistent with the hypothesis that the suppressor mutations result in an increased number of Fo complexes. Therefore, the more likely interpretation is that the suppressor mutations directly affect Fo function.
We suggest that the amino terminus of b and the carboxyl terminus of a interact and that this interaction is important for FO function. The amino terminus of b can be labeled with hydrophobic labeling reagents, demonstrating that this region of b is inserted into the membrane (17,18). The determination of the amino acid sequence of a has shown that it is a very hydrophobic protein (1,3); most of a is probably also located within the membrane. Therefore, these two subunits may be interacting within the membrane bilayer. Results from chemical cross-linking have also shown that at least some parts of a and b are close to one another in Fo (19). A model depicting putative structures of the Fo polypeptides is shown in Fig. 4 are based on analysis of their primary sequences, their accessibility to labeling reagents, and the properties of mutants (1, 2, 17, 20-23). The structure of the a subunit with five transmembrane helices has been proposed based on hydropathy analysis of the primary sequence and placement of evolutionarily conserved region^.^ Other similar structures have also been proposed (1, 2, 20, 24). We suggest that the region of a containing proline 240 is near residue 9 of b, resulting in the orientation of a shown in Fig. 4.
A similar analysis was conducted by Jans et al. (25). Under the conditions that they used (single copy of gene for bwpg), mutant b subunits and ATPase activity were not found associated with the membrane, and they concluded that there was a defect in assembly. The defect in assembly is not seen as strongly in our experiments in which the mutant bmPg is carried on a plasmid. Jans et al. (25) also isolated a mutation that suppressed the effects of the bmPg mutation and suggested that this suppressor mutation affected the gene for a. The a suppressor appeared to reverse the defect in assembly that they observed. Therefore, the mechanism of suppression by the a mutant may differ from the mechanism of suppression by the a suppressor mutants.
Our results support the idea that the b subunit affects the Fo proton channel through interaction with the a subunit. Studies in uiuo and in vitro have shown that b is essential for proton translocation. If b is extracted from Fo, the resultant ac complex is unable to translocate protons (26). In addition, membranes prepared from strains carrying mutations that cause premature termination of b are not capable of translocating protons through Fo (4, 27). However, once Fo has formed, the carboxyl terminus of b can be removed with protease, and the modified Fo can still translocate protons (28, 29). Therefore the amino terminus of b is sufficient for proton conductivity under these conditions. While the hydrophobic amino terminus of b may not itself carry protons, its association may affect the conformation of another protein that does, the a subunit. Cain and Simoni3 suggest that the carboxyl terminus of a may be directly involved in proton conduction. The conformation of this region of the a polypeptide may be influenced by interaction with the b subunit.
In the course of these studies, we observed an effect of a overproduction on the phenotype of the b, g mutant strain. Jans et al. (25) observed that membranes prepared from a strain carrying this mutation in single copy did not translocate protons in response to ATP. When the unc operon from this strain, including the mutation, was cloned onto a high copy plasmid and transformed into the same strain (carrying the bmpg mutation on its chromosome), the transformants regained ATPase .function. We suggest that this effect was due to the overproduction of a. This effect may be similar to that of the a suppressor mutations. At high concentrations of a, an equilibrium may shift to favor more of (I in a functional conformation despite the mutation in b.