The yeast F(1)F(0)-ATP synthase: analysis of the molecular organization of subunit g and the importance of a conserved GXXXG motif.

The F(1)F(0)-ATP synthase enzyme is located in the inner mitochondrial membrane, where it forms dimeric complexes. Dimerization of the ATP synthase involves the physical association of the neighboring membrane-embedded F(0)-sectors. In yeast, the F(0)-sector subunits g and e (Su g and Su e, respectively) play a key role in supporting the formation of ATP synthase dimers. In this study we have focused on Su g to gain a better understanding of the function and the molecular organization of this subunit within the ATP synthase complex. Su g proteins contain a GXXXG motif (G is glycine, and X is any amino acid) in their single transmembrane segment. GXXXG can be a dimerization motif that supports helix-helix interactions between neighboring transmembrane segments. We demonstrate here that the GXXXG motif is important for the function and in particular for the stability of Su g within the ATP synthase. Using site-directed mutagenesis and cross-linking approaches, we demonstrate that Su g and Su e interact, and our findings emphasize the importance of the membrane anchor regions of these proteins for their interaction. Su e also contains a conserved GXXXG motif in its membrane anchor. However, data presented here would suggest that an intact GXXXG motif in Su g is not essential for the Su g-Su e interaction. We suggest that the GXXXG motif may not be the sole basis for a Su g-Su e interaction, and possibly these dimerization motifs may enable both Su g and Su e to interact with another mitochondrial protein.

The F 1 F 0 -ATP synthase enzyme is located in the inner mitochondrial membrane, where it forms dimeric complexes. Dimerization of the ATP synthase involves the physical association of the neighboring membrane-embedded F 0 -sectors. In yeast, the F 0 -sector subunits g and e (Su g and Su e, respectively) play a key role in supporting the formation of ATP synthase dimers. In this study we have focused on Su g to gain a better understanding of the function and the molecular organization of this subunit within the ATP synthase complex. Su g proteins contain a GXXXG motif (G is glycine, and X is any amino acid) in their single transmembrane segment. GXXXG can be a dimerization motif that supports helix-helix interactions between neighboring transmembrane segments. We demonstrate here that the GXXXG motif is important for the function and in particular for the stability of Su g within the ATP synthase. Using sitedirected mutagenesis and cross-linking approaches, we demonstrate that Su g and Su e interact, and our findings emphasize the importance of the membrane anchor regions of these proteins for their interaction. Su e also contains a conserved GXXXG motif in its membrane anchor. However, data presented here would suggest that an intact GXXXG motif in Su g is not essential for the Su g-Su e interaction. We suggest that the GXXXG motif may not be the sole basis for a Su g-Su e interaction, and possibly these dimerization motifs may enable both Su g and Su e to interact with another mitochondrial protein.
The F 1 F 0 -ATP synthase enzyme plays a pivotal role in the aerobic production of ATP (adenosine triphosphate) in eukaryotic cells. This enzyme is located in the inner membrane of mitochondria and is composed of a large number of different subunits (1)(2)(3)(4). Functionally, the enzyme can be divided into two parts or subcomplexes, the inner membrane embedded F 0 -sector and the associated hydrophilic F 1 -sector, which protrudes into the mitochondrial matrix. The F 0 -sector is responsible for the passage of protons from the intermembrane space into the matrix, a step coupled to the F 1 -catalyzed synthesis of ATP from ADP (adenosine diphosphate) and phosphate (4).
In mitochondria the F 0 -sector is composed of a number of different polypeptides, which are both nuclear-and mitochondria-encoded. In yeast, for example, at least 10 different sub-units make up this subcomplex, three of which (subunits 6, 8, and 9) are encoded by the mitochondrial DNA (mtDNA) 1 (2,3). The F 0 -sector of the mitochondrial enzyme is, thus, more complex than that of its bacterial counterpart, where only three subunits (subunits a, b, and c) have been described. In addition to mediating the passage of H ϩ ions across the inner membrane, the F 0 -sector also has been described to physically associate with neighboring F 0 -sectors, which results in the dimerization of ATP synthase complexes in the mitochondrial inner membrane (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15). Three F 0 -sector subunits, subunit 4 (Su 4, equivalent to bacterial subunit b) and the two recently identified subunits, subunits e and g (Su e and Su g, respectively) have been shown to play a critical role in the process of F 0sector dimerization (5)(6)(7)(8)(15)(16)(17)(18). A number of independent lines of evidence for the ATP synthase dimerization have been presented in the literature over the recent years. First, molecular sizing analysis, either gel filtration or native gel electrophoresis of the ATP synthase complexes solubilized from mitochondrial membranes with a mild detergent, demonstrated that the ATP synthase can be isolated as a complex whose mass (ϳ1000 kDa) was consistent with that of a dimeric complex (5,6,9,14). Subsequent chemical cross-linking approaches adopted by Velours and co-workers (7,17), demonstrated the close physical proximity of two Su 4 proteins in the mitochondrial inner membrane. Because the stoichiometry of Su 4 is known to be 1 Su 4 per F 0 -sector, the observed Su 4-Su 4 cross-linking provides evidence for F 0 -F 0 interaction in the mitochondrial membrane (7,17). More recently, in vivo studies using fluorescent resonance energy transfer technology has also supported a close physical association of one F 1 F 0 -ATP synthase complex with another (12,13).
Gene knock-out studies in yeast combined with biochemical analysis of the ATP synthase complex have indicated that the conserved Su e and Su g proteins are not essential for the enzymatic activity of the complex (6,9,17). The ATP synthesis capacity of the complex has been reported to be partially compromised in the absence of Su e and Su g, however (9,14,18,19), and consequently, the null mutants exhibit a weak growth phenotype on non-fermentable carbon sources (6,9,18,19). Further analysis of the mitochondria from the null mutant strains, ⌬su e and ⌬su g, respectively, however, revealed that these mutants exhibit two interesting phenotypes. First, in the absence of Su e and/or Su g, the ATP synthase complex failed to assemble into detergent stable dimers, as indicated by native gel electrophoresis (6,9,14,15). However, Su 4-Su 4 crosslinking has been reported in ⌬su e and ⌬su g mitochondria, indicating that a close physical proximity of ATP synthase complexes can still exist in the absence of Su e and Su g proteins (9). These Su 4-Su 4-mediated interactions appear not to be stable in the absence of Su e and Su g, however, as only monomeric ATP synthase complexes are observed after detergent solubilization (6,9,14,15). Second, the importance of the Su g and Su e proteins have been further highlighted by the recent demonstration that the presence of these proteins are required to develop normal mitochondrial morphology, in particular development of the characteristic cristae structure of the inner membrane (9,14,16). Cristae are finger-like invaginations of the inner mitochondrial membrane, protruding into the mitochondrial matrix and are distinct from the inner boundary membrane, which is in close proximity to the outer mitochondrial membrane. In the absence of Su g or Su e, the mitochondrial inner membrane fails to partition into cristae invaginations but, rather, accumulates as onion-type membranes (9,14,16).
To further understand how Su e and Su g may affect the ATP synthase and its organization in the inner mitochondrial membrane, it is important to gain a better understanding of the molecular environment of these subunits within the ATP synthase and to identify the proteins they interact with. This present study focuses on the Su g protein. Su g is a relatively small protein (115 amino acid residues) and spans the inner mitochondrial membrane in an N in -C out orientation (6,19). Su g has a single transmembrane segment (residues 88 -106) that is located close to the C-terminal end of the protein (Fig. 1A). The bulk of the protein, the N-terminal hydrophilic domain (residues 1-87), is located in the mitochondrial matrix ( Fig. 1A) (6). Alignment of Su g amino acid sequences indicates a high level of conservation in the C-terminal region of Su g, in particular the transmembrane segment (Fig. 1B). A tyrosine residue (Tyr-112) located immediately after the transmembrane region and two glycine residues (Gly-101 and Gly-105) within the transmembrane segment are conserved in all known Su g proteins (20). These conserved glycine residues form a motif GXXXG (G is Gly, and X represents any amino acid), which has been shown to be involved in the homo-or heterodimerization of other membrane proteins, e.g. glycophorin A (21)(22)(23)(24)(25). GXXXG motifs can form the framework for transmembrane helix-helix interactions, where the Gly residues and the spacing of three residues between them, have been shown to be essential aspects of the motif as they ensure a close helix-helix packing interface. The highly conserved nature of the GXXXG motif in Su g is indicative of its potential importance for the function of Su g, and it is possible that this motif may be involved in transmembrane helix-helix interactions between neighboring Su g proteins (homodimerization) or with another F 0 -sector subunit (heterodimerization).
Using site-directed mutagenesis combined with a cross-linking approach, we have addressed here the molecular environment of Su g within the F 0 -ATPase complex. We provide evidence for a close proximity between Su g and the Su e protein and also for a putative Su g homodimerization. Furthermore, we show that alterations in the Su g protein can cause significant changes in the molecular environment of Su e. We also demonstrate the importance of the conserved GXXXG motif for the stability of Su g and its interaction with the ATP synthase.
Cloning and Expression of Mutated and HA-tagged Su g Derivatives-The entire open reading frame encoding Su g (ATP20 gene) was amplified by the PCR as a SpeI-PstI fragment. When indicated, the reverse primer used contained the sequence information to encode the hemagglutinin (HA) epitope before the translational stop codon in addition to a Su g-specific sequence for priming. In addition, residues Gly-101, Gly-l05, Val-110, and Tyr-112 were individually mutated as indicated by incorporating the mutated sequence into the corresponding Su g reverse primer. The recombinant PCR products were cloned into a yeast integration vector Yip351, which contained a galactose-inducible GAL10 promoter, and the LEU2 auxotrophic marker gene (15,27). The resulting plasmids were linearized at a unique ClaI restriction site located in the 5Ј region of the LEU2 gene locus and transformed into the ⌬su g null mutant using the protocol described by Knop et al. (28), and leucine-positive transformants were selected. Mitochondria were isolated from each of the transformants, and the expression and mitochondrial localization of the Su g derivatives were verified by Western blotting using an antibody specific for the HA epitope (Covance Research Products) or Su g antiserum (6).
Growth Curve Analysis-All strains were initially grown overnight at 30°C in YPG medium supplemented with 0.2% galactose and histidine (0.06 mg/ml) and leucine (0.26 mg/ml). The cells were inoculated in the same medium at an A 580 nm of 0.2, and growth of the cells was monitored by measuring the A 580 nm at 2-h intervals over a period of 10 h. rho 0 /rho Ϫ Cell Conversion Detection Assay-All yeast strains were maintained on YPG plates and were used to inoculate YP-lactate (0.5% lactate) supplemented with 2% galactose and allowed to grow overnight at 30°C. The A 580 nm of the overnight cultures was measured, and an equivalent number of cells from each culture was plated onto YPD (2% glucose) or YPG (3% glycerol, supplemented with 0.1% galactose) plates. After incubation at 30°C, the colonies were counted, and the number of rho 0 /rho Ϫ cells (i.e. petite cells) was calculated and expressed as a % of total colonies on the YPD plate.
Chemical Cross-linking of Su g Protein with 5,5Ј-Dithiobis(2-nitrobenzoic Acid) (DTNB)-Isolated mitochondria (100 g of protein) were resuspended in SH buffer (0.6 M sorbitol, 20 mM Hepes, pH 7.2) at a protein concentration of 0.5 mg/ml. Cross-linking was performed on ice for 30 min in the presence of the sulfhydryl-specific homobifunctional reagent DTNB ((0.2 mM). After quenching with 10 mM N-ethylmaleimide and 10 mM EDTA, pH 8.0, mitochondria were re-isolated by centrifugation, lysed in the presence of SDS-containing sample buffer (without ␤-mercaptoethanol), and cooked at 95°C for 5 min. Samples were analyzed by SDS-PAGE using a 16% acrylamide, 0.6% bisacrylamide gel followed by Western blotting (10).
Miscellaneous-Protein determinations and SDS-PAGE were performed according to published methods (29,30). The Western blot analysis and antibody decoration were performed using available Su e, Su g, and Su k antisera raised against peptides corresponding to the C-terminal region of each of the proteins, respectively (6). The F 1antibody was raised in chicken after injection of purified yeast F 1 -sector and was obtained from Dr. David Mueller (The Chicago Medical School). Clear native-PAGE analysis of F 1 F 0 -ATP synthase complexes after solubilization with digitonin was performed essentially as described previously (15).

RESULTS
Expression of the Mutated Su g Proteins-We adopted a cysteine mutagenesis and cross-linking approach to identifying proteins that may interact with Su g and also address the functional significance of the GXXXG motif in Su g. Su g contains one Cys residue at position 75, which is located in the matrix-exposed, hydrophilic region of Su g (Fig. 1A). We introduced an additional Cys residue into the C terminus of Su g, close to the transmembrane region and the conserved GXXXG motif. Two Su g mutants, Su gY112C and Su gV110C, were created whereby the Tyr and Val residues at positions 112 and 110 (Fig. 1, A and B) were individually mutated to Cys residues. We specifically chose to introduce the Cys residue into the C-terminal region of Su g as we wanted to analyze if Su g is in close proximity to Su e. Su e contains a unique Cys residue (Cys-27) immediately C-terminal to its membrane anchor segment, i.e. located in the intermembrane space, a position similar to that of the introduced Cys residues (V110C and Y112C) of Su g. The two resulting Su g derivatives, Su gY112C and Su gV110C, were cloned either in the absence or presence of an additional C-terminal hemagglutinin tag (HA tag) (Su gY112C HA and Su gV110C HA , respectively) and were expressed in a yeast su g null mutant. As a control we also expressed the wild-type Su g protein (i.e. no additional Cys residues) with the C-terminal HA tag, Su g HA .
Mitochondria were isolated from the resulting transformants, and the presence of the Su g derivatives was verified using either Su g-or HA-specific antiserum, as appropriate (Fig. 1C). The derivatives Su gY112C (Fig. 1C) and in particular the Su gV110C (results not shown) were poorly recognized by the Su g antisera, in contrast to the Su gY112C HA and Su gV110C HA derivatives, which were clearly detectable with the HA antiserum (Fig. 1C). Because the existing Su g antibody was raised against a synthetic peptide corresponding to the C-terminal 13 amino acid residues (residues 103-115, respectively), it is possible that mutation of Tyr-112 or Val-110 to a Cys residue may have adversely affected the ability of the epitope to be recognized by the Su g antibody. The steady state levels of both Su gY112C HA and Su gV110C HA were similar to those of the wild-type HA-tagged construct. We, therefore, conclude that the alteration of residues Val-110 or Tyr-112 to cysteine did not hinder the mitochondrial targeting or stability of the Su g protein.
The steady state levels of other F 0 -sector subunits, such as Su e, Su k, and Su 4 proteins did not appear to be affected by the expression of the HA-tagged or mutated derivatives of the Su g protein (Fig. 1C). The levels of these proteins each appeared to be similar to those of wild-type mitochondria, thus demonstrating the assembly of the F 0 -sector subunits was not hindered through the expression of the mutated Su g derivatives. In addition, growth curve analysis indicated that the strains harboring the Cys-modified Su g derivatives grew in a similar manner as the control wild-type strain on a non-fermentable carbon source, e.g. glycerol (Table I). As previously described the su g null strain, ⌬su g, displayed a slightly slower growth phenotype under these conditions (6,19).
Yeast cells lacking the Su g protein display an increased potential to undergo loss of their mtDNA (14,18) (Table I).
Expression of the wild-type Su g bearing the HA tag or the Su gV110C HA and Su gY112C HA derivatives in the su g null mutant background prevented the high frequency of spontaneous rho 0 /rho Ϫ formation, which was observed in the ⌬su g strain in the absence of an expressed Su g protein (Table I).
Taken together these results suggest that introduction of a Cys residue at position 110 or 112 does not appear to adversely affect the stability of Su g. Thus, despite being conserved among the Su g protein family, mutation of residue Tyr-112 to a Cys does not appear to adversely affect the function of Su g protein, at least under the conditions tested here.
Su g and Su e Proteins Are Found in Close Proximity to Each Other-Cross-linking studies using the sulfhydryl-specific reagent DTNB (Ellman's reagent) were performed with isolated mitochondria harboring the wild-type or the Cys-modified Su g derivatives. DTNB was chosen as a cross-linking agent due to fact that it is a "zero-length" spacer arm cross-linking agent; thus, the ability to be cross-linked by DTNB is indicative of the close proximity of two neighboring sulfhydryl residues. Wildtype control, Su gY112C, or Su gY112C HA mitochondria were incubated with DTNB or were mock-treated, and Su g and its cross-linked adducts were subsequently resolved on SDS-PAGE, Western-blotted, and subjected to immune-decoration with antibodies specific for Su g or the HA epitope ( Fig. 2A). In the presence of DTNB, Su gY112C formed a cross-linked adduct of ϳ23 kDa. This adduct corresponds to a product of Su g (ϳ12 kDa) cross-linked to an ϳ11-kDa protein. The Su gY112C HA derivative also formed a similar adduct in the presence of DTNB, which was ϳ24 kDa in mass, the difference in size between the Su gY112C and Su gY112C HA adducts accounted for by the presence of the HA tag (ϳ1 kDa). In contrast the wild-type Su g, which lacks the additional Cys in the C-terminal region, did not form a similar cross-linked adduct and Schizosaccharomyces pombe (S.p., Q09774, Swiss-Prot). The positions of the conserved Gly and Tyr residues are highlighted, and the numbering of the amino acid residues is indicated. C, mitochondria were isolated from wild-type (WT) or the ⌬su g null mutant strain harboring the mutated Su g derivatives as indicated and were analyzed by SDS-PAGE, Western blotting, and immune-decoration with a Su g-specific antiserum (for WT, ⌬su g, and ⌬su g bearing the Su gY112C derivative) or with an HA-specific polyclonal antiserum (for ⌬su g bearing the Su g HA , Su gY112C HA , and Su gV110C HA derivatives). The steady state levels of other mitochondrial proteins Su e, Su k, and Su 4 of the ATP synthase and Tim23 (loading control) are indicated. Functional Analysis Su g of the Yeast F 1 F o -ATP Synthase when incubated with DTNB ( Fig. 2A). We, therefore, conclude that the Cys at position 112, located in the intermembrane space, is responsible for the cross-linking to the ϳ11-kDa protein. In addition, a cross-linked adduct of ϳ26 kDa, albeit with low efficiency, was observed in the Su gY112C HA mitochondria when analyzed with the HA-specific antibody. From the size of this adduct, this may correspond to a Su g-Su g homodimer. Su e, another F 0 -dimer specific subunit, also contains a conserved GXXXG motif in its single transmembrane segment (16). The presence of the GXXXG motif in Su e has been shown to be important for it to attain correct molecular positioning within the F 0 -sector of the ATP synthase complex (16). Because Su g depends on the presence of Su e for its stability, it is conceivable that these two proteins may physically interact in the membrane (15,16). To address if the 11-kDa protein, which becomes cross-linked to the Cys-modified Su g derivatives, represents Su e, a parallel DTNB-cross-linking experiment was immune-decorated with antibodies specific for Su e (Fig. 2B,  upper panel). In addition to forming the previously described Su e-Su e homodimer of 22 kDa (10) and an 18-kDa adduct (see the legend of Fig. 4 for further discussion), Su e was also observed to form cross-linked adducts of 23-and 24-kDa size in the presence of Su gY112C and Su gY112C HA derivatives, respectively. These Su e-containing adducts were notably absent in the wild-type mitochondria, which contained the authentic, i.e. non-Cys-modified Su g protein. Taking together, the mass of the Su e-containing adducts, the fact that they occur only in the presence of the Cys-modified Su g, and the observed increase in size of the adduct when cross-linking was performed in the HA-tagged Su g mutant mitochondria allow us to conclude that the observed 23-and 24-kDa adducts represent Su g-Su e-cross-linked partners. We conclude, therefore, that Su e and Su g can be found in close proximity to each other in the mitochondrial inner membrane. Interestingly, the Su e-Su g adducts were observed to form with a significantly greater efficiency than either the Su e-Su e homodimer or the putative Su g-Su g dimer.
Cross-linking of Su g to Su e was also observed using the second independent Su g Cys mutant, Su gV110C HA (Fig. 2B,  lower panel). However, the efficiency of cross-linking of Su g-Su e appeared to be greater in the case of the Su gY112C HA derivative. The observed difference in the cross-linking efficiency between Su gY112C HA and Su gV110C HA may be accounted for by a positioning effect of the residues. In the folded structure of Su g, residue Val-110 may be positioned away from the Cys-27 residue of Su e, whereas the residue 112 may be physically closer to it.
Mutations in the Conserved GXXXG Motif of Su g-As previously indicated, Su g contains a highly conserved GXXXG motif in its transmembrane segment. To address the importance of this motif for the stability of Su g and its molecular organization within the F 0 -sector, a series of mutated Su g derivatives was constructed. First, the conserved Gly residues of the 101 GXXXG 105 motif were individually replaced by bulkier hydrophobic residues, either Leu or Val. The resulting Su g mutants were cloned as HA-tagged proteins, Su gG101L HA , Su gG101V HA , Su gG105L HA , and Su gG105V HA . In addition, another Su g mutant was constructed whereby the GXXXG motif was disrupted by the insertion of an extra Ala residue between residues 103 and 104, i.e. conversion of GEIIG to GEIAIG, thus resulting in the creation of a HA-tagged Su g103A104 HA derivative.
The resulting Su g GXXXG motif mutants were expressed in the ⌬su g strain, and mitochondria were isolated and analyzed for the presence of the mutated Su g derivatives (Fig. 3, A and B). Immune-decoration with HA-specific antiserum indicated that the Su gG105V HA derivative accumulated in the mitochondria at levels similar to those of the wild-type Su g HA derivative, whereas the levels of Su gG105L HA were slightly lower (Fig. 3A). The reduced levels of the Su gG105L HA relative to the Su gG105V HA derivative may be due to the fact that a Leu residue is bulkier than Val and may, therefore, be less favorable for the stability of Su g. The steady state levels of the Su g103A104 HA protein and in particular, the Su gG101L HA and Su gG101V HA derivatives, were severely reduced when compared with the Su g HA levels, however (Fig. 3B).
Yeast strains harboring the Su gG101L HA , Su gG101V HA , and Su g103A104 HA derivatives, like the ⌬su g parent strain, displayed an enhanced frequency of rho 0 /rho Ϫ formation (Fig.  3C). The observed instability of the mtDNA is most likely a result of the reduced and, therefore, limiting levels of the Su g protein in these strains. Yeast cells harboring the Su

FIG. 2. Chemical cross-linking of Su g-Su e heterodimers.
A and B, upper panel, mitochondria (100 g of protein) isolated from the wild type (WT) and the ⌬su g null mutant strains expressing the Su gY112C (Y112C), or Su gY112C HA (Y112C HA ) derivatives were treated with the chemical cross-linker DTNB (ϩ) or were mock-treated and received Me 2 SO (Ϫ), as described under "Experimental Procedures." After crosslinking and quenching of free DTNB, mitochondria were re-isolated by centrifugation, divided in half, and subjected to non-reducing SDS-PAGE and Western blotting. One of the resulting blots (A) was used to identify Su g and the Su g HA derivatives and their respective crosslinked adducts by decoration with Su g antisera or with HA-specific antisera, as indicated. Adducts with electrophoretic mobilities corresponding to Su e-Su g (e-g), Su e-Su g HA (e-g HA ), and Su g HA -Su g HA (g HA -g HA ) are indicated. Su e and cross-linked adducts were identified after decoration of the second blot with Su e-specific antisera (B, upper panel). Adducts with electrophoretic mobilities corresponding to Su e-Su e (e-e), Su e-Su g (e-g), Su e-Su g HA (e-g HA ), and Su e-7 kDa (e-7kDa) are indicated. A protein that cross-reacts with the Su e antiserum (present also in the ⌬su e null mutant (results not shown)) is indicated (*). B, lower panel, mitochondria (100 g of protein) isolated from the wild-type (WT) and the ⌬su g null mutant strains expressing the Su gV110C HA (V110C HA ) or Su gY112C HA (Y112C HA ) derivatives were subjected to cross-linking with DTNB and further processed as described above in (B, upper panel). The resulting blot was decorated with Su e-specific antisera. The mobilities of the protein standards (kDa) are indicated. gG105V HA and Su gG105L HA derivatives also displayed an increased formation of rho 0 /rho Ϫ cell formation, suggesting this aspect of Su g function was partially compromised by the mutation in the residue Gly-105. The observed instability of mtDNA in the Gly-105 mutants, however, was not as significant as with those harboring the Gly-101 mutants or the Su g103A104 HA derivative.
Role of the Conserved GXXXG Motif for the Interaction of Su g with Su e-To analyze the significance of the GXXXG motif of Su g for its interaction with Su e or its own possible homodimerization, a double mutant of Su g was created whereby the Su gG105V HA and Su gG105L HA mutants were cloned in context of the Tyr-112 to Cys mutation, i.e. resulting in the creation of the Su gG105V,Y112C HA and Su gG105L,Y112C HA derivatives, which were then expressed in the yeast null su g mutant, ⌬su g (Fig. 3A).
Chemical cross-linking using DTNB was pursued to analyze whether the ability of Su g to interact with Su e (or Su g) was compromised by the introduction of these mutations into the conserved GXXXG motif (Fig. 4). The level of Su g-Su e-cross-linked adducts formed in both of the Gly-105 mutants was similar to that observed for the Su gY112C HA derivative (Fig. 4, upper panel). Furthermore, the level of cross-linked adduct corresponding to the putative Su g-Su g dimer was increased in the both of the Gly-105 mutants relative to the Su gY112C HA derivative, where the GXXXG motif was intact (Fig. 4, upper panel).
Analysis of a parallel Western blot which was immune-decorated with Su e-specific antisera provided independent support for the conclusion that the Su g-Su e interaction was not affected by the mutation of Gly-105 in Su g (Fig. 4, lower panel). The efficiency of formation of the Su g-Su e cross-linked adducts, as evidence by the production of the 24-kDa adduct immune-reactive to Su e antisera, was not adversely affected when the Su g protein harbored a mutation in the conserved GXXXG motif. Interestingly the formation of the Su e-Su e homodimer, as evidence by the formation of the 22-kDa adduct upon incubation with DTNB, was more pronounced in both of the Gly-105-mutated Su g proteins. A similar increase in the levels of the Su e-Su e homodimer was also observed in the absence of the Su g protein, i.e. in the ⌬su g mitochondria (Fig. 4, lower panel). Significant cross-linking of Su e to a smaller protein of ϳ7 kDa, as evidenced by the generation of a 18-kDa Su e-reactive adduct, was observed in the absence of Su g (i.e. in the ⌬su g mitochondria) and also in the ⌬su g mitochondria bearing either of the Gly-105-mutated forms of Su g. Formation of this 18-kDa Su e-adduct was also observed in the wild-type and Su gY112C HA mitochondria, albeit at significantly reduced levels (Fig. 4, lower panel, see also Fig.  2B). Although the identity of this Su e-interacting protein is currently unknown, the greatly enhanced formation of the Su e-18-kDa adduct indicates that the molecular environment of Su e is greatly altered in the absence of Su g or in the presence of the Gly-105-mutated derivatives of Su g.
Conserved GXXXG Motif of Su g and ATP Synthase Dimerization-Using the mild detergent digitonin, the ATP synthase complex can be solubilized from wild-type mitochondria membranes and resolved on a native electrophoresis gel as a dimeric complex (ϳ1000 kDa) (6). The presence of Su g and Su e is required for the formation of detergent-stable ATP synthase dimers, as only monomeric complexes are observed after detergent solubilization of mitochondria from the ⌬su g and ⌬su e strains (6, 9, 15). We, therefore, analyzed the dimeric state of the ATP synthase complex from mitochondria harboring the mutations in the conserved Gly-105 residue of the Su g protein.

FIG. 3. Expression of Su g derivatives containing mutations in GXXXG motif.
A and B, mitochondria were isolated from the ⌬su g null mutant strain harboring the wild-type HA-tagged Su g (Su g HA ) or the GXXXG mutated Su g derivatives as indicated and were analyzed by SDS-PAGE and Western blotting, essentially as described in Fig. 1A. G105V HA , Su gG105V HA ; G105L HA , Su gG105L HA ; G105V,Y112C HA , Su gG105V,Y112C HA ; G105L,Y112C HA , Su gG105L,Y112C HA ; G101L HA , Su gG101L HA ; G101V HA, Su gG101V HA ; 103A104 HA , Su g103A104 HA . C, the ⌬su g strain, ⌬su g bearing mutated Su g derivatives and wild-type control strains as indicated were maintained on YP-glycerol media, transferred to YP-galactose media, and grown overnight. The number of cells that became rho 0 /rho Ϫ (i.e. exhibited loss of mt-DNA) was determined, as described under "Experimental Procedures." For each strain type the number of cells that exhibited loss of mtDNA was expressed as a % of total cells analyzed (% rho 0 /rho Ϫ ).
FIG. 4. Intact GXXXG motif in Su g is not required for the formation Su g-Su e heterodimers. Chemical cross-linking with DTNB was performed essentially as described in Fig. 2. The mitochondria used for the analysis were isolated from ⌬su g null strain or the ⌬su g null strain expressing the Su gHA, Su gY112C HA (Y112C HA ), Su gG105V,Y112C HA (G105V,Y112C HA ), and Su gG105L,Y112C HA (G105L,Y112C HA ) derivatives. After cross-linking, samples were divided, analyzed on parallel SDS-PAGE gels, and Western-blotted. Immune-decoration of resulting blots with an HA-specific antiserum (upper panel) or a Su e-specific antiserum (lower panel) was performed. Cross-linked adducts with electrophoretic mobilities corresponding to Su e-Su g HA (e-g HA ), Su g HA-Su g HA ( g HA -g HA ), Su e-Su e (e-e), and Su e to an unknown ϳ7-kDa protein (e-7kDa) are indicated. A protein that cross-reacts with the Su e antiserum, is indicated (*). The mobilities of the protein standards (kDa) are indicated.
Isolated mitochondria were treated with increasing concentrations of the mild detergent digitonin, and the solubilized proteins were analyzed by clear native-PAGE. The assembly state of the ATP synthase complex was analyzed after Western blotting and immune-decoration with antibodies raised against the purified F 1 -sector (Fig. 5, upper panel). In the absence of Su g, i.e. in the ⌬su g mitochondria, the ATP synthase complex was solubilized as the monomeric size complex, indicating that the presence of Su g is required for the formation of detergentstable ATP synthase dimers (Fig. 5, upper panel). The expression of the HA-tagged wild-type Su g protein, Su g HA in the ⌬su g strain fully restored the ability of the ATP synthase to form stable dimeric complexes (Fig. 5, upper panel). We conclude, therefore, that the addition of the HA tag to the C terminus of Su g did not adversely affect the function of Su g in this respect. Clear native-PAGE analysis of the detergentsolubilized ATP synthase from the mitochondria harboring the Su gG105V HA or Su gG105L HA derivatives indicated the presence of both dimeric and monomeric forms of the enzyme (Fig. 5, lower panel). The ATP synthase dimers formed in the presence of Gly-105-mutated Su g derivatives were not as stable as those supported by the wild-type Su g HA , as evidenced by the increased presence of monomeric ATP synthase particularly at the higher detergent concentrations. Moreover, ATP synthase monomers were observed in the G105L mutant relative to the G105V mutant, and this may reflect the reduction in the steady state levels of the Su gG105L HA derivative relative to the Su gG105V HA protein. Analysis of the assembly state of the ATP synthase in the mitochondria harboring the Su gG101L HA , Su gG101V HA , or Su g103A104 HA derivatives indicated the presence of the monomeric form of the enzyme (results not shown). The significant reduction in the levels of dimeric complexes in this case simply may reflect the strongly reduced steady state levels of the Gly-101-mutated or Su g103A104 HA derivatives relative to wild-type or the Gly-105-mutated derivatives. DISCUSSION To gain more insight into the importance and the role(s) of Su g of the yeast ATP synthase, we used site-directed mutagenesis and a sulfhydryl-specific cross-linking approach to map the molecular environment of Su g within the F 0 -sector. Using the zero-length spacer arm cross-linker DTNB, we provide evidence here for the close proximity of the Su g and Su e proteins in the mitochondrial inner membrane. Moreover, our data sup-port that Su g and Su e are positioned in the inner membrane in such a manner that their membrane anchor domains are physically close together. Specifically, we demonstrate that a cysteine positioned at residue 112 of Su g can be cross-linked to Cys-27 of Su e using DTNB. These residues are located in the intermembrane space and are in close proximity (ϳ7 and 6 residues, respectively) to the membrane anchor regions of their respective proteins.
Our conclusion that the membrane anchor regions of Su g and Su e are physically close to each other is consistent with recent observations concerning the stabilization of Su g by Su e. The presence of Su e is required for the stability of Su g (6,15,16), and recent findings indicate that the membrane anchor region of Su e and/or the hydrophilic residues immediately C-terminal to it, are critical for the stability of Su g (15). Moreover, mutation in the conserved GXXXG motif in the membrane anchor region of Su e had a pronounced effect on the stability of Su g (16). This together with our findings in this work confirm that Su g and Su e and in particular their membrane anchor regions are in close physical proximity to each other within the F 0 -sector of the ATP synthase complex.
The GXXXG motif in the transmembrane segment of Su g is highly conserved from lower eukaryotes, such as yeast, to higher eukaryotes, e.g. humans (20), indicating the potential importance of this motif for the function of Su g. GXXXG motifs can play essential roles in the homo-and heterodimerization of a number of membrane proteins, as they form the basis for helix-helix interactions, with the homodimerization of glycophorin A being one of the best-characterized model proteins (21)(22)(23)(24). Given that there are ϳ3.6 residues per ␣-helical turn, the conserved Gly residues of the motif would be arranged on the same face of the helix, and the number of three-spacer residues separating them is critical to preserve this arrangement. The small nature of the side chain of Gly is compatible with the formation of a close helix-helix association (21)(22)(23)(24). Exchange of Gly residues for hydrophobic ones with bulkier side chains, such as Ala, Val, Leu, or Ile, interfere with the packing interface of neighboring helixes and consequently have been shown to prevent the dimerization of model membrane proteins such as glycophorin A and integrin ␣ IIb (23,31,32). Similarly, the introduction of an extra residue between the Gly residues of this motif prevent helix-helix association of glycophorin A, as the conserved Gly residues are no longer be arranged on the same face of the ␣-helix (21). Our analysis here demonstrates that the conserved 101 GXXXG 105 motif in Su g plays an important role in the function and stability of Su g in the mitochondria. The disruption of the GXXXG motif by the insertion of an additional Ala residue after amino acid 103 had severe effects on the stability of Su g, as evidenced by the reduced steady state levels of the Su g103A104 HA derivative. Similarly, our data show that the mutation of the conserved Gly residues affects both the stability and molecular organization of Su g within the ATP synthase complex. In both the case of Gly-101 and Gly-105, substituting the Gly residue for a Leu residue was observed to be more deleterious than substituting with a Val residue. Leu has a bulkier side chain than Val and, therefore, may adversely affect the helix-packing capacity of Su g transmembrane region to a greater extent. Of the two conserved Gly residues (Gly-101 and Gly-105), our data support a more critical role of residue Gly-101, which relative to Gly-105, is located more toward the center of the transmembrane helix of Su g. Substitution of Gly-101 by a bulkier residue, Val, and in particular Leu, was deleterious for the stability of Su g, whereas the mutation of Gly-105 did not have such a pronounced adverse affect. Although the steady state levels of Su g appeared FIG. 5. Clear native gel analysis of the dimeric state of the F 1 F o -ATP synthase complexes. Mitochondria were isolated from the ⌬su g strain or the ⌬su g strain harboring the Su g HA , Su gG105V HA , or Su gG105L HA derivatives, as indicated. The ⌬su g null strain had been grown on YP-glycerol medium, whereas the other strains were grown on YP-galactose. Isolated mitochondria (200 g of protein) were solubilized with digitonin (dig) at the concentrations indicated and, after a clarifying centrifugation, were directly analyzed by clear-native-PAGE, as described under "Experimental Procedures." The native gels were then Western-blotted, and immune-decoration was performed using a yeast F 1 -specific antisera. The positions of the ATP synthase dimer (V dim ) and monomer (V mon ) are indicated. not to be as significantly reduced by mutation of the Gly-105 residue, we observed that the function of Su g was clearly compromised in Su gG105V HA and Su gG105L HA derivatives. Specifically, we demonstrated that the Su g derivatives bearing a mutation in residue Gly-105 were less able to support the stable dimerization of the ATP synthase complex and that cells harboring this Su g mutant displayed a reduced capacity to prevent loss of the mtDNA when compared with the wild-type Su g HA protein. Furthermore, our cross-linking data indicate that mutation of Gly-105 to either a Val or Leu residue had a pronounced affect on the ability of Su g to influence the organization of Su e within the ATP synthase complex. In the absence of Su g, i.e. in the ⌬su g mitochondria, we report here that Su e displays an increased ability to form homodimers and also to become cross-linked to an unknown protein of ϳ7 kDa. Mitochondria harboring Su gG105V,Y112C HA or Su gG105L,Y112C HA derivatives both displayed a similar alteration in the molecular organization of Su e, as was observed in the ⌬su g mitochondria. Thus, although the mutation of Gly-105 of Su g did not have such a pronounced affect on the stability of Su g as the Gly-101 mutation, it clearly compromised the function of Su g, in particular its ability to stabilize the mtDNA and to modulate the environment of Su e. All of these observations together allow us to conclude that the conserved GXXXG motif plays a critical role in the stability of Su g and its ability to modulate the molecular organization of components of the ATP synthase in particular Su e. We propose, therefore, that the GXXXG motif is required to support the interaction of Su g with another membrane protein. As discussed below, one possible candidate for this interacting protein would be Su e.
Su e also has a conserved GXXXG motif in its membrane anchor segment, and the presence of Su e is required for the stability of Su g (15,16). Velours and co-workers (16) have previously shown that mutation of the conserved Gly residues of the GXXXG motif in Su e or the disruption of this motif by the insertion of an extra hydrophobic residue results in the inability of Su e to ensure the stabilization of Su g. These comments, together with our findings reported here, allow us to predict that the transmembrane region of Su g, and in particular the conserved GXXXG motif, could be the foundation for the observed Su g-Su e interaction. In turn, this interaction with Su e would ensure the stability of Su g in the membrane. Contrary to this notion, however, we demonstrate here that an intact GXXXG motif in Su g is not required to support an interaction between Su g and Su e. The exchange of Gly-105 for Val or Leu residues did not have a measurable effect on the ability of Su g to become cross-linked to Su e despite the fact that this mutation compromises a number of functions of Su g as outlined above. If the GXXXG motif of Su g operates as a dimerization motif and in a similar manner as in glycophorin A or integrin ␣ IIb (23,31,32), mutation of Gly-105 to Leu or Val would have disrupted helix-helix packing and, hence, the ability of Su g to interact with Su e. In addition, our data would support that the GXXXG motif is not the basis for the possible Su g-Su g homo-dimerization we observed. An increase in the level of Su g-Su g interaction was observed in the case of the Su g derivative bearing the Gly-105 mutations.
Hence, we conclude that an intact GXXXG motif in Su g is not essential for the observed Su g-Su e interaction or the putative Su g-Su g homodimerization. Although it remains possible that the GXXXG motif is somehow involved in the interaction of the transmembrane segments of Su g with Su e, it does not appear act as a canonical GXXXG helix-helix interaction motif, as has been described for glycophorin A. We conclude, therefore, that the GXXXG motif of Su g may not be the basis (or sole basis) for the Su g-Su e interaction. While interacting together in the membrane, it is possible that Su g and Su e may also interact with yet another protein and in a manner that requires the GXXXG motifs of Su g and Su e. In the absence of Su g (or in mitochondria bearing the Gly-105-mutated Su g derivatives), our cross-linking data that indicate a closer association between Su e and an unknown protein of ϳ7 kDa is favored. The identity of this 7-kDa protein is currently unknown. The mass of this protein is comparable with that of two other known F 0 -subunits, Atp8 and Atp9. Because Atp8 does not contain a Cys residue, we can exclude it as a candidate, as the cross-linking to Su e was observed with the sulfhydryl-specific cross-linking agent DTNB. In addition, immune-decoration of Western blots of cross-linking experiments with Atp9 antisera have not provided support for the possibility that the 7-kDa candidate may represent Atp9. We, therefore, are adopting other approaches to purify and identify this 7-kDa interacting protein. We propose by interacting together (and possibly with the 7-kDa protein also), Su g and Su e exert functions that are distinct from their ability to stabilize the ATP synthase dimers and by doing so ensure the maintenance of a tightly coupled ATP synthase, as suggested by the reduced growth and the loss of mtDNA phenotypes observed with the su e and su g null mutant strains. Understanding the molecular organization of Su g and Su e within the F 0 -sector represents an essential step toward fully appreciating the significance of these conserved proteins.