Structure-Function Studies of Adenine Nucleotide Transport in Mitochondria OF DISTINCT AACl AND AAC2 PROTEINS IN YEAST*

AAC1 and AAC2 genes in yeast each encode functional ADP/ATP carrier (AAC) proteins of the mitochondrial inner membrane. In the present study, mitochondria harboring distinct AAC proteins and the pet9 Arg96 to HIS mutant (Lawson, J., Gawaz, M., Klingenberg, M., and Douglas, M. G. (1990) J. Biol. Chem. 265, 14195-14201) protein have been characterized. In addition, properties of the different AAC proteins have been defined following reconstitution into proteoliposomes. Deletion of AAC2 but not AAC1 causes a major reduction in the mitochondrial cytochrome content and respiration, and this level remains low even when the level of AAC1 protein is increased to 20% that of the AAC2 gene product. In reconstitution studies, the rate of nucleotide transport by isolated AAC1 protein is approximately 40% that of the AAC2 protein. Thus, the lack of mitochondrial-dependent growth supported by the AAC1 gene product alone may be due to the combination of low abundance and reduced activity. Surprisingly, analysis of the Arg96 to His mutant protein revealed binding and transport activities similar to the functional AAC1 and AAC2 gene products. These observations are discussed in relation to a molecular analysis of this highly conserved small transporter and its function in conjunction with other proteins in the mitochondrial membrane.


AACl
and AAC2 genes in yeast each encode functional ADP/ATP carrier (AAC) proteins of the mitochondrial inner membrane.
In addition, properties of the different AAC proteins have been defined following reconstitution into proteoliposomes.
Deletion of AAC2 but not AACl causes a major reduction in the mitochondrial cytochrome content and respiration, and this level remains low even when the level of AACl protein is increased to 20% that of the AAC2 gene product.
In reconstitution studies, the rate of nucleotide transport by isolated AACl protein is approximately 40% that of the AAC2 protein.
Thus, the lack of mitochondrial-dependent growth supported by the AACl gene product alone may be due to the combination of low abundance and reduced activity.
Surprisingly, analysis of the Argss to His mutant protein revealed binding and transport activities similar to the functional AACl and AAC2 gene products.
These observations are discussed in relation to a molecular analysis of this highly conserved small transporter and its function in conjunction with other proteins in the mitochondrial membrane.
The ADP/ATP carrier protein of mitochondria is an abundant and relatively simple membrane transporter of the mitochondrial inner membrane (Klingenberg, 1985). It is a relatively small protein of approximately 300 amino acids which spans either mitochondrial inner or artificial membranes to promote energy linked exchange of adenine nucleotides. As a simple highly conserved transmembrane transporter it is ideally suited to define the molecular features of nucleotide transport and its control. The AAC' proteins recently defined in yeast provide the basis for a detailed structure function analysis of different AAC molecules in both native and artificial membranes (Adrian et aZ., 1986;Lawson and Douglas, 1988;Lawson et al., 1990).
The genetic complementation of mutants blocked in adenine nucleotide function (Lawson and Douglas, 1988;Lawson et al., 1990) and the level of AAC protein present in mitochondrial membranes indicate that the AAC2 gene product (or the yeast pet9 gene product) is essentially the only AAC translation product present in mitochondrial membranes grown under derepressing conditions. The reasons for the presence of a silent AACl gene capable of encoding a functional translocator has not been defined at present. To gain some insight into the relative contributions of each protein to adenine nucleotide transport, we have examined the biochemical and biophysical properties of the individual AAC proteins including the Argg6 to His, pet9, protein ifi mitochondrial membranes as well as reconstituted into artificial bilayers. Sufficient levels of the AACl protein are expressed from a yeast multicopy vector for preparation of this translocator isoform. In addition, we have characterized the consequences of chromosomal deletion of both AAC genes on the resulting oxidative capacities and cytochrome content of mitochondria. Their absence from the membrane dramatically reduces the extent to which biogenesis of the respiratory chain will occur. These studies confirm that the individual translocators including the Arg96 to His pet9 exhibit little difference in nucleotide binding and less than a 2-fold difference in their translocation properties. Thus, the failure of the pet9 mutant to grow on a nonfermentable carbon source cannot be explained solely on the basis of a defect in binding and transport of adenine nucleotides. were grown on galactose (Sherman et al., 1979). Yeast cells were obtained bv overnight growth in a-liter flasks (New Brunswick) at 30 "C with vigorous shaking and maximum aeration. Cells were harvested at mid-logarithmic phase (A57sn,,, = 5.5). Preparation of Mitochondria-Protoplasts were formed by enzymatic digestion of the cell wall (O'Malley et al., 1982, Daum et al., 1982. For this purpose yeast cells were suspended in 0.1 M Tris, 10 mM dithiothreitol, pH 9.4, for 10 min at 32 "C (0.5 g wet weight/ml), centrifuged 5 min at 3000 x g, washed twice with 1.2 M sorbitol, and resuspended at 0.15 g wet weight/ml in 1.2 M sorbitol, 20 mM KHZPO,, pH 7.4, (0.15 g wet weight/ml under addition of zymolyase 20,000 (1 mg/l g wet weight). The conversion to protoplasts occurred within 60-90 min at 35 "C with gentle shaking.
Protoplasts were harvested by centrifugation at 3000 x g for 5 min and washed twice with 1.2 M sorbitol.
For lysis protoplasts were suspended at 4 "C in a buffer containing 0.6 M mannitol, 10 mM Tris, pH 7.4, 0.1% bovine serum albumin, 1 mM PMSF at 0.15 g wet weight/ml with a "Dounce" homogenizer (12 strokes) and then incubated for 15 min under moderate stirring. After centrifugation at 1000 X g for 10 min the supernatant was recentrifuged (1000 x g, 10 min). The mitochondrial fraction was obtained by centrifugation of the resulting supernatant at 9700 X g for 10 min, resuspension in 0.5 M mannitol, 10 mM Tris, pH 7.4, and stirred for 5 min. After centrifugation for 10 min at 1000 X g, the final mitochondrial preparation was obtained by recentrifugation of the supernatant at 9700 X g for 10 min. If necessary mitochondria were loaded with carboxyatractylate (CAT) or atractylate (ATR) in the presence of 50 @M ADP and 2 mM MgClz before storage at -70 "C in liquid nitrogen.
Respiration-Oxygen consumption was determined for cells and mitochondria with the platinum electrode at 25 "C (Kramer and Klingenberg, 1982 et al. (1951) in the presence of 1% SDS (Schagger and Von Jagow, 1987) and by a modified Biuret method (Lowry and Zitomer, 1984). CNBr cleavage, immunoblot, and preparation of antisera were performed as published earlier (Knirsch et al., 1989). For immunoblot and Coomassie Blue-stained SDS-polyacrylamide gel electrophoresis, proteins were run on 12.5% Laemmli gels with 5% stacking gel (Laemmli, 1970 (Silverman 1987, Daum et al., 1982. All other chemicals were of analytical grade.

RESULTS
Yeast Strains-The yeast strains were constructed for growth and characterization of the individual AACl andAAC2 gene products (Lawson et al., 1990). The strains used are listed in Table I. The mutants were constructed by deletion of either or both the AACl and AAC2 genes. Into the deletion mutants either AACl or AAC2 genes were reintroduced on plasmids. By using the centromere plasmid pSEYc63, AAC2 was expressed in the LIAAC~AAAC~ host JLY1053 at one to two copies/cell. Using the 2 micron plasmid pSEY8, AACl was expressed at approximately 20-50 copies/cell (Table I). This was necessary since AACl is not expressed at a detectable level as a unit copy gene (Lawson andDouglas, 1988, Lawson et al., 1990). Only from the 2 micron plasmid YEp-AACI) could the AACl product be isolated in amounts sufficient for reconstitution.
Growth Characteristics of the Yeast Cells-Yeast mutants which lack the AACl gene (aacl::LEU2) but contain AAC2 either in the chromosome or on a CEN plasmid could all be grown under the same conditions using galactose as a carbon source. AAC2 deletion mutants were unable to grow on glycerol but grow at rates similar to wild type on galactosecontaining medium (Table II). Although slower on galactose, the AAAClAAAC2 double deletion mutant grew to cell density comparable with the AAAC2 mutant alone for biochemical studies.
Respiration and Cytochrome Content of Mitochondria-Mitochondria were isolated from the various mutant strains after growth to log phase on galactose (see "Experimental Procedures"). The cytochrome content noted in isolated mitochondria reflected that measured in whole cells (not shown). The cytochrome au3 content of AAAC2 mitochondria was reduced to about 40% of wild type mitochondria (Table III). The content of cytochromes aa and c was lowest in the aAAClAAAC2 double deletion mutant but measurable when compared with mitochondria from a rho-strain lacking any cytochrome aas. These cytochrome contents also reflect the relative respiration rates for each form. The level of cytochrome b was also lowest in the AAACfAAAC2 mutant, (40 mmol/g) but increased approximately 3-and 12-fold in the (AAAC2 host and UC1 host, respectively (Table III)). The respiratory capacity of mitochondria was measured with added NADH, succinate, and malate plus glutamate (Table IV). In the AAAC2 mutant the respiratory activity was decreased about 7-fold for NADH and 5-fold for succinate (Table IV) compared with wild type yeast. This was the same pattern observed in whole cells (not shown) for these mutants. The AAAClAAAC2 double mutant was further reduced but in the same range as the oxidative activity of AAAC2 or pet9 alone with all hydrogen donors tested (Table IV). On the other hand, the respiration of all mitochondria harboring a functional AAC2 gene product differed little from that of wild type. Thus, mitochondrial respiration rates with different substrates were consistent with the rates measured in whole cells (not shown) and reflect a large reduction in cytochrome content due to absence of AAC2 protein. This was confirmed by measurement of difference spectra on mitochondrial suspensions (Kramer and Klingenberg, 1982). The content of cytochrome aa3 in the MAC2 mutants decreased up to 5-fold and cytochrome c decreased d-fold from the wild type mitochondria (Table III). This reduced cytochrome aa content results in an elevated cytochrome turnover for the UAClAAAC2 mitochondria (Table IV). We have also observed that the pet9 strain, although fully competent for AACl function and blocked for AAC2 function has only one-fifth the cytochrome content of the wild type mitochondria (Table  III). Thus, it would appear that assembly of the respiratory chain with its complete complement of cytochromes is limited in some manner by the activity of the AACl protein.
It is noteworthy that expression of the AACl protein in amounts sufficient to support growth on glycerol (YEpAACI) only marginally increased the respiratory activity and cytochrome content of the AAAClAAAC2 and AAAC2 strains. The loss of cytochromes and of respiratory activity is the result of the loss of the AAC2 gene product which is not compensated for by over expression of the AACl protein. It would, therefore, appear that the AAC2 protein may serve a structural as well as a nucleotide transport role in the membrane.
Content of AAC in Mitochondria-The carrier protein content in the various yeast mitochondria was determined by measuring the binding of [3H]CAT. The binding of [3H]CAT was determined as a function of the concentration of added CAT in order to evaluate the saturation and the binding constants.
The binding values are summarized in    reflect the reduction in cytochrome au3 content upon deletion of the AAC2 gene (Table III) product from the membrane and the inability of the AACl protein alone to restore the cytochrome content.
The corrected binding values were plotted in mass action graphs and evaluated for the dissociation constant of [3H] CAT and the maximum binding. The KD for the AACl product (3.7 X lo-* M) is higher but not significantly different from that of the wild type (2.1 x lo-' M). The Ko of the pet9 mutant remained wild type indicating that the Ar$'j to His mutation has no effect on [3H]CAT binding. In wild type mitochondria the measured KD and site number is determined predominately by that of the AAC2 gene product. Isolation ofAAC1 and AAC2-To characterize the individual AAC proteins in the absence of other inner membrane proteins, the AACl and AAC2 proteins were purified for additional study. The isolation of AACl and AAC2 from isolated mitochondria followed the procedure recently established for the wild type AAC (Knirsch et al., 1989). The Isolation of the AAC2 gene product followed the same procedure as its isolation from wild type. Isolation of AACl from a host containing the YEpAACl plasmid required solubilization with Triton X-100 in the presence of Na2S04 followed by HTS chromatography (Fig. 1). A band which comigrates with AACl (AAAC2, YEpAACl) is noted in the partially purified mitochondrial membrane extract prepared by the HTS passthrough.
If this protein is AAClp, then the content of this isoform is higher than indicated by antibody studies (Lawson et al., 1990) and more comparable with the content revealed by carboxyatractylate binding.
We also note the presence of a band comigrating with the AAClp in an extract prepared from the AAACl.
This most likely represents a proteolytic fragment of the AAC2p, since the extract of the double deletion lacks this band (Fig. 1, lane 1). The level of AAC proteins in the pet9 host appear comparable with wild type. The separation of co-purifying porin protein from the AACl or AAC2 was achieved by chromatography of the HTS eluate on Sephacryl S-300 (see "Experimental Procedures"). As shown in Fig. 2 the purified AACl migrates below the AAC2, slightly above porin. For AAC2 and AACl the observed difference in the migration rate corresponded to a molecular weight difference of KD 1.5. This was slightly higher than the difference of KD 0.7 calculated from the DNA sequence (Lawson and Douglas, 1988).
Cyanogen bromide cleavage of AAC proteins purified from wild type yeast mitochondria compared to the cleavage pattern of purified AACl and AAC2 proteins confirmed that the AAC2 gene product is expressed exclusively under normal growth conditions (not shown). The cleavage products of AAC proteins purified from wild type mitochondria correspond HTS -Passthrough Mitochondria SDS-polyacrylamide gel electrophoresis of detergent-extracted mitochondrial proteins (right) and corresponding hydroxylapatite pass-throughs (left) of wild type, UACP LAACl,YEpAACl, and L4ACI yeast strains. 12.5% Laemmli separating gel was loaded with (80 rg) mitochondrial protein and (15 pg) HTS protein. 1 mg of CAT-loaded mitochondria were solubilized in 4% SDS, 3% glycerol, 10 mM Tris-Cl, pH 6.8, in the presence of 1 mM PMSF. HTS proteins were obtained as described under "Experimental Procedures" using C12E8 for solubilization.
with the fragment lengths predicted from the AAC2 sequence. This agrees with all data published to date that AAC2 is the primary carrier found in aerobically grown wild type yeast (Lawson and Douglas, 1988;Lawson et al., 1990). The level of ["HICAT binding noted for AACl protein in the AAAC2 host reflects increased AACl protein due in some manner to the loss of the AAC2 gene product. This observation is currently under investigation.
In the presence of AAC2 the AACl product is not detectable by either antibody or CNBr cleavage pattern.
Reconstitution of AACl and AAC2 Gene Products-To further examine the kinetic properties of the AACI, AAC2, and pet9 gene products, the proteins were isolated and reconstituted into liposomes. For reconstitution studies the carrier proteins were isolated unliganded to the inhibitor CAT (see "Experimental Procedures").
The KM and V,,,,, values for the AAC2 protein alone and the AAC protein from the wild type strain were the same. We observed that unliganded AACl carrier protein is more unstable than AAC2 carrier under these reconstitution conditions. After surveying several detergents and conditions it was noted that the non-ionic detergent Ci2ER in the presence of a high concentration of ammonium acetate was useful to solubilize the AACl carrier in reasonable yield. These were the conditions used previously for isolating the aspartate-glutamate carrier from bovine heart mitochondria (Kramer and Heberger, 1986) and also rendered a more Isolation from mpAAC1 and AAC2 strains. Purification followed the procedure described under "Experimental Procedures." CAT-loaded mitochondria were solubilized with Triton X-100 (ratio of detergent/protein = 2.5) in a standard buffer (Kramer and Klingenberg, 1982). After HTS chromatography the eluate was concentrated by pressure dialysis and then applied to Sephacryl S-300. which seoarated the persistent porin. eluted in the HTS passthroughs (Fig: 3). .
advantage that less porin was extracted (see Fig. 1).
In reconstitution studies using the freeze-thaw method, it was necessary to minimize the presence of residual detergent in the proteoliposomes and partial inactivation of the reconstituted protein by sonication. To overcome these difficulties a modification of the method developed for the reconstitution of the uncoupling protein from brown adipose mitochondria was utilized. This led to a reproducibly high exchange activity after reconstitution of both AACl and AAC2 proteins (Klingenberg and Winkler, 1986). In this procedure phospholipids dispersed in Ci2E8 were added to the solubilized carrier protein followed by the addition of Amberlite XAD-2 to gradually remove the detergent (see "Experimental Procedures"). The kinetics of transport were obtained by measuring the uptake of [14C]ADP and ["'CIATP at four time intervals at each of four different ADP or ATP concentrations.
The initial rates were plotted in v versus v/s graphs for the evaluation of V,,, and KM (Fig. 3). In order to directly compare reconstitution among the different carriers, the molar exchange rates were normalized to the [3H]CAT binding capacity. We noted earlier that ["HICAT binding was essentially the same for both AACl and AAC2 proteins. Therefore the ["HICAT binding of the HTS protein fraction was determined for each reconstitution experiment (see "Experimental Procedures"). The results of the exchange measurements obtained with the reconstituted AAC from the various strains are summarized in Table VI. The rates are expressed according to protein content as well as to the ["HICAT binding capacity. The latter more precisely reflects the carrier content. The KM and V,,,,, values determined were virtually the same whether AAC protein was isolated from wild type yeast or from strains expressing the plasmid encoded or integrated AAC2 by itself. Sur-prisingiy the pet9 protein exhibited only a 2-fold reduction in V,,,., compared with wild type. On the other hand, the AACl carrier exhibited a V,,,,, value which was less than that of the pet9 protein. There was only a small variation in the KM values determined for the AACl, AACZ, and pet9 proteins. Thus, the only major distinction between the activities of the isolated AACl, AAC2, and pet9 proteins is their turnover number.
For estimating the exchange capacity in mitochondria harboring different carrier proteins, the molecular activity of the AAC proteins was multiplied with the number of carrier sites determined in the mitochondria. As shown in Table VI the exchange activity is reduced to approximately 25% in the mitochondria from yeast expressing YEpAACr or pet9 for both the ADP and ATP exchange. It is noteworthy that the respiratory capacity of mitochondria is markedly influenced by the type of AAC protein which is present in the membrane. This is expressed as the turnover number normalized to the cytochrome aa content (Table VI). The values where wild type mitochondria are compared with mitochondria containing AAC protein exclusively from YCpAAC2 or YEpAACl are similar, supporting the previous observation that the decrease in AAC activity is accompanied by a correspondingly larger decrease in cytochrome content.

DISCUSSION
The two AAC proteins encoded by AACl and AAC2 in Saccharomyces cereukiue differ much more from each other than those of the mammalian AAC. When the yeast AAC2 andAAC1 are compared, the difference is about 25% (Lawson and Douglas, 1988). With the additional 10 residues at the N terminus in AAC2, the difference is 27%. The rationale that the isoforms may have adapted to tissue specific metabolic requirements is not evident. However, it is clear that both isoforms will support growth of yeast on a nonfermentable carbon source. The possibility that the isoforms may be expressed in yeast under different conditions has not been observed in preliminary studies. At all vegetative stages of the wild yeast only the AAC2 appears to be expressed, whereas the AACl gene is essentially silent.3 [3H]CAT binding data indicate that in the absence of AAC2, AACl protein is present to about IO-20% of the AAC2 level in wild type. The mechanism for this increase in AACl expression in the absence of AAC2 must await further study. The level of AACl protein in mitochondrial membranes increases to 45% that of wild type when the plasmid YEpAACI, 20-50 copies/cell, is the source of AACl gene product. The molecular basis for this reduced AACl content is currently under study.
In animals it has not yet been possible to examine functional differences of the isoforms, nor to isolate these isoforms in a functional state. The present characterization of the AACl and AAC2 proteins in yeast is the first example where the two isoforms could be compared and their kinetic and binding properties differentiated.   Wild type   1150  840  600  450  28  6  480  350  1910  1590  AAc2  1100  850  620  450  25  I  460  345  2190  1640  YCpAAC2  1040  820  630  500  23  6  430  340  3070  2430   YEpAACl   480  360  230  190  20  10  120  90  2400  1800  pet9  690  410  350  210  31  12  125  75  3570  2130 protein be prepared for the characterization. The exchange activity of the AACl and AACZ could not be determined with sufficient reliability in isolated mitochondria from the various yeast strains. The AAACZ host produced only small and fragile mitochondria with low endogenous nucleotide content. Therefore the exchange rates were determined after isolation of carriers and reconstitution into proteoliposomes.
Exchange rates were quantitated in relation to the number of CAT binding sites present for both AACl and AACZ proteins. The AACl protein both in mitochondria and in the liposome has only about 40% of the exchange activity of the AACZ protein.
There was, however, no significant difference for the KM for ADP/ATP between both AAC isoforms as well as pet9.
The AACl protein expressed from the YEpAACl plasmid is present in the mitochondrial membrane to 45% the level of AACZ protein in the wild type (Lawson et al., 1990). At this level of AACl protein in the mitochondria one can calculate that the exchange capacity of the mitochondria is less than 25% that of the wild type. On the other hand AACZ in single copy yields mitochondria with essentially the same activity as the wild type. We conclude from these studies that the only distinction between AACl and AACZ proteins is from their relative exchange capacities.
Yeast cells still grow at near normal rates on a fermentable carbon source in the absence of both AACl and AACZ gene products. This is surprising in view of the vital importance of nucleotide exchange between the cytoplasmic and mitochondrial compartments.
It was anticipated that the loss of all nucleotide transport function would be a lethal event. The maintenance of cell viability could result from some low level of residual adenine nucleotide transport. Respiration and cytochrome content is drastically diminished but still definitely present in mitochondria.
The possibility must also be considered that in these cells the very small pool of intramitochondrial nucleotides is segregated from the cytosolic pool. In this case some mitochondrial functions would be sustained by low level endogenous mitochondrial ATP production. It is now well documented that the loss of ATP-dependent protein folding activities within mitochondria are not viable (reviewed in Hart1 and Neupert, 1990). It is anticipated that the loss of all nucleotide transport function would be a lethal event. The maintainance of cell viability likely results from some low level residual adenine nucleotide transport by an as yet unidentified transporter or that the cell adopts low level translocator activity. It is noteworthy in this regard that AACl which appears to be below the limits of detection in membranes containing AAC2 protein is increased to about lo-20% the level of AACZ protein in membranes which lack an AACZ gene product. These data suggest an interesting feedback mechanism for sensing the presence of the AACZ gene product. Alternatively, during rapid growth of yeast there may be a mechanism for segregating pools of adenine nucleotides from the cytoplasm which are sufficient for maintainance of basic vital activities. Reconstitution studies with the Arg9'j to His pet9 protein show that nucleotide transport is still present in the artificial membrane. A 2-fold reduction in turnover under conditions in which the KM for both nucleotides remains essentially unchanged should not reduce the transport of nucleotides below the threshold necessary for mitochondrial-dependent growth. It is possible that the combination of reduced turnover in addition to the 55% reduction in membrane content of the pet9 gene product (Table V) was sufficient to prevent growth on a nonfermentable carbon source. However, a pet9 mutant cannot be restored for growth on glycerol under conditions in which the pet9 allele is also expressed on a multicopy plasmid. Therefore, lack of complementation by the pet9 gene product from a high copy plasmid cannot be ascribed to a simple reduction in nucleotide transport kinetics. It is possible that transport through the translocator in the mitochondrial membrane is much reduced compared with that observed in the reconstituted membrane. It is also possible that the mutation at Arg6 affects the association of AAC with other proteins in the inner membrane and that the constraints imposed or released by the mutation would not be apparent when the protein is the only protein in the membrane. One approach to determining if the mutation effects the association of AAC with other components of the inner membrane is the potential selection of extrageneic suppressors of pet9. One class of extragenic suppressors might be expected to encode compensatory changes in proteins assembled with pet9 AAC in the membrane. These studies are currently in progress.