ADP/ATP transport is the terminal step of oxidative phosphorylation in mitochondria. In this paper seven mutants of AAC2 from Saccharomyces cerevisiae are studied on the cellular and mitochondrial level. Six conspicuously located arginines were mutated into mostly neutral residues [Nelson, D. R., Lawson, J. E., Klingenberg, M., & Douglas, M. G. (1993) J. Mol. Biol. 230, 1159-1170]. R96A, R96H, R204L, and R294A are located in the second transmembrane helix of each repeat while R252I, R253I, and R254I are in the arginine triplet of the last domain. All six arginine residues are conserved in all known ADP/ATP carrier sequences. At the cellular level, oxidative phosphorylation in R96H and R294A retains 8% of the wild-type rate, but it is virtually zero in the other mutants. However, cytochrome c, a parameter of oxidative capacity, remains at 4-42% of wt. The weak coordination of respiratory chain and AAC expression indicates that respiration is needed also for other purposes. In mitochondria the AAC-linked ATP synthesis is measured and segregated by using the AAC inhibitor bongkrekate (BKA). Only the R96H and R294A mutants express a significant rate of AAC-dependent ATP synthesis amounting to 2-18% of the plasmid-borne wild-type AAC2 mitochondria. In all other mutants it is virtually zero. However, respiratory capacity and cytochrome c content are reduced only by 20-70%. Whereas in immunoblots the presence of AAC is detected in all mutant mitochondria, by quantitative ELISA no AAC can be measured down to 0.05 mumol of AAC dimer/g of protein in R96A and R204L, whereas in R96H, R2521, R2531, and R254I the content is around 0.2 and in R294A the content is 0.46 as compared to 0.6 in the plasmid wild type. Also the [3H]CAT and [3H]BKA binding is virtually zero in some mutants and closely parallels the ELISA-determined AAC content, indicating that the mutations did not affect the inhibitor binding site. The turnover of AAC [V(ATP)/AAC content] in oxidative phosphorylation is reduced to 10% or 20% except for the two intrahelical mutants R96H and R294A. In the three Arg triplet mutants, it is nearly zero. In conclusion, the first two intrahelical arginines R96 and R204, are essential for expression but probably also for the activity of AAC. R294A still retains good transport activity and remarkably high expression of AAC. All arginines in the triplet 252, 253, 254 are essential. Extrapolation of the in vitro phosphorylation rates to the cellular level by the cytochrome c factor reveals a large discrepancy to the in vivo rates in particular for R294A. This indicates that these mutations render the AAC more sensitive to the regulatory intracellular ATP/ADP ratio than the wt AAC.