The rules that govern complementation of mutant and wild-type mitochondrial genomes in human cells were investigated under different experimental conditions. Among mitochondrial transformants derived from an individual affected by the MERRF (myoclonus epilepsy associated with ragged red fibers) encephalomyopathy and carrying in heteroplasmic form the mitochondrial tRNALYS mutation associated with that syndrome, normal protein synthesis and respiration was observed when the wild-type mitochondrial DNA exceeded 10% of the total complement. In these transformants, the protective effect of wild-type mitochondrial DNA was shown to involve interactions of the mutant and wild-type gene products. Very different results were obtained in experiments in which two mitochondrial DNAs carrying nonallelic disease-causing mutations were sequentially introduced within distinct organelles into the same human mitochondrial DNA-less (p°) cell. In transformants exhibiting different ratios of the two genomes, no evidence of cooperation between their products was observed, even 3 months after the introduction of the second mutation. These results pointed to the phenotypic independence of the two genomes. A similar conclusion was reached in experiments in which mitochondria carrying a chloramphenicol resistance-inducing mitochondrial DNA mutation were introduced into chloramphenicol-sensitive cells. A plausible interpretation of the different results obtained in the latter two sets of experiments, compared with the complementation behavior observed in the heteroplasmic MERRF transformants, is that in the latter, the mutant and wild-type genomes coexisted in the same organelles from the time of the mutation. This would imply that the way in which mitochondrial DNA is sorted among different organelles plays a fundamental role in determining the oxidative-phosphorylation phenotype in mammalian cells. These results have significant implications for mitochondrial genetics and for studies on the transmission and therapy of mitochondrial DNA-linked diseases.Despite the large number of studies on the expression of artificially produced mitochondrial DNA (mtDNA) mutations and the phenotype of somatic cell hybrids and cybrids, information on mitochondrial genome interactions in mammalian cells is still very limited. The dependence of such interactions on the distribution of the mtDNA molecules among the mitochondria, which changes continuously throughout division and, possibly, fusion of the organelles, and the difficulty of investigating mtDNA sorting in a cell account in part for the present lack of understanding of how much the mitochondrial genomes interact in mammalian cells. The recent discovery of a variety of mtDNA mutations associated with diseases in humans (47) and the growing evidence of accumulation of mtDNA mutations with aging in somatic tissues (12-14, 19, 43) have raised a number of questions about the occurrence and frequency of intermixing of mutant and wild-type mtDNA and/or their products within a cell and the role that co...