Bearing in mind a large body of evidence on proton conductivity of certain proteinaceous channels, it seems reasonable to consider peptide protonophores as candidates for low-toxic uncouplers. The ionic channel formed by the pentadecapeptide gramicidin A is known to effectively conduct protons. Unfortunately, it cannot be used as a protonophoric drug in mammalian cells because of high toxicity associated with its high conductivity for all monovalent cations, in BU 4061T particular, potassium and sodium. Controlled uncoupling that prevents undesirable consequences of excessive mitochondrial membrane potential should be one of the major goals of pharmacology. Gramicidin A conventionally used as an uncoupler in isolated mitochondria and chloroplasts could be considered as a candidate for mitochondria uncoupling in tissues. However, gA is characterized by high toxicity, in particular, it perturbs ion balance across plasma membrane even at low concentrations. The gA toxicity also observed here is apparently associated with the high potency of the peptide to form channels that selectively conduct monovalent cations, in particular, potassium and sodium. Therefore, to avoid the gA toxicity, one could try to alter its ionic selectivity by modifying the amino acid sequence or attaching certain groups to its N-termini. As shown in our previous work, substitutions of lysine for position 1 or 3 at neutral pH when this residue is positively charged, strongly diminished the peptide channel-forming potency which could suppress its toxicity for cells. In the present study we described the uncoupling activity of the glutamate-substituted gA analogue in cells and isolated mitochondria which was supported by the data on its protoncarrying activity in a series of model systems. Remarkably, having even higher than gA uncoupling activity, gA was nontoxic for rat kidney primary cell culture, in contrast to gA. The difference in toxicity could be associated with the fact that glutamate residues are mostly deprotonated at pH 7 and thus the majority of gA molecules would carry negative charges at their N-termini, which prevents formation of the conducting parthway for monovalent cations through head-to-head association of two monomers. Moreover, even the uncharged form of gA might form a distorted dimer in overall b6.3–helical conformation with reduced potassium conductance. This follows from the two-orders of magnitude difference in the concentrations of gA and gA required to induce channel formation in BLM. Besides, in liposomes, gA was also much less active than gA. To reconcile the lower activity of gA as compared to gA in model membranes with the higher activity of the analogue as compared to the parent peptide in isolated mitochondria, one should take into account additional membrane barriers encountered by peptide molecules on their way to the target, i.e. the outer membrane of mitochondria and plasma membrane of cells, on one hand, and the absence of such barriers in the case of liposomes and planar bilayers, on the other hand. If one assumes that membrane permeability of gA exceeds that of gA, then the uncoupling potency of gA would be higher than that of gA especially in cells, where the peptide should permeate through at least two membranes before reaching the inner mitochondrial membrane. The fact that the effective concentrations were higher in cells than in mitochondria by three orders of magnitude for gA and to a larger degree for gA proves the permeation through the membrane to be a limiting stage in the uncoupling action of the peptides. This idea is in line with the results of the experiments with peptides preincubated with liposomes.