An analysis of dimethylsulfoxide-induced action potential block: a comparative study of DMSO and other aliphatic...

A series of water soluble aliphatic solutes were chosen for study. Fifty percent effective doses (ED50) to block propagated compound action potentials (AP's) were obtained by examining dose-response relations for each solute. All solutes used were liquids at room temperature and are typically used as solvents. The solutes studied were dimethylsulfoxide (DMSO), dimethylformamide (DMF), dimethylacetamide, acetone, and hexamethylphosphoramide (HMPA); the octanol/water partition coefficients for these test substances form an ordered sequence that increased 40-fold from DMSO to HMPA. AP's were recorded from desheathed frog sciatic nerves using the sucrose-gap technique; test solutes were added to Ringer's solution and applied externally to the nerve. ED50's for the solutes could be predicted as a function of the molar volume (dV/dn), polarity (P), and the hydrogen bond acceptor basicity (beta). Voltage-clamp experiments employing the vaseline-gap technique on single muscle fibers showed that each solute reduced Na+ current with little change in their kinetics at all voltages studied. Experiments using DMSO or DMF showed that Na+ channel block alone is insufficient to explain the respective ED50 values of AP block. Experiments conducted using a chloride transport-sensitive membrane fluidity assay, using rat pancreas secretory granules, suggested that each of the solutes act to increase membrane fluidity at doses below and above ED50 values. Light microscopic observations of fixed thick sections of whole nerves previously exposed to DMSO or DMF show structural changes; however, ED50 values cannot be simply explained by osmotic alterations of nerve structure. ED50's are likely to be produced by a combination of effects including osmotically induced nerve structural changes, ion channel block, and fluidity changes. The toxicity (lethal doses or toxic concentrations) of each of these five solutes correlates well with the ED50 and could be predicted as a function of dV/dn, P, and beta.