The dynamics of interactions between rat liver cytosolic proteins and arsenicals were examined in an in vitro methylation system that contained cytosol, glutathione, S-adenosylmethionine, and 1 mu M [As-73]arsenite. After incubation at 37 degrees C for up to 90 min, low-molecular-weight components of the assay system (<10 kDa) were removed by ultrafiltration and cytosolic proteins were separated by size-exclusion chromatography on Sephacryl S-300 gel. Five As-73-labeled protein peaks were found in chromatographic profiles. The estimated molecular masses of As-73-labeled proteins eluting in the three earliest peaks were as follows: V-o, greater than or equal to 1000 kDa; A, 135 kDa; and B, 38 kDa. Peak C eluted immediately before the total volume (V-T) of the chromatographic column; peak D eluted after the V-T. As-73 bound to proteins was released by CuCl treatment and speciated by thin-layer chromatography. Amounts and ratios of inorganic As, methyl As, and dimethyl As associated with cytosolic proteins depended upon the incubation interval. Inorganic As was present in all protein peaks. Methyl As was primarily associated with peaks A and C; dimethyl As was associated with peaks B and C. To examine the effect of valence on the binding of methylarsenicals to cytosolic proteins, trivalent or pentavalent C-14-labeled methyl As or dimethyl As was incubated in an in vitro system designed to minimize the enzymatically catalyzed production of methylated arsenicals. Proteins in peaks A, B, and C bound preferentially trivalent methyl and dimethyl As. Peak D bound either trivalent or pentavalent methyl and dimethyl As. Protein-bound inorganic and methyl As were substrates for the production of dimethyl As in an in vitro methylation system, suggesting a role for protein-bound arsenicals in the biomethylation of this metalloid. (C) 1997 Academic Press.