Models of F-actin structure predict the importance of hydrophobic loop 262-274 at the interface of subdomains 3 and 4 to interstrand interactions in filaments. If this premise is correct, prevention of the loop conformational change--its swinging motion--should abort filament formation. To test this hypothesis, we used site-directed mutagenesis to create yeast actin triple mutant (LC)2CA (L180C/L269C/C374A). This mutation places two cysteine residues in positions potentially enabling the locking of loop 262-274 to the monomer surface via disulfide formation. Exposure of the purified mutant to oxidation catalysts resulted in an increased electrophoretic mobility of actin on SDS PAGE and a loss of two cysteines by DTNB titrations, consistent with disulfide formation. The polymerization of un-cross-linked mutant actin by MgCl2 was inhibited strongly but could be restored to wild type actin levels by phalloidin and improved greatly through copolymerization with the wild-type actin. Light scattering measurements revealed nonspecific aggregation of the cross-linked actin under the same conditions. Electron microscopy confirmed the absence of filaments and the presence of amorphous aggregates in the cross-linked actin samples. Reduction of the disulfide bond by DTT restored normal actin polymerization in the presence of MgCl2 and phalloidin. These observations provide strong experimental support for a critical role of the hydrophobic loop 262-274 in the polymerization of actin into filaments.