Interpretation of a new electron-density map at 3.25-A resolution has led to a somewhat revised model for the free radical (semiquinone) structure of flavodoxin from Clostridium MP. Although the general conformation of the molecule is very similar to that of oxidized Desulfovibrio vulgaris flavodoxin, flavin mononucleotide-protein interactions are not identical in the two flavodoxins. In the Cl. MP semiquinone molecule, the isoalloxazine ring appears to retain the essentially planar conformation characteristic of oxidized flavins; within the limits imposed by the resolution of the data, the map shows no evidence for bending of the isoalloxazine ring about N5-N1O.Analysis of the structure of flavodoxin crystals is expected to provide a detailed picture of flavin mononucleotide-protein interactions, which account for the characteristic chemical properties of these model flavoproteins (1, 2). Earlier we reported an interpretation of an electron-density map of Clostridium MP flavodoxin semiquinone at 3.25-A resolution. The model constructed on the basis of that map was considered tentative in several regions, and the orientation of flavin mononucleotide could not be assigned unequivocally (3).Recently, we calculated another map of the semiquinone form of flavodoxin at 3.25-A resolution, using three heavyatom derivatives, Sm+3, Pt+2, and Au+', and giving more weight to anomalous scattering differences in the phasing. The resulting map had a much improved figure of merit, 0.80, and has proven easier to interpret. The new map clearly suggested a flavin mononucleotide position different from that proposed earlier and required some revisions in the chain tracing ( Fig. 1). At about the same time, Watenpaugh et al. (4) succeeded in determining the structure of the oxidized form of a very similar protein from Desulfovibrio vulgaris at 2.5-A resolution. Part of our current model of the semiquinone form was constructed after a preliminary sketch of the model of Watenpaugh et al. had been kindly sent to us by Dr. Jensen. We describe our new molecular model briefly here in order to facilitate comparison of the two structures (5).Despite the greater chain length of flavodoxin from D.vulgaris (about 10 additional residues), the three-dimensional structures of the two molecules are clearly homologous. The prosthetic group is similarly situated in both proteins, but the flavin mononucleotide-protein interactions appear to differ in certain respects. Some differences are not unexpected in view of the known dissimilarities in amino-acid composition (3, 6), in optical and circular dichroic spectra (7), and in affin'ty for modified flavins (7,8 Fig. 1. In the matching regions, the average displacement of a total of 113 alpha carbons in constructing the new model was 1.7 A; in the best-defined helix, HI, the change was only 0.5 A. However, altered chain connections in the vicinity of the Nterminus and the flavin mononucleotide have led to a quite different ordering of the amino-acid sequence along the model. In the interpretati...