The catalytic action of 5-deazaflavin in the photochemical reduction of flavin and iron proteins [Massey, V. and Hemmerich, P. (1978) Biochemistry, 17, 9-17] is shown to be due to the highly reactive 5-deazaflavosemiquinone. This radical is generated in a complex sequence of reactions, which involves (a) covalent photoaddition of the substrate residue to the deazaflavin, (b) fast secondary photoreaction of this adduct with starting deazaflavin to yield a covalent radical dimer, accompanied by the liberation of the oxidized substrate, and (c) deazaflavin-sensitized cleavage of the radical dimer to the monomers. The structure and properties of this radical (redimerisation or dismutation) and the precursor intermediates as well as the mechanism of the photoreaction are described. Deazaflavins and their natural parent compounds are compared with respect to their different redox behavior and radical stability. The syntheses of 5-deuterated deazaflavins are described and their redox reactions are compared with those of normal deazaflavins.Among the three principal types of flavoprotein activities, i.e. (de)hydrogenation, 0 2 activation and e-transfer, only the first is retained in deazaflavoproteins, and even this to rather limited extent in most cases [l -71. The initial overestimation of deazaflavins as flavin analogs must be ascribed to the lack of chemical knowledge about this heterocyclic system. For example, Briistlein and Bruice [8] were among the first to carry out chemical studies and they concluded, from the retention of H label in the course of reduction, that deazaflavin was transferring hydride and that, by analogy, flavin was a hydride-transferring entity also. Edmondson et al. [2] have been studying the system more extensively, but they have been deceived by the fact that there is a physical analogy between the oxidized species Fl,, and dF1, which, however, does not extend to the radical HdFl nor to the fully reduced state HZdFl,,d. These authors, however, were the first to state the lack of radical stability in deazaflavoproteins as compared to natural flavoproteins. This was corroborated by the inactivity of modified D-amino acid oxidase radical produced photochemically, as reported by Jorns and Hersh [6]. Based upon theoretical studies, Sun and Song [9] claimed that 'the reactivity of position 5 is approximately independent of the identity of atom (N or C)'.On the contrary, we want to show that the analogy between flavin and deazaflavin is rather limited, deazaflavin being a 'flavin-shaped nicotinamide model ' [ 11. Spencer et al. [lo] were the first to see 'carbanion adducts' and 'two-electron disproportionation', a phenomenon which should rather be termed 'interdeazaflavin hydride transfer'. This study does not yield much information on the half-reduced deazaflavin. Radicals do not occur as essential intermediates in deazaflavin-linked oxidoreduction, unless enforced le--transfer methods are applied, as we will show below.Based on our new and simple synthesis of the deazaflavin skeleton [ll], ...