Diphthamide, the target of diphtheria toxin, is a unique posttranslational modification on eukaryotic and archaeal translation elongation factor 2 (EF2). The proposed biosynthesis of diphthamide involves three steps and we have recently found that in Pyrococcus horikoshii (P. horikoshii), the first step uses an S-adenosyl-L-methionine (SAM)-dependent [4Fe–4S] enzyme, PhDph2, to catalyze the formation of a C–C bond. Crystal structure shows that PhDph2 is a homodimer and each monomer contains three conserved cysteine residues that can bind a [4Fe–4S] cluster. In the reduced state, the [4Fe–4S] cluster can provide one electron to reductively cleave the bound SAM molecule. However, different from classical radical SAM family of enzymes, biochemical evidence suggest that a 3-amino-3-carboxypropyl radical is generated in PhDph2. Here we present evidence supporting that the 3-amino-3-carboxypropyl radical does not undergo hydrogen abstraction reaction, which is observed for the deoxyadenosyl radical in classical radical SAM enzymes. Instead, the 3-amino-3-carboxypropyl radical is added to the imidazole ring in the pathway towards the formation of the product. Furthermore, our data suggest that the chemistry requires only one [4Fe–4S] cluster to be present in the PhDph2 dimer.