Vitamin B6 is an essential metabolite in all organisms, being required as a cofactor for a wide variety of biochemical reactions. De novo biosynthesis of the vitamin occurs in microorganisms and plants, but animals must obtain it from their diet. Two distinct and mutually exclusive de novo pathways have been identified to date, namely deoxyxylulose 5-phosphate dependent, which is restricted to a subset of eubacteria, and deoxyxylulose 5-phosphate independent, present in archaea, fungi, plants, protista, and most eubacteria. In these organisms, pyridoxal 5'-phosphate (PLP) formation is catalyzed by a single glutamine amidotransferase (PLP synthase) composed of a glutaminase domain, PDX2, and a synthase domain, PDX1. Despite plants being an important source of vitamin B6, very little is known about its biosynthesis. Here, we provide information for Arabidopsis thaliana. The functionality of PDX2 is demonstrated, using both in vitro and in vivo analyses. The expression pattern of PDX2 is assessed at both the RNA and protein level, providing insight into the spatial and temporal pattern of vitamin B6 biosynthesis. We then provide a detailed biochemical analysis of the plant PLP synthase complex. While the active sites of PDX1 and PDX2 are remote from each other, coordination of catalysis is much more pronounced with the plant proteins than its bacterial counterpart, Bacillus subtilis. Based on a model of the PDX1/PDX2 complex, mutation of a single residue uncouples enzyme coordination and in turn provides tangible evidence for the existence of the recently proposed ammonia tunnel through the core of PDX1.