|Abstract:Eukaryotic phosphofructokinase, a key regulatory enzyme in glycolysis, has homologous N- and C-terminal domains thought to result from duplication, fusion, and divergence of an ancestral prokaryotic gene. It has been suggested that both the active site and the fructose 2,6-P2 allosteric site are formed by opposing N- and C-termini of subunits oriented antiparallel in a dimer. On the contrary, we show here that in fact the N-terminal halves form the active site, since expression of the N-terminal half of the enzymes from Dictyostelium discoideum and human muscle in phosphofructokinase-deficient yeast restored growth on glucose. However, the N-terminus alone was not stable in vitro. The C-terminus is not catalytic but is needed for stability of the enzyme, as is the connecting peptide that normally joins the two domains (here included in the N-terminus). Co-expression of homologous, but not heterologous, N- and C-termini yielded stable, fully active enzymes in vitro with sizes and kinetic properties similar to those of the wild type tetrameric enzymes. This indicates that the separately translated domains can fold sufficiently well to bind to each other, that such binding of complementary domains is stable and that the alignment is sufficiently accurate and tight as to preserve metabolite binding sites and allosteric interactions.