Reconstitution of active octameric mitochondrial creatine kinase from two genetically engineered fragments

Creatine kinase (CK) has been postulated to consist of two flexibly hinged domains. A previously demonstrated protease-sensitive site in M-CK (Morris & Jackson, 1991) has directed our attempts to dissect mitochondrial CK (Mi-CK) into two protein fragments encompassing amino acids [1-167] and [168-380]. When expressed separately in Escherichia coli, the two fragments yielded large amounts of insoluble inclusion bodies, from which the respective polypeptides could be purified by a simple two-step procedure. In contrast, co-expression of the two fragments yielded a soluble, active, and correctly oligomerizing enzyme. This discontinuous CK showed nearly full specific activity and was virtually indistinguishable from native Mi-CK by far- and near-UV CD. However, the positive cooperativity of substrate binding was abolished, suggesting a role of the covalent domain linkage in the crosstalk between the substrate binding sites for ATP and creatine. The isolated C-terminal fragment refolded into a native-like conformation in vitro, whereas the N-terminal fragment was largely unfolded. Prefolded [168-380] interacted in vitro with [1-167] to form an active enzyme. Kinetic analysis indicated that the fragments associate rapidly and with high affinity (1/K1 = 17 microM) and then isomerize slowly to an active enzyme (k2 = 0.12 min-1; k-2 = 0.03 min-1). Our data suggest that the C-terminal fragment of Mi-CK represents an autonomous folding unit, and that the folding of the C-terminal part might precede the conformational stabilization of the N-terminal moiety in vivo.