Fusion constructs incorporating structural elements from mammalian isozymes of hexokinase, Types I-IV, in frame with sequence encoding the green fluorescent protein (GFP) have been made and expressed in hexokinase-deficient M + R 42 cells. Fusion proteins incorporating catalytically active regions from the Type II isozyme, or the entire Type IV sequence, were expressed in catalytically active form. The intracellular localization of the fusion proteins was determined using confocal microscopy. Fusion proteins including the N-terminal halves of the Type I or Type II isozymes were targeted to mitochondria, while the N-terminal half of the Type III isozyme did not confer mitochondrial targeting. The mitochondrial targeting signal was represented by the hydrophobic sequence at the extreme N-termini ("binding domain") of the Type I and Type II isozymes. Inclusion of the binding domain from the Type I isozyme was sufficient to confer mitochondrial binding on GFP itself as well as on constructs including the N-terminal half of Type III hexokinase. However, the Type I hexokinase binding domain was not sufficient to cause mitochondrial targeting of a construct containing the Type IV sequence. These results suggest that, although the binding domain is critical for mitochondrial targeting, other interactions involving an adjacent structure might also play a role. Fusion proteins including the N-terminal half of Type I hexokinase became dissociated from mitochondria under conditions favorable for accumulation of intracellular Glc-6-P. The 2-deoxy analog was much less effective than Glc in causing mitochondrial dissociation of the fusion construct, in accord with previous studies showing 2-deoxy-Glc-6-P to be much less effective than Glc-6-P at promoting release of Type I hexokinase from mitochondria. Dissociation, induced by formation of Glc-6-P or 2-deoxy-Glc-6-P, did not occur with the fusion protein including only the binding domain of Type I hexokinase. This is consistent with previous studies indicating that Glc-6-P-dependent dissociation results from binding of this ligand to a site in the N-terminal half of the enzyme, but which is not likely to be present in the small segment represented by the binding domain. These studies demonstrate the usefulness of this approach in defining structural elements involved in targeting hexokinase isozymes to specific subcellular locations and modulation of that intracellular location by perturbations of metabolic status.