Topographical association of a glucokinase with a Golgi-rich fraction from rat liver.

In addition to the well-established cytoplasmic glucokinase (EC 2.7.1.2) (Walker, 1966; Walker & Parry, 1966; Niemeyer et al., 1975), some glucokinase activity possessing different kinetic characteristics is known to be associated with microsomal fractions of rat hepatocytes (Berthillier et al., 1970; Berthillier & Got, 1972). Most of this activity could be recovered in a Golgi-rich microsomal subfraction prepared by zonal centrifugation (Berthillier et al., 1973). We have now prepared a Golgi-rich fraction directly from livers of rats by the procedure of Morrk (1971), thus permitting the study of the topographical relationship of this glucokinase to the Golgi material. This Golgi-rich fraction was enriched in thiamin pyrophosphatase and UDPgalactose-N-acetylglucosamine galactosyltransferase activities and contained the particulate glucokinase. Electron microscopy showed that the preparation was composed largely of vesicles having the same profile as that shown by Morrk (1971). The purity of the fraction was checked by the assay of marker enzymes for contaminating membranes. Negligible amounts of the total 5’-nucleotidase, cytochrome oxidase and glucose 6-phosphatase activities remained in the fraction. Soluble enzymes (glucose &phosphate dehydrogenase, phosphofructokinase, aldolase, pyruvate kinase and lactate dehydrogenase) were not detectable. Several treatments for the liberation of the glucokinase activity from the Golgi membranes were studied. First, extensive washing with a high salt concentration ( 3 ~ KCl) did not liberate any glucokinase activity. This result suggests that, as previous reports had indicated (Berthillier & Got, 1974), the particulate glucokinase is not adsorbed to the outside of the Golgi vesicles. As Triton X-100 is always required to measure the Golgi-bound glucokinase activity, several concentrations of this neutral detergent were tried. A typical latency curve was obtained, latency being lost at a concentration close to the critical micellar concentration (0.015 % of Triton X-100). Four successive freezing-and-thawing operations were necessary to liberate all the latent activity from the Golgi vesicles. Each of the first three operations liberated a portion of the activity bound to the pellet; the last one released the whole of the remaining activity. The total recovery of glucokinase activity was nearly 100 %. The same treatment did not liberate more than 7 % of the galactosyltransferase activity, and no phospholipids were detectable in the supernatants. As phospholipids are constituents of the membranes, their absence from these supernatants suggests that no membrane fragments were present, thus indicating that the glucokinase had been liberated by the treatment in soluble form. This behaviour, which is completely different from that of the galactosyltransferas, indicates that the glucokinase is not, as for the latter enzyme (Bergeron et al., 1973), associated with the membranes of the Golgi elements, but with their contents. This localization was confirmed by sonication. Two sonications for 10s were required for liberation of the glucokinase activity. The total recovery of released glucokinase was 70 %, and phospholipids were again absent from the supernatants. What is the significance of this localization of glucokinase activity inside the Golgi vesicles? As indicated above, it does not appear to be associated with a complete glycolytic pathway, because other glycolytic enzymes are not present. A pathway for the biosynthesis of UDP-glucose from glucose was demonstrated in a microsomal preparation of rat liver (Berthillier & Got, 1974). The first enzyme of this pathway is glucokinase. We find that phosphoglucomutase and UTP-glucose 1-phosphate uridylyl-