Membrane adhesion dictates Golgi stacking and cisternal morphology.

Two classes of proteins that bind to each other and to Golgi membranes have been implicated in the adhesion of Golgi cisternae to each other to form their characteristic stacks: Golgi reassembly and stacking proteins 55 and 65 (GRASP55 and GRASP65) and Golgin of 45 kDa and Golgi matrix protein of 130 kDa. We report here that efficient stacking occurs in the absence of GRASP65/55 when either Golgin is overexpressed, as judged by quantitative electron microscopy. The Golgi stacks in these GRASP-deficient HeLa cells were normal both in morphology and in anterograde cargo transport. This suggests the simple hypothesis that the total amount of adhesive energy gluing cisternae dictates Golgi cisternal stacking, irrespective of which molecules mediate the adhesive process. In support of this hypothesis, we show that adding artificial adhesive energy between cisternae and mitochondria by dimerizing rapamycin-binding domain and FK506-binding protein domains that are attached to cisternal adhesive proteins allows mitochondria to invade the stack and even replace Golgi cisternae within a few hours. These results indicate that although Golgi stacking is a highly complicated process involving a large number of adhesive and regulatory proteins, the overriding principle of a Golgi stack assembly is likely to be quite simple. From this simplified perspective, we propose a model, based on cisternal adhesion and cisternal maturation as the two core principles, illustrating how the most ancient form of Golgi stacking might have occurred using only weak cisternal adhesive processes because of the differential between the rate of influx and outflux of membrane transport through the Golgi.