Takayoshi Kirisako

Autophagy and the Cvt pathway are examples of nonclassical vesicular transport from the cytoplasm to the vacuole via double-membrane vesicles. Apg8/Aut7, which plays an important role in the formation of such vesicles, tends to bind to membranes in spite of its hydrophilic nature. We show here that the nature of the association of Apg8 with membranes changes depending on a series of modifications of the protein itself. First, the carboxy-terminal Arg residue of newly synthesized Apg8 is removed by Apg4/Aut2, a novel cysteine protease, and a Gly residue becomes the carboxy-terminal residue of the protein that is now designated Apg8FG. Subsequently, Apg8FG forms a conjugate with an unidentified molecule "X" and thereby binds tightly to membranes. This modification requires the carboxy-terminal Gly residue of Apg8FG and Apg7, a ubiquitin E1-like enzyme. Finally, the adduct Apg8FG-X is reversed to soluble or loosely membrane-bound Apg8FG by cleavage by Apg4. The mode of action of Apg4, which cleaves both newly synthesized Apg8 and modified Apg8FG, resembles that of deubiquitinating enzymes. A reaction similar to ubiquitination is probably involved in the second modification. The reversible modification of Apg8 appears to be coupled to the membrane dynamics of autophagy and the Cvt pathway.

We characterized Apg8/Aut7p essential for autophagy in yeast. Apg8p was transcriptionally upregulated in response to starvation and mostly existed as a protein bound to membrane under both growing and starvation conditions. Immunofluorescence microscopy revealed that the intracellular localization of Apg8p changed drastically after shift to starvation. Apg8p resided on unidentified tiny dot structures dispersed in the cytoplasm at growing phase. During starvation, it was localized on large punctate structures, some of which were confirmed to be autophagosomes and autophagic bodies by immuno-EM. Besides these structures, we found that Apg8p was enriched on isolation membranes and in electron less-dense regions, which should contain Apg8p-localized membrane- or lipid-containing structures. These structures would represent intermediate structures during autophagosome formation. Here, we also showed that microtubule does not play an essential role in the autophagy in yeast. The result does not match with the previously proposed role of Apg8/Aut7p, delivery of autophagosome to the vacuole along microtubule. Moreover, it is revealed that autophagosome formation is severely impaired in the apg8 null mutant. Apg8p would play an important role in the autophagosome formation.