Macroautophagy is a very well known process in mammalian cells, and it is mechanistically identical to the process that occurs in yeast cells (21, 22, 25, 59, 75, 103, 104, 119). Macroautophagy has also been demonstrated in plants (4, 16, 120).
Several molecular components of the mammalian machinery have been cloned because of their homology to genes identified in S. cerevisiae (Table 1). In particular, the two UBL systems are conserved. Aut7 is the core of one of these two systems, and its rat homologue, LC3, has identical functional characteristics. LC3 is cytosolic, but after processing it becomes membrane bound and localizes to both the inside and the outside of forming autophagosomes (44, 74). In addition, the human LC3 homologue is also activated by the Apg7 homologue (hApg7) (115). The human homologue of the second UBL, Apg12 (hApg12), is similarly activated by hApg7 and finally forms a complex with the Apg5 homologue (hApg5) (73, 115). The mouse Apg12-Apg5 conjugate is essential for macroautophagy and localizes to the forming autophagosomes (74). As in yeast cells, the Apg12-Apg5 complex is necessary for the last step of the LC3/Aut7 conjugation system that leads to the tight association of LC3 with the nascent autophagosome (51, 74). LC3/Aut7 is probably one of the principal structural elements of the autophagosome, and the absence of its targeting to membranes in Apg5-deficient mouse embryonic stem cells can explain why those cells are impaired in autophagosome formation (2, 37, 56, 74).
The human Bcl-2-interacting protein, Beclin 1, is the functional homologue of yeast Vps30/Apg6 (see below) (68). Beclin 1 is able to complement the autophagy deficiency of yeast cells lacking Vps30/Apg6, but it also restores autophagy in human MCF7 breast carcinoma cells and reduces the tumorigenicity of various other malignant cell lines (68). In yeast cells, Vps30/Apg6 is associated with another protein required for macroautophagy: the phosphatidylinositol (PI) 3-kinase Vps34 (see below) (49). PI 3-phosphate also has been shown to be essential for autophagy in mammalian cells (11, 86).
Even if the basic machinery for autophagy is identical between yeast and mammalian cells, the possibility that this phenomenon is more complex in higher eukaryotes cannot be excluded. For example, in addition to LC3, there are two more human Aut7 counterparts that play a role in other cellular processes (44, 66, 85, 91, 115, 133). Interestingly, all three human Aut7 homologues are activated by hApg7, suggesting that various mammalian pathways are coordinated during the induction of autophagy (115). This variegation of Aut7-like proteins is also present in other higher eukaryotes (Table 1). In yeast cells these multiple functions are probably accomplished by a single protein (66). Aut2 is the protease involved in the processing of Aut7 and then in its subsequent release from membranes (see above) (56). It seems that there are also several Aut2 homologues in every higher eukaryotic organism (Table 1). This finding may reflect the necessity to process different Aut7-like factors that are ultimately linked to different substrates. In yeast cells, Aut7 is covalently bound to PE (see above) (40); further investigations are required to demonstrate if the same is true for all the mammalian Aut7 homologues.
In addition to S. cerevisiae, several other yeasts can modulate their peroxisome population to better exploit nutrient conditions. When methylotrophic yeasts such as Pichia pastoris, Pichia methanolica, Hansenula polymorpha, and Candida boidinii are grown in media containing methanol, peroxisome biogenesis is induced in order to optimize the utilization of this carbon source (125, 127). Upon adaptation to an alternative carbon source, such as glucose or ethanol, peroxisomes are rapidly degraded in the vacuole (13, 35, 62, 92, 122, 126, 130). An identical fate is also reserved for the peroxisomes of Yarrowia lipolytica and Aspergillus nidulans after shifting of those fungi from a medium containing oleic acid to one containing glucose (3, 30). Pexophagy occurs by two mechanisms. Macropexophagy is morphologically identical to macroautophagy in S. cerevisiae, while micropexophagy involves uptake of the peroxisomes directly at the vacuole surface (reviewed in references 54 and 111) (Fig. 1).
Genetic screens to identify mutants defective in pexophagy have been performed with H. polymorpha, P. pastoris, and Y. lipolytica (30, 92, 118, 122). The identified genes, PDD1, PpVPS15/GSA19, PDD7, GSA7, GSA9, GSA10, GSA11, GSA12, and GSA14, are functional homologues of the S. cerevisiae genes VPS34, VPS15, APG1, APG7, CVT9, APG1, APG2, CVT18, and APG9, respectively, showing that pexophagy in yeasts employs the same machinery as that required for macroautophagy and the Cvt pathway (29, 48, 53, 108, 110, 129, 140). The P. pastoris GSA1 gene is also essential for pexophagy and encodes the regulatory subunit of phosphofructokinase, an enzyme that itself is not required for pexophagy (141). The role of Gsa1 is not clear, and it is not known if mutations in the phosphofructokinases of other yeasts have the same effect. Pexophagy has also been observed in mammalian cells (12, 69), but there are no reports characterizing the machinery used for this process.
The existence of the Cvt pathway has only been demonstrated in the yeast S. cerevisiae. Even if the autophagy machinery is present in all eukaryotic cells, homologues of the factors specific for the Cvt pathway, such as Cvt9 or Vac8, seem to be restricted only to other yeasts (Table 1). The major difficulty in the identification of this transport route in other organisms is that no putative cargo molecules are known, even if it seems that Ape1 and Ams1 have counterparts. Ape1 has clear homologues in other eukaryotic cells, but it is difficult to predict if they follow a Cvt pathway because alignments show that those proteins have divergent N termini. This region contains the information to target prApe1 to the Cvt pathway (82). These divergences may reflect differences in the targeting determinant but may also be symptomatic of a diverse localization. For example, the closest mammalian homologue of Ape1 localizes to the cytosol (138). Ams1 also has homologues in Schizosaccharomyces pombe, A. nidulans, humans, and rats. None of these homologues has a classical signal sequence or transmembrane domains, suggesting they may not travel through the secretory pathway (23, 38). Confusingly, the rat homologue localizes to the endoplasmic reticulum (10). More detailed studies are required to shown where and how Ape1 and Ams1 homologues are localized in other organisms. The absence of Cvt9 or Vac8 but also of the Cvt19 receptor in higher eukaryotes may reflect only the fact that the Cvt pathway in these organisms selects other cargo molecules.