Autophagosome formation and cargo sequestration in the absence of LC3/GABARAPs

ABSTRACT

It has been widely assumed that Atg8 family LC3/GABARAP proteins are essential for the formation of autophagosomes during macroautophagy/autophagy, and the sequestration of cargo during selective autophagy. However, there is little direct evidence on the functional contribution of these proteins to autophagosome biogenesis in mammalian cells. To dissect the functions of LC3/GABARAPs during starvation-induced autophagy and PINK1-PARK2/Parkin-dependent mitophagy, we used CRISPR/Cas9 gene editing to generate knockouts of the LC3 and GABARAP subfamilies, and all 6 Atg8 family proteins in HeLa cells. Unexpectedly, the absence of all LC3/GABARAPs did not prevent the formation of sealed autophagosomes, or selective engulfment of mitochondria during PINK1-PARK2-dependent mitophagy. Despite not being essential for autophagosome formation, the loss of LC3/GABARAPs affected both autophagosome size, and the efficiency at which they are formed. However, the critical autophagy defect in cells lacking LC3/GABARAPs was failure to drive autophagosome-lysosome fusion. Relative to the LC3 subfamily, GABARAPs were found to play a prominent role in autophagosome-lysosome fusion and recruitment of the adaptor protein PLEKHM1. Our work clarifies the essential contribution of Atg8 family proteins to autophagy in promoting autolysosome formation, and reveals the GABARAP subfamily as a key driver of starvation-induced autophagy and PINK1-PARK2-dependent mitophagy. Since LC3/GABARAPs are not essential for mitochondrial cargo sequestration, we propose an additional mechanism of selective autophagy. The model highlights the importance of ubiquitin signals and autophagy receptors for PINK-PARK2-mediated selectivity rather than Atg8 family-LIR-mediated interactions.

The highly conserved Atg8 family of ubiquitin-like proteins localizes to phagophore (the precursor to the autophagosome) membranes during autophagy via conjugation to the lipid phosphatidylethanolamine. Since the discovery of Atg8 lipid conjugation, the degree of lipidation and localization of Atg8 proteins have become the most widely used measures for autophagosome formation in vitro and in vivo. Yeast express a single ATG8 gene, which has been reported to be essential for autophagosome formation. The mammalian Atg8 protein family consists of 6 proteins divided into the LC3 and the GABARAP subfamilies. Both subfamilies are thought to play essential roles during autophagosome biogenesis, including autophagosome expansion and closure, and sequestration of selective autophagy cargo. However, empirical evidence directly linking mammalian Atg8 family proteins to a specific function has remained elusive, owing to the large number of family members in mammals hindering their analysis in isolation.

To determine the roles of the mammalian Atg8 family proteins, we used CRISPR/Cas9 gene editing technology to generate knockouts of the LC3 subfamily (LC3 TKO), the GABARAP subfamily (GBRP TKO), and the combined set (hexa KO) in HeLa cells. The hexa KO cell line is incapable of degrading the autophagic substrate SQSTM1/p62 during starvation-induced autophagy, and mitochondria during PINK1-PARK2-dependent mitophagy. Interestingly, loss of all GABARAPs causes severe autophagy defects whereas LC3 TKO cells display similar levels of starvation-induced autophagy and PINK1-PARK2-dependent mitophagy as wild-type (WT) cells. Although LC3s were found to be dispensable for autophagy activity, they can partly contribute since hexa KO cells have a slightly greater autophagy defect than GABARAP TKOs. Furthermore, despite playing a minor role during the autophagy types tested in our study, LC3s are important for basal autophagy, as are the GABARAPs. The low autophagy activities observed for LC3 subfamily members indicates that they may play more specialized roles for other types of autophagy, including xenophagy. Our results also showed that individual GABARAP proteins are sufficient to rescue autophagy defects in hexa KO cells, although not to levels observed in WT cells. It is therefore possible that functional cooperation between the GABARAPs themselves or between GABARAPs and LC3s improves autophagic efficiency.

To test the prevailing hypothesis that LC3/GABARAPs are required for the generation of sealed autophagosomes, we used transmission electron microscopy and immuno-electron microscopy to visualize autophagosomal structures in hexa KO cells during starvation-induced autophagy and PINK1-PARK2-dependent mitophagy. To our surprise, correctly sealed autophagosomes are present in cells lacking LC3/GABARAPs in equivalent or greater numbers than seen in WT cells. Three-dimensional reconstruction of autophagosomes in hexa KOs and biochemical protease protection assays also supported the conclusion that the membranes are sealed across the entire autophagosomal surface. These results clearly indicate that the mammalian Atg8 family proteins are not essential for autophagosome formation and closure. It therefore remains to be determined precisely how autophagosomes seal, but suggestions have been made that a membrane scission event is required for closure.

Although the absence of LC3/GABARAPs does not prevent autophagosome biogenesis, these proteins play an important role in expanding phagophore membranes, as has been previously proposed. This is best exemplified by the fact that in hexa KO cells, starvation-induced autophagosomes and PINK1-PARK2-dependent mitophagosomes were smaller and therefore had less capacity for cargo. For example, hexa KO mitophagosomes rarely contain more than one mitochondrion. Mitophagosomes in WT, GBRP TKO and LC3 TKO cells contain multiple mitochondria within the same delimiting membrane. Elongation of the sequestering membrane is an underappreciated rate limiting factor in autophagic flux, as it determines the rate of autophagosome closure and the total size of the autophagosome. Indeed, autophagosomes form at a slower rate in the absence of LC3s and GABARAPs.

Successful formation of autophagosomes containing cargo in hexa KO and GBRP TKO cells indicates a defect during later stages of autophagy involving the fusion of autophagosomes with lysosomes. Analyses using fluorescence microscopy, immuno-gold electron microscopy and the mtKeima mitophagy assay demonstrated that LC3/GABARAPs are crucial for mediating autophagosome-lysosome fusion events. Loss of LC3/GABARAPs abolishes the recruitment of the adaptor protein PLEKHM1 to the autophagosomal surface. PLEKHM1 plays a critical role in recruiting the HOPS complex, which together with RAB7 promotes delivery of sequestered cargo to endolysosomal compartments for degradation (Fig. 1). Notably, PLEKHM1 is preferentially recruited by GABARAPs, potentially explaining why a significant block in autophagic activity was observed in GBRP TKO cells but not LC3 TKO cells. Given that PLEKHM1 contains an LC3 interacting region (LIR) motif, it is likely that LIR-mediated interactions with Atg8 family proteins are important to recruit PLEKHM1 and drive autophagosome-lysosome fusion, as has been reported in recent studies.

Figure 1.

Model of autophagosome biogenesis during PINK1-PARK2-dependent mitophagy in WT and hexa KO cells. 1) Initiation: During PINK1-PARK2-dependent mitophagy, autophagosome biogenesis begins after the accumulation of S65 phosphorylated ubiquitin conjugated to mitochondrial outer membrane proteins. The S65 phosphorylated ubiquitin chains promote recruitment of the autophagy receptors CALCOCO2 and OPTN, which initiate autophagosome biogenesis by promoting recruitment of ULK1. 2) Nucleation: The ULK1 complex and phosphatidylinositol 3-kinase (PtdIns3K) complexes are recruited to the initiation site and govern the nucleation of an autophagosome on the surface of the mitochondrion. The specific role of each kinase is unclear, however, PtdIns3K activity is essential for autophagosome biogenesis and promotes recruitment of phosphatidylinositol-3-phosphate (PtdIns3P) binding proteins including ZFYVE1/DFCP1 and the WIPIs. After nucleation of the sequestering membrane, the ATG12–ATG5-ATG16L1 complex is recruited to conjugate Atg8 family proteins to lipid. 3) Expansion: The sequestering membranes elongate around the mitochondria using lipid from various membrane sources, including the endoplasmic reticulum. The ATG12–ATG5-ATG16L1 complex proceeds to conjugate Atg8 family proteins to phosphatidylethanolamine in the membrane. This does not occur in the hexa KO cell line due to a lack of the 6 mammalian Atg8 family proteins. Membrane expansion is therefore impaired in the hexa KO cell line, resulting in smaller autophagosomes and slow autophagosome formation efficiency. 4) Closure: Through an undefined mechanism, that may involve membrane scission, the phagophore membranes seal to encapsulate their mitochondrial cargo. 5) Maturation: Upon recruitment of PLEKHM1 through LIR-mediated interactions with Atg8 family proteins, the sealed autophagosome enters its terminal stages of maturation. Autophagosomal maturation occurs via HOPS complex-dependent fusion with RAB7-positive endolysosomal organelles, enabling autophagosome acidification and the acquisition of lysosomal proteases required to degrade sequestered content. Autophagosomes in the hexa KO cell line do not undergo autophagosomal maturation, as they do not recruit PLEKHM1.

In current models of selective autophagy, LC3/GABARAPs function to recruit and incorporate specific cargo into pre-formed phagophore membranes. The selectivity is achieved via LIR-mediated interactions between LC3/GABARAPs and autophagy receptors on the cargo. Our finding that autophagosomes in hexa KOs correctly sequester mitochondria following PINK1-PARK2 activation was therefore highly unexpected. It also raised an important question; what is the role of the LIR motif during PINK1-PARK2-dependent mitophagy? The answer lies in the mechanism of autophagosome biogenesis after PINK1-PARK2 activation. Rather than recruiting pre-formed phagophore membranes, PINK1-PARK2-induced autophagosomes are constructed locally on the surface of damaged mitochondria. S65 phosphorylated ubiquitin chains generated by PINK1 and PARK2 preferentially recruit CALCOCO2/NDP52 and OPTN (optineurin) to allow subsequent recruitment of the ULK1 complex and initiate autophagosome biogenesis on the mitochondrial surface. Thus, it is the presence of the S65 phosphorylated ubiquitin chains, not Atg8 family proteins, that determines the selectivity of PINK1-PARK2-dependent mitophagy. The likely role of the LIR in OPTN and CALCOCO2 is to recruit LC3/GABARAPs and promote phagophore expansion and formation efficiency. LIR interactions also recruit PLEKHM1 for autolysosome formation to degrade mitochondria (Fig. 1). In contrast, ubiquitin-independent pathways of selective autophagy, such as BNIP3L/Nix-mediated mitophagy critically depend on LC3/GABARAP-LIR-mediated interactions for selectivity and cargo sequestration. Whether the LIR interactions recruit preformed autophagosome precursors or initiate autophagosome biogenesis on the cargo remains to be determined. Notably, ULK1 has a LIR motif that preferentially interacts with GABARAPs suggesting a possible mechanism for the latter scenario. Taken together, we speculate that 2 mechanistically distinct paths of selective autophagy exist: 1) ubiquitin-mediated selective autophagy pathways in which LIR-mediated interactions are dispensable for selectivity; and 2) selective autophagy that relies on receptors resident on cargo that critically depends on LIR-mediated interactions to sequester the cargo within phagophores.

Our work provides a platform for future studies to investigate the role of each Atg8 subfamily as well as individual LC3s and GABARAPs in other types of selective autophagy such as xenophagy or aggrephagy. Furthermore, since Atg8 family proteins are not crucial for autophagosome formation/closure, further investigation is required to dissect the mechanisms of autophagosome biogenesis.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.