ATG9A controls all stages of autophagosome biogenesis

ABSTRACT

Canonical autophagy is an intracellular pathway that degrades and recycles cellular components. A key step of this pathway is the formation of double-membraned organelles, known as autophagosomes, an emblematic feature of macroautophagy. For convenience, the formation of autophagosomes can be categorized into sequential steps, initiation (X), expansion (Y) and closure (Z). ATG9A is an integral membrane protein known for its role in the X and Y steps. whereby it organizes phagophore membrane assembly and its growth. Here, we report a previously unappreciated function of mammalian ATG9A in directing the last step Z. In particular, ATG9A partners with the key ESCRT-III component CHMP2A through IQGAP1 to facilitate autophagosome closure. Thus, ATG9A orchestrates all stages of autophagosome membrane biogenesis, from phagophore initiation to its closure. This makes ATG9A a unique ATG factor that works as a central hub in autophagosome biogenesis.

Abbreviation: ATG9A autophagy related 9A; CCCP carbonyl cyanide m-chlorophenylhydrazone; Co-IP co-immunoprecipitation; ESCRT endosomal sorting complexes required for transport; EBSS Earle’s balanced salt solution; ER endoplasmic reticulum; HCM high-content microscopy; HT HaloTag; LC-MS/MS liquid chromatography-tandem mass spectrometry; KO knockout; MPL membrane permeant ligand; MIL membrane impermeant ligand; Mtb Mycobacterium tuberculosis; SolVit sealing of organellar limiting membranes in vitro; TMR tetramethylrhodamine; WT wild type

Canonical autophagy entails the generation of double-membrane organelles called autophagosomes that sequester intracellular material. Autophagosomes are formed through a process involving a suite of ATG (autophagy related) proteins and additional factors taking turns during the sequential stages of initiation (X), expansion (elongation) (Y), and closure (Z) of the phagophore membrane (Figure 1). The final step of phagophore closure is key to a full sequestration of the captured cargo. Whereas much is known about ATG9A involvement in the initiation of phagophore formation and its expansion via lipid transfer, the exact molecular mechanisms governing phagophore closure remain poorly understood.

Figure 1.

Schematic representation of ATG9A’s role at different stages of the autophagy pathway (initiation also known as nucelation, stage X; expansion also known as elongation, stage Y; and closure, stage Z). At stage X, ATG9A vesicles serve as a seeding membrane and interact with components of the autophagy initiation machinery, such as the ULK kinase complex and the phosphatidylinositol 3-kinase complex I (not depicted). During membrane expansion stage Y, ATG9A organizes the phagophore-endoplasmic reticulum membrane contact sites and scrambles lipids transferred to growing phagophores by ATG2 proteins. During stage Z, ATG9A partners with IQGAP1 and recruits CHMP2A to catalyze phagophore closure, forming the double-membrane autophagosome. This sequesters the cytoplasmic material targeted for recycling or elimination such as via lysosomal degradation (not depicted).

ATG9A is a transmembrane protein best known as providing the seeding membrane to initiate phagophore formation and as a lipid scramblase facilitating lipid transfer during phagophore expansion. Using proximity biotinylation LC-MS/MS with APEX2-ATG9A [1], we confirmed that ATG9A interacts with factors participating in the initiation step X (RB1CC1/FIP200, ATG13, and ULK1) and expansion step Y (ATG2B) as well as in membrane atg8ylation (WIPI2, ATG3, and ATG7). In addition, we detected among ATG9A interactors ESCRT proteins shown to participate in the phagophore closure (step Z).

Despite the presence of the majority of ESCRTs in the ATG9A interactome, CHMP2A is conspicuously absent in proteomic data with APEX2-ATG9A. CHPM2A is a component of the ESCRT-III complex that stimulates VPS4-dependent membrane scission, necessary for constriction and severing of negative-curvature membrane necks. However, there is an abundance of IQGAP1 in the ATG9A interactome. Elsewhere, we have shown that ATG9A directly interacts with IQGAP1 and that IQGAP1 directly interacts with CHMP2A. To test whether IQGAP1 acts as an intermediate, bringing ATG9A and CHMP2A together for phagophore closure, we knocked down IQGAP1 and employed the HaloTag-LC3B based MIL/MPL high-content microscopy (HCM) assay to quantify phagophore closure, whereby closed phagophores are MPL-positive and unclosed phagophores are MIL-positive. HCM allows for unbiased, operator-independent, algorithm-based image collection and quantification. The HCM data indicated that IQGAP1 knockdown results in an increase of open phagophores and reduction of closed phagophores/autophagosomes. This was confirmed by additional methods including ^TMRHT release assay and proteinase K protection of autophagic substrates. Thus, IQGAP1 is a hitherto unappreciated factor in phagophore closure.

To address the question of whether IQGAP1 is bridging ATG9A and CHMP2A, we carried out co-IP experiments with GFP-ATG9A and endogenous CHMP2A and observed a decrease of CHMP2A in complexes with ATG9A in cells depleted of IQAGP1. Thus, IQGAP1 bridges ATG9A with CHMP2A, to mediate phagophore closure. Taken together these and additional localization data indicate that ATG9A, IQGAP1, and CHMP2A cooperate in this step of the process.

Quantificative electron microscopy results reveal fewer double-membrane autophagosomal structures in ATG9A KO cells, whereas knockdowns of IQGAP1 and CHMP2A result in accumulation of unclosed phagophores. There is furthermore an unanticipated accumulation of unclosed phagophores containing mitochondria in IQGAP1 and CHMP2A knockdown cells, even though mitochondria were not specifically depolarized/perturbed and thus this is attributed to basal mitophagy. The ultrastructural findings confirmed that ATG9A, CHMP2A and IQGAP1 are critical for phagophore closure.

In the MIL/MPL assay, Huh7 ATG9A KO HT-LC3B cells show reduced MPL levels and elevated MIL levels, a phenotype that is complemented by transient transfection of Huh7 ATG9A KO HT-LC3B cells with FLAG-ATG9A (WT). To test whether the scramblase activity of ATG9A, known to facilitate step Y, is involved in phagophore closure, we employed an ATG9A scramblase mutant. We observed that the scramblase mutant ATG9A[M33] suppresses excess formation of MIL⁺ profiles similarly to WT ATG9A, indicating that the lipid scramblase is not a requisite for progression toward closure. We next employed a technique termed SolVit, developed to study phagophore closure in a cell-free system. This involves sequential MIL/MPL staining of HT-LC3B vesicles in vitro. The SolVit data support ATG9A’s role in phagophore closure.

The interaction between ATG9A and IQGAP1 was modeled by AlphaFold and point mutations generated in the C-terminal region of ATG9A. Two candidate interaction sites were mutationally analyzed: Site 1 ₆₈₉WEGQLQSLVLSEY₇₀₁ (W689S, E690K, L693R, Y701N); and Site 2 ₈₀₁FSRLP₈₀₅ (₈₀₁AAAAA₈₀₅). The ATG9A variant with mutated Site 1 was termed ATG9A[CD1] (CD, closure deficient) while the variant with mutated Site 2 was termed ATG9A[CD2]. Both ATG9A[CD1] and ATG9A[CD2] display reduced binding to IQGAP1. In MPL/MIL HCM experiments with cells induced for autophagy by starvation, we observed recovery of closed autophagosomes in Huh7 ATG9A KO HT-LC3B cells complemented with FLAG-ATG9A (WT). Cells transfected with FLAG-ATG9A[CD1] or FLAG-ATG9A[CD2] retain reduced MPL puncta numbers. In the ^TMRHT release assay, a decline in ^TMRHT fragment release was observed in the Huh7 ATG9A KO cells complemented with FLAG-ATG9A[CD1] or FLAG-ATG9A[CD2] relative to FLAG-ATG9A. Thus, CD1 and CD2 mutations disrupt ATG9A-IQGAP1 interactions and prevent phagophore closure.

Based on the role of ATG9A in all steps (X,Y and Z) of autophagosome biogenesis we realized that it is an excellent choice to probe in vivo roles of the canonical autophagy pathway in various physiological and pathological contexts. We applied this to a controversial area of whether canonical autophagy plays a significant role in protection against Mycobacterium tuberculosis (Mtb) infection using conditional atg9a knockout mice. Uninfected atg9af^l/fl Lyz2-Cre⁺ mice, which lack Atg9a in the myeloid lineage, showed spontaneous mortality for yet to be determined reasons. Mtb infection does not further exacerbate this condition, suggesting that canonical autophagy may not be critical for control of tuberculosis but that instead membrane atg8ylation factors shared with, but not restricted to, the canonical autophagy pathway, as shown elsewhere, are the link between the ATG system and control of Mtb in vivo.

In conclusion, our findings reveal a hitherto unknown function of ATG9A as a regulator of phagophore closure. We identify ATG9A as a hub that sequentially governs all principal steps in autophagosomal membrane biogenesis (Figure 1). This makes ATG9A a special ATG protein overseeing all stages of autophagosome formation.

Funding Statement

The work was supported by the National Institute of Allergy and Infectious Diseases [R37AI042999, R01AI111935] and National Institute of General Medical Sciences [P20GM121176].

Disclosure statement

No potential conflict of interest was reported by the author(s).