Autophagy-independent effects of autophagy-related-5 (Atg5) on exosome production and metastasis

ABSTRACT

Autophagy-related-5 (Atg5) and Autophagy-related-16-Like-1 (Atg16L1) canonically participate in autophagy. Recent research demonstrates that apart from this, they also control production of extracellular vesicles called exosomes by regulating acidification of late endosomes. Atg5-mediated exosome production increased migration and metastasis of breast cancer cells suggesting exosomes may perform some functions ascribed to autophagy.

Autophagy is a process of cytoplasmic degradation that promotes cellular survival. In canonical macroautophagy, a de novo membrane engulfs cytoplasmic materials and then seals itself to form an autophagosome.¹ The autophagosome then fuses with lysosomes leading to the degradation of its contents. Over thirty proteins encoded by Autophagy-related genes (Atg) are required for autophagy. Many, including Autophagy-related-5 (Atg5), Autophagy-related-7 (Atg7) and Autophagy-related-16-like-1 (Atg16l1) participate in an enzymatic cascade that culminates in the covalent modification of Lc3-I (official gene name, Microtubule-associated protein 1 light chain-3 [Map1lc3]) with a lipid to form Lc3-II.¹

Deleting genes like Atg7 or Atg5 promotes tumorigenesis in mice.² On the other hand, autophagy can support survival of rapidly growing cancer cells exposed to nutrient-poor environments and chemotherapeutics.^2,3 Autophagy also participates in metastasis and formation of metastatic niches in part by controlling release of matrix metalloproteinases and other proteins through mechanisms which remain less thoroughly investigated.⁴

Intercellular transmission of material by extracellular vesicles of 40–120 nm called exosomes has also been attributed important roles in promoting metastasis and establishing metastatic niches.⁵ Exosomes correspond to vesicles which bud into and accumulate in the lumen of late endosomes then known as multivesicular bodies (MVB). Fusion of MVB with the plasma membrane releases the intraluminal vesicles into the extracellular space where they are referred to as exosomes.⁶ We noticed that significant amounts of Lc3 were detected inside exosomes defined by characteristic density and morphology.⁷ The question that posed itself was how Lc3 was selectively packaged into exosomes as they form at late endosomes. Intriguingly, the Lc3 found in exosomes was in most cases almost entirely lipid-modified Lc3-II. Cells in which Atg7 was downregulated or deleted only produced Lc3-I as expected, but this Lc3-I was still packaged into exosomes.⁷ Exosomes also contained Atg5 and Atg16l1 which bind Lc3 suggesting that this complex recruits Lc3 into exosomes. Supporting this hypothesis, cells genetically deleted of Atg5 or Atg16L1 produced exosomes that did not contain Lc3. Paralleling effects on Lc3 packaging into exosomes, cells lacking Atg5 or Atg16l1 produced few exosomes, while cells depleted of Atg7 demonstrated no deficits in exosome numbers.⁷ This demonstrates that exosome production is controlled by a complex of Atg5 and Atg16l1 independent of canonical autophagy.

To understand how Atg5 and Atg16l1 promoted exosome production we systematically evaluated steps of exosome biogenesis. In Atg5^−/− cells late endosomes still produced intraluminal vesicles as assessed by electron microscopy but these organelles exhibited increased size and acidification, whereas acidification of lysosomes or early endosomes was not similarly affected.⁷ This suggested that Atg5 normally slows acidification of late endosomes. Acidification of endosomes and lysosomes is a function of the V₁V₀-ATPase, a multi-molecular proton pump whose activity is regulated by its subunit composition.⁸ We found that Lc3 associates with ATPase H+ transporting V1 subunit E1 (Atp6v1e1), a cytoplasmic component of the V₁V₀-ATPase, which is released inside exosomes. In Atg5^−/− cells, both Lc3 and Atp6v1e1 were no longer sorted into exosomes, Atp6v1e1 remained associated with the V₁V₀-ATPase at late endosomes, and the pH of these organelles decreased.⁷ This suggested that Lc3 and Atg5 control assembly of the V₁V₀-ATPase and late endosomal acidification by removing Atp6v1e1 (Figure 1). Mimicking this process by knocking-down Atp6v1e1, chemically inhibiting the V₁V₀-ATPase or neutralizing endolysosomal pH, rescued production of exosomes in Atg5^−/− cells.⁷ Cumulatively, this suggests that Atg5 promotes exosome production by regulating the pH of MVBs: it recruits Lc3 to MVB where the latter associates with the V₁V₀-ATPase which it disassembles by sorting Atp6v1e1 into exosomes (Figure 1).

Figure 1.

Autophagy-related-5 (Atg5) and Autophagy-related-16-Like-1 (Atg16L1) promote exosome formation and metastasis of breast cancer independent of Atg7 and canonical autophagy. The V₁V₀-ATPase complex acidifies endolysosomal compartments including multivesicular bodies. LC3 (official gene name, Microtubule-associated protein 1 light chain-3 [MAP1LC3]) can interact with ATP6V1E1 (V1E) a regulatory component of the V₁V₀-ATPase, removing it from the V₁V₀-ATPase into exosomes in a process that requires Atg5. Removal of V1E from the V₁V₀-ATPase causes the pH of multivesicular bodies but not other endolysosomal compartments to increase in wild-type cells and more exosomes are released. In cells lacking Atg5 the V₁V₀-ATPase remains intact, multivesicular body pH is low and few exosomes are released. Neutralizing endolysosomal pH in these cells lacking Atg5 can rescue exosome release. Released exosomes act on neighboring cells to promote migration, invasion and metastasis of breast cancer cells, independent of effects on metastatic niches.

We sought to test the impact of this autophagy-independent effect of Atg5 on exosome production in cancer models. ATG5^−/− MDA-MB-231 breast cancer cells exhibited less migration, wound healing and invasion in vitro and this was rescued by replenishing these cells with levels of exosomes produced by wild-type cells. In turn, in a syngeneic mouse tumor model using 4T1 cells, supplementing Atg5^−/− cells with exosomes up to wild-type levels before injection into the mammary pad decreased tumor size but increased metastasis to the lung.⁷ This suggests that Atg5 is required for exosome production and this promotes migration, invasion and metastasis in models of breast cancer in a process that is independent of autophagy and independent of the role of exosomes in establishing metastatic niches⁵ (Figure 1).

This research has several implications for the field of autophagy, extracellular vesicles and cancer. Atg5, Atg16l1 and Lc3 control acidification of other endolysosomal compartments to regulate phagocytosis, secretory granule release and antigen presentation independent of autophagy.⁹ This research adds to this list of autophagy-independent functions and suggests interactions between Lc3 and the V₁V₀-ATPase may control these and other processes which are dependent on endolysosomal acidification such as signaling from mTOR or receptor tyrosine kinases.^9,10

This research also suggests that in addition to targeting substrates for degradation by autophagy, Lc3 has a novel autophagy-independent function¹⁰ in sorting proteins into exosomes. In this capacity Lc3 may shape the tumor environment and progression by packaging important regulators of oxidative stress (sequestome-1 also called p62, Kelch-like ECH-associated protein 1 [KEAP1]) or epithelial-mesenchymal transition (cateninβ-1) among others into exosomes.

Finally, this research suggests that effects observed in Atg5^−/− or Atg16l1^−/− conditions or after treatment inhibitors of endolysosomal acidification like Chloroquine or Bafilomycin could be due to effects on exosome production rather than autophagy, particularly when corresponding effects are not observed in Atg7^−/− cells or animals. This could include effects of autophagy genes on metastasis and establishing metastatic niches,^2,4 but may also extend to control of receptor signaling, virus production, metal transport, drug export and other processes that are also known to occur at MVB. Resolving the respective roles of exosomes and autophagy in these processes may help direct the pursuit of autophagy- and exosome-modulating drugs as cancer therapeutics.

Funding Statement

National Sciences and Engineering Research Council of Canada (436104) Canadian Cancer Society Research Institute (CCSRI) (702978). R.S. was funded by Fondation France Alzheimer.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.