Abstract
The phagophore membrane is highly curved along the rim of the open cup, suggesting that the molecular mechanisms governing its formation and growth could rely on membrane curvature-dependent events. To this end, we recently reported that lipidation of the LC3 protein family is facilitated on highly curved membranes in vitro. We further showed that the conjugating enzyme ATG3 contains an amphipathic helix that is responsible for this membrane curvature dependency, and that the maintenance of this amphipathic structure is essential for ATG3 function in vivo.
Keywords: curvature, Atg3, lipidation, LC3, phosphatidylethanolamine
The autophagosome precursor or phagophore (PG) adopts a cup-like structure that grows and closes on itself to form the double-membrane autophagosome, engulfing cytoplasmic substrates targeted for macroautophagic degradation (Fig. 1). Transmission electron microscopy suggests that at the leading edge of the PG, the curvature radius may be as little as 10 nanometers. The protein-membrane dynamics driving autophagosome biogenesis remain unclear; however, recognition of unique membrane curvatures by membrane remodeling proteins is well described in membrane biology, and is already implicated in the early stages of autophagosome growth. Because PG structures accumulate when LC3 lipidation is blocked, we investigated whether the lipidation event itself is a curvature-sensitive process.
Figure 1. How limiting LC3 lipidation to curved membranes could impart spatial and temporal specificity. The N-terminal 20 amino acids of mouse and human ATG3 form a membrane curvature-sensing amphipathic helix. As a result, lipidation of the LC3 protein family is strongly favored on highly curved membranes. In vivo, the leading edge of the growing phagophore (PG) exhibits an extreme local curvature that distinguishes the PG spatially from most other intracellular membranes and temporally from the later stages of autophagosome maturation, and thus may constitute a unique intracellular target for ATG3 function.
ATG3 Senses Membrane Curvature
Membrane association of LC3 family proteins is achieved by formation of a covalent bond to phosphatidylethanolamine (PE) through a ubiquitination-like sequence of enzymatic reactions. ATG7 acts as the LC3 activating enzyme, while ATG3 acts as the LC3 conjugating enzyme that transfers LC3 to PE to form LC3–PE/LC3-II. This lipidation reaction can be entirely recapitulated in vitro on liposomes with recombinant proteins. Using this reconstituted system, we observed that lipidation is dramatically improved on membranes harboring lipid packing defects like those that arise with unnaturally high local concentrations of cone-shaped lipids, of which PE is one example.
In eukaryotes, the distribution of lipids is carefully controlled, effectively minimizing the potential for lipid-dependent defects to accumulate. However, similar membrane stresses exist biologically at regions of very high curvature, where the packing of lipids is made suboptimal by the geometric organization of the severely bent bilayer. Indeed, in our reconstituted system, lipidation can also be made very efficient with low densities of PE provided that the underlying membrane is highly curved. The increase in efficiency is not apparent until the radius of curvature is below 50 nm and continues to improve as the curvature becomes more extreme. Thus, the basic components of the lipidation machinery are designed to work optimally on membranes that resemble small vesicles or tubules.
This curvature-dependence is encoded in an N-terminal amphipathic helix comprising the first 20 amino acids of human and mouse ATG3. Mutations reducing or disrupting the hydrophobic face of the helix diminish or abrogate LC3 lipidation, while mutations that expand the hydrophobic face support ATG3 targeting to all membranes independent of curvature, allowing for LC3 lipidation even on a flat membrane surface. As all of the mutants continued to support the formation of an ATG3-LC3 intermediate, we think that the amphipathic organization of the helix is only involved in membrane recognition. Importantly, the helix mutants that disrupt lipidation in vitro also fail to rescue LC3 lipidation in Atg3 knockout cells, suggesting that this curvature sensing is an integral part of the normal functioning of ATG3.
Why Should LC3 Lipidation Occur on Highly Curved Membranes?
Restricting lipidation to the curved edge of the PG would impart both spatial and temporal specificity to the reaction (Fig. 1): 1) Spatially, lipidation would be diminished or inhibited altogether on flat surfaces, mimimizing off-target LC3 accumulation at ectopic membranes. These spatial limits could work in concert with ATG4-mediated delipidation at off-target sites to insure that the capture of autophagic proteins and cargo occur only on bona fide autophagic membranes; 2) temporally, lipidation would be limited to only the PG. After encapsulation is complete, the fission event that closes the structure also eliminates the strident curvature, and thus would prohibit further LC3 deposition on the mature autophagosome. Again, this design could synergize with the ATG4-mediated delipidation that is thought to remove LC3 just before fusion with the lysosome.
There is also the possibility that deposition of LC3 at the rim would be advantageous with respect to LC3 function itself. Indeed, LC3–PE is directly involved in PG growth and closure, as well as autophagic cargo capture. Implicit in a role in closure is a localization to the dilating rim. Likewise, cargo capture must necessarily involve movement of cargo from the open face/rim of the cup into the lumen of the growing organelle. Whether either of these activities would benefit from an initial and perhaps transitory concentration of LC3–PE at the rim during lipidation is unknown.
Perspective
Moving forward, an interesting possibility is that many of the proteins involved in autophagosome growth will recognize and target the curved rim, such that the temporal and spatial specificity proposed in Figure 1 could be of general relevance. Indeed, both ATG14 and the Atg1-Atg17 complexes have already been suggested to rely on curvature sensing. Further, proteins that sense membrane curvature can, in principle, lead to membrane curvature stabilization or even induction if many molecules are implicated. Whereas it is not clear whether ATG3 could induce membrane curvature by itself, the likely presence of other factors at the tip of the growing PG, collectively at high concentration, may participate in a feed-forward mechanism of rim stabilization and expansion, thus allowing for PG growth. How ATG3 interacts with other protein components on the PG also requires additional study. In particular, in our in vitro system, the protein complex ATG12–ATG5-ATG16L1 is not required to efficiently lipidate LC3; however, this complex appears to play a more important role in vivo. ATG12–ATG5-ATG16L1 is proposed to directly activate ATG3 and also scaffold the growing PG, perhaps along the relatively flat convex face of the organelle. Thus, establishing where these 2 protein units assemble is essential. Unraveling how individual proteins are distributed on mammalian autophagosomes remains challenging, but given the recent progress in nanoscale fluorescence microscopy, perhaps the biochemistry occurring at the leading edge of the cup will soon be amenable to real-time observation.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.