Complete set of the Atg8-E1-E2-E3 conjugation machinery forms an interaction web that mediates membrane shaping.

Atg8, a ubiquitin-like protein, is conjugated with phosphatidylethanolamine (PE) via Atg7 (E1), Atg3 (E2) and Atg12-Atg5-Atg16 (E3) enzymatic cascade and mediates autophagy. However, its molecular roles in autophagosome formation are still unclear. Here we show that Saccharomyces cerevisiae Atg8-PE and E1-E2-E3 enzymes together construct a stable, mobile membrane scaffold. The complete scaffold formation induces an in-bud in prolate-shaped giant liposomes, transforming their morphology into one reminiscent of isolation membranes before sealing. In addition to their enzymatic roles in Atg8 lipidation, all three proteins contribute nonenzymatically to membrane scaffolding and shaping. Nuclear magnetic resonance analyses revealed that Atg8, E1, E2 and E3 together form an interaction web through multivalent weak interactions, where the intrinsically disordered regions in Atg3 play a central role. These data suggest that all six Atg proteins in the Atg8 conjugation machinery control membrane shaping during autophagosome formation.

The mitochondrial intermembrane space protein mitofissin drives mitochondrial fission required for mitophagy.

Mitophagy plays an important role in mitochondrial homeostasis by selective degradation of mitochondria. During mitophagy, mitochondria should be fragmented to allow engulfment within autophagosomes, whose capacity is exceeded by the typical mitochondria mass. However, the known mitochondrial fission factors, dynamin-related proteins Dnm1 in yeasts and DNM1L/Drp1 in mammals, are dispensable for mitophagy. Here, we identify Atg44 as a mitochondrial fission factor that is essential for mitophagy in yeasts, and we therefore term Atg44 and its orthologous proteins mitofissin. In mitofissin-deficient cells, a part of the mitochondria is recognized by the mitophagy machinery as cargo but cannot be enwrapped by the autophagosome precursor, the phagophore, due to a lack of mitochondrial fission. Furthermore, we show that mitofissin directly binds to lipid membranes and brings about lipid membrane fragility to facilitate membrane fission. Taken together, we propose that mitofissin acts directly on lipid membranes to drive mitochondrial fission required for mitophagy.

Membrane perturbation by lipidated Atg8 underlies autophagosome biogenesis.

Autophagosome biogenesis is an essential feature of autophagy. Lipidation of Atg8 plays a critical role in this process. Previous in vitro studies identified membrane tethering and hemi-fusion/fusion activities of Atg8, yet definitive roles in autophagosome biogenesis remained controversial. Here, we studied the effect of Atg8 lipidation on membrane structure. Lipidation of Saccharomyces cerevisiae Atg8 on nonspherical giant vesicles induced dramatic vesicle deformation into a sphere with an out-bud. Solution NMR spectroscopy of Atg8 lipidated on nanodiscs identified two aromatic membrane-facing residues that mediate membrane-area expansion and fragmentation of giant vesicles in vitro. These residues also contribute to the in vivo maintenance of fragmented vacuolar morphology under stress in fission yeast, a moonlighting function of Atg8. Furthermore, these aromatic residues are crucial for the formation of a sufficient number of autophagosomes and regulate autophagosome size. Together, these data demonstrate that Atg8 can cause membrane perturbations that underlie efficient autophagosome biogenesis.

In vitro reconstitution of autophagic processes.

Autophagy is a lysosomal degradation system that involves de novo autophagosome formation. A lot of factors are involved in autophagosome formation, including dozens of Atg proteins that form supramolecular complexes, membrane structures including vesicles and organelles, and even membraneless organelles. Because these diverse higher-order structural components cooperate to mediate de novo formation of autophagosomes, it is too complicated to be elaborated only by cell biological approaches. Recent trials to regenerate each step of this phenomenon in vitro have started to elaborate on the molecular mechanisms of such a complicated process by simplification. In this review article, we outline the in vitro reconstitution trials in autophagosome formation, mainly focusing on the reports in the past few years and discussing the molecular mechanisms of autophagosome formation by comparing in vitro and in vivo observations.

Phase separation organizes the site of autophagosome formation.

Many biomolecules undergo liquid-liquid phase separation to form liquid-like condensates that mediate diverse cellular functions1,2. Autophagy is able to degrade such condensates using autophagosomes-double-membrane structures that are synthesized de novo at the pre-autophagosomal structure (PAS) in yeast3-5. Whereas Atg proteins that associate with the PAS have been characterized, the physicochemical and functional properties of the PAS remain unclear owing to its small size and fragility. Here we show that the PAS is in fact a liquid-like condensate of Atg proteins. The autophagy-initiating Atg1 complex undergoes phase separation to form liquid droplets in vitro, and point mutations or phosphorylation that inhibit phase separation impair PAS formation in vivo. In vitro experiments show that Atg1-complex droplets can be tethered to membranes via specific protein-protein interactions, explaining the vacuolar membrane localization of the PAS in vivo. We propose that phase separation has a critical, active role in autophagy, whereby it organizes the autophagy machinery at the PAS.

Liquidity Is a Critical Determinant for Selective Autophagy of Protein Condensates.

Clearance of biomolecular condensates by selective autophagy is thought to play a crucial role in cellular homeostasis. However, the mechanism underlying selective autophagy of condensates and whether liquidity determines a condensate's susceptibility to degradation by autophagy remain unknown. Here, we show that the selective autophagic cargo aminopeptidase I (Ape1) undergoes phase separation to form semi-liquid droplets. The Ape1-specific receptor protein Atg19 localizes to the surface of Ape1 droplets both in vitro and in vivo, with the "floatability" of Atg19 preventing its penetration into droplets. In vitro reconstitution experiments reveal that Atg19 and lipidated Atg8 are necessary and sufficient for selective sequestration of Ape1 droplets by membranes. This sequestration is impaired by mutational solidification of Ape1 droplets or diminished ability of Atg19 to float. Taken together, we propose that cargo liquidity and the presence of sufficient amounts of autophagic receptor on cargo are crucial for selective autophagy of biomolecular condensates.

Membrane-binding domains in autophagy.

Autophagy is an intracellular degradation system conserved among eukaryotes that mediates the degradation of various biomolecules and organelles. During autophagy, a double membrane-bound organelle termed an autophagosome is synthesized de novo and delivers targets from the cytoplasm to the lysosomes for degradation. Autophagosome formation involves complex and dynamic membrane rearrangements, which are regulated by dozens of autophagy-related (Atg) proteins. In this review, we summarize our current knowledge of membrane-binding domains and motifs in Atg proteins and discuss their roles in autophagy.

Elementary processes of antimicrobial peptide PGLa-induced pore formation in lipid bilayers.

Antimicrobial peptide PGLa induces the leakage of intracellular content, leading to its bactericidal activity. However, the elementary process of PGLa-induced leakage remains poorly understood. Here, we examined the interaction of PGLa with lipid bilayers using the single giant unilamellar vesicle (GUV) method. We found that PGLa induced membrane permeation of calcein from GUVs comprised of dioleoylphosphatidylcholine (DOPC) and dioleoylphosphatidylglycerol (DOPG) and its rate increased with time to reach a steady value, indicating that PGLa induced pores in the bilayer. The binding of PGLa to the GUV membrane raised its fractional area change, δ. At high PGLa concentrations, the time course of δ showed a two-step increase; δ increased to a value, δ1, which was constant for an extended period before increasing to another constant value, δ2, that persisted until aspiration of the GUV. To reveal the distribution of PGLa, we investigated the interaction of a mixture of PGLa and carboxyfluorescein (CF) -labeled PGLa (CF-PGLa) with single GUVs. The change of the fluorescence intensity of the GUV rim, I, over time showed a two-step increase from a steady value, I1, to another, I2, concomitant with the entering of CF-PGLa into the lumen of the GUV prior to AF647 leakage. The simultaneous measurement of δ and I indicated that their time courses were virtually the same and the ratios (δ2/δ1 and I2/I1) were almost 2. These results indicated that CF-PGLa translocated across the bilayer before membrane permeation. Based on these results, the elementary processes of the PGLa-induced pore formation were discussed.

Activation Energy of the Low-pH-Induced Lamellar to Bicontinuous Cubic Phase Transition in Dioleoylphosphatidylserine/Monoolein.

Electrostatic interaction is an important factor for phase transitions between lamellar liquid-crystalline (Lα) and inverse bicontinuous cubic (QII) phases. We investigated the effect of temperature on the low-pH-induced Lα to double-diamond cubic (QII(D)) phase transition in dioleoylphosphatidylserine (DOPS)/monoolein (MO) using time-resolved small-angle X-ray scattering with a stopped-flow apparatus. Under all conditions of temperature and pH, the Lα phase was directly transformed into an intermediate inverse hexagonal (HII) phase, and subsequently the HII phase slowly converted to the QII(D) phase. We obtained the rate constants of the initial step (i.e., the Lα to HII phase transition) and of the second step (i.e., the HII to QII(D) phase transition) using the non-negative matrix factorization method. The rate constant of the initial step increased with temperature. By analyzing this result, we obtained the values of its apparent activation energy, Ea (Lα → HII), which did not change with temperature but increased with an increase in pH. In contrast, the rate constant of the second step decreased with temperature at pH 2.6, although it increased with temperature at pH 2.7 and 2.8. These results indicate that the value of Ea (HII → QII(D)) at pH 2.6 increased with temperature, but the values of Ea (HII → QII(D)) at pH 2.7 and 2.8 were constant with temperature. The values of Ea (HII → QII(D)) were smaller than those of Ea (Lα → HII) at the same pH. We analyzed these results using a modified quantitative theory on the activation energy of phase transitions of lipid membranes proposed initially by Squires et al. (Squires, A. M.; Conn, C. E.; Seddon, J. M.; Templer, R. H. Soft Matter 2009, 5, 4773). On the basis of these results, we discuss the mechanism of this phase transition.

Antimicrobial Peptide Lactoferricin B-Induced Rapid Leakage of Internal Contents from Single Giant Unilamellar Vesicles.

Enzymatic digestion of bovine lactoferrin generates lactoferricin B (Lfcin B), a 25-mer peptide with strong antimicrobial activity of unknown mechanism. To elucidate the mechanistic basis of Lfcin B bactericidal activity, we investigated the interaction of Lfcin B with Escherichia coli and liposomes of lipid membranes. Lfcin B induced the influx of a membrane-impermeant fluorescent probe, SYTOX green, from the outside of E. coli into its cytoplasm. Lfcin B induced gradual leakage of calcein from large unilamellar vesicles (LUVs) of dioleoylphosphatidylglycerol (DOPG)/dioleoylphosphatidylcholine (DOPC) membranes. To clarify the cause of Lfcin B-induced leakage of calcein from the LUVs, we used the single giant unilamellar vesicle (GUV) method to investigate the interaction of Lfcin B with calcein-containing DOPG/DOPC-GUVs. We observed that a rapid leakage of calcein from a GUV started stochastically; statistical analysis provided a rate constant for Lfcin B-induced pore formation, kp. On the other hand, phase-contrast microscopic images revealed that Lfcin B induced a rapid leakage of sucrose from the single GUVs with concomitant appearance of a spherical GUV of smaller diameter. Because of the very fast leakage, and at the present time resolution of the experiments (33 ms), we could not follow the evolution of pore nor the process of the structural changes of the GUV. Here we used the term "local rupture" to express the rapid leakage of sucrose and determined the rate constant of local rupture, kL. On the basis of the comparison between kp and kL, we concluded that the leakage of calcein from single GUVs occurred as a result of a local rupture in the GUVs and that smaller pores inducing leakage of calcein were not formed before the local rupture. The results of the effect of the surface charge density of lipid membranes and that of salt concentration in buffer on kp clearly show that kp increases with an increase in the extent of electrostatic interactions due to the surface charges. Analysis of Lfcin B-induced shape changes indicated that the binding of Lfcin B increased the area of the outer monolayer of GUVs. These results indicate that Lfcin B-induced damage of the plasma membrane of E. coli with its concomitant rapid leakage of internal contents is a key factor for the bactericidal activity of LfcinB.

Stretch-activated pore of the antimicrobial peptide, magainin 2.

Antimicrobial peptide magainin 2 forms pores in lipid membranes and induces membrane permeation of the cellular contents. Although this permeation is likely the main cause of its bactericidal activity, the mechanism of pore formation remains poorly understood. We therefore investigated in detail the interaction of magainin 2 with lipid membranes using single giant unilamellar vesicles (GUVs). The binding of magainin 2 to the lipid membrane of GUVs increased the fractional change in the area of the membrane, δ, which was proportional to the surface concentration of magainin 2, X. This indicates that the rate constant of the magainin 2-induced two-state transition from the intact state to the pore state greatly increased with an increase in δ. The tension of a lipid membrane following aspiration of a GUV also activated magainin 2-induced pore formation. To reveal the location of magainin 2, the interaction of carboxyfluorescein (CF)-labeled magainin 2 (CF-magainin 2) with single GUVs containing a water-soluble fluorescent probe, AF647, was investigated using confocal microscopy. In the absence of tension due to aspiration, after the interaction of magainin 2 the fluorescence intensity of the GUV rim due to CF-magainin 2 increased rapidly to a steady value, which remained constant for a long time, and at 4-32 s before the start of leakage of AF647 the rim intensity began to increase rapidly to another steady value. In contrast, in the presence of the tension, no increase in rim intensity just before the start of leakage was observed. These results indicate that magainin 2 cannot translocate from the outer to the inner monolayer until just before pore formation. Based on these results, we conclude that a magainin 2-induced pore is a stretch-activated pore and the stretch of the inner monolayer is a main driving force of the pore formation.

The single GUV method for revealing the functions of antimicrobial, pore-forming toxin, and cell-penetrating peptides or proteins.

We recently developed the single giant unilamellar vesicle (GUV) method for investigating the functions and dynamics of biomembranes. The single GUV method can provide detailed information on the elementary processes of physiological phenomena in biomembranes, such as their rate constants. Here we describe the process of pore formation induced by the antimicrobial peptide (AMP), magainin 2, and the pore-forming toxin (PFT), lysenin, as revealed by the single GUV method. We obtained the rate constants of several elementary steps, such as peptide/protein-induced pore formation in lipid membranes and the membrane permeation of fluorescent probes through the pores. Information on the entry of the cell-penetrating peptide (CPP), transportan 10 (TP10), into a single GUV and its induced pore formation in lipid membranes was also obtained. We compare the single GUV method with other methods for investigating the interaction of peptides/proteins with lipid membranes (i.e., the large unilamellar vesicle (LUV) suspension method, the GUV suspension method, and single channel recording), and discuss the pros and cons of the single GUV method. On the basis of these data, we discuss the advantages of the single GUV method.

Initial step of pH-jump-induced lamellar to bicontinuous cubic phase transition in dioleoylphosphatidylserine/monoolein.

Electrostatic interactions (EI) are an important factor for phase transitions between lamellar liquid-crystalline (L(α)) and inverse bicontinuous cubic (Q(II)) phases. We investigated the low pH-induced L(α) to double-diamond cubic (Q(II)(D)) phase transition in dioleoylphosphatidylserine (DOPS)/monoolein (MO) using time-resolved small-angle X-ray scattering. Using a stopped-flow apparatus, a suspension of liposomes (multilamellar vesicles (MLVs) or large unilamellar vesicles (LUVs)) of 20%-DOPS/80%-MO membrane at neutral pH was rapidly mixed with a low pH buffer, and then the structural change of the membranes in the resultant suspension was observed as a function of time (i.e., pH-jump experiment). At the initial step, the L(α) phase was directly transformed into the hexagonal II (H(II)) phase, and subsequently, the H(II) phase slowly converted into the Q(II)(D) phase. We obtained the rate constants of the initial step (i.e., the L(α) to H(II) phase transition) and of the second step (i.e., the H(II) to Q(II)(D) phase transition) using the non-negative matrix factorization method. The rate constant of the initial step was independent of the MLV concentration, indicating that single MLVs can convert into the HII phase without any interaction with other MLVs. On the other hand, the rate constant of the initial step increased with a decrease in pH, 0.041 s(-1) at pH 2.6 and 0.013 s(-1) at pH 2.8, and also exhibited a size dependence; for smaller vesicles such as LUVs and smaller MLVs with diameters of ~1 µm, the rate constant was smaller. They were reasonably explained by the classical nucleation theory. These results provide the first experimental evidence of the total kinetics of EI-induced L(α)/Q(II) phase transitions.

Entry of cell-penetrating peptide transportan 10 into a single vesicle by translocating across lipid membrane and its induced pores.

The cell-penetrating peptide, transportan 10 (TP10), can translocate across the plasma membrane of living cells and thus can be used for the intracellular delivery of biological cargo such as proteins. However, the mechanisms underlying its translocation and the delivery of large cargo remain unclear. In this report we investigated the entry of TP10 into a single giant unilamellar vesicle (GUV) and the TP10-induced leakage of fluorescent probes using the single GUV method. GUVs of 20% dioleoylphosphatidylglycerol (DOPG)/80% dioleoylphosphatidylcholine (DOPC) were prepared, and they contained a water-soluble fluorescent dye, Alexa Fluor 647 hydrazide (AF647), and smaller vesicles composed of 20% DOPG/80% DOPC. The interaction of carboxyfluorescein (CF)-labeled TP10 (CF-TP10) with these loaded GUVs was investigated using confocal microscopy. The fluorescence intensity of the GUV membrane increased with time to a saturated value, then the fluorescence intensity due to the membranes of the smaller vesicles inside the GUV increased prior to leakage of AF647. This result indicates that CF-TP10 entered the GUV from the outside by translocating across the lipid membrane before CF-TP10-induced pore formation. The rate constant of TP10-induced pore formation in lipid membranes increased with an increase in TP10 concentration. Large molecules such as Texas Red Dextran 40,000, and vesicles with a diameter of 1-2 µm, permeated through the TP10-induced pores or local rupture in the lipid membrane. These results provide the first direct experimental evidence that TP10 can deliver large cargo through lipid membranes, without the need for special transport mechanisms such as those found in cells.

The single-giant unilamellar vesicle method reveals lysenin-induced pore formation in lipid membranes containing sphingomyelin.

Lysenin is a sphingomyelin (SM)-binding pore-forming toxin. To reveal the interaction of lysenin with lipid membranes, we investigated lysenin-induced membrane permeation of a fluorescent probe, calcein, through dioleoylphosphatidylcholine(DOPC)/SM, DOPC/SM/cholesterol(chol), and SM/chol membranes, using the single-giant unilamellar vesicle (GUV) method. The results clearly show that lysenin formed pores in all the membranes, through which membrane permeation of calcein occurred without disruption of GUVs. The membrane permeation began stochastically, and the membrane permeability coefficient increased over time to reach a maximum, steady value, Ps, which persisted for a long time(100--500 s), indicating that the pore concentration increases over time and finally reaches its steady value, NP s . The Ps values increased as the SM/lysenin ratio decreased, and at low concentrations of lysenin, the Ps values of SM/DOPC/chol (42/30/28)GUVs were much larger than those of SM/DOPC (58/42) GUVs. The dependence of Ps on the SM/lysenin ratio for these membranes was almost the same as that of the fraction of sodium dodecyl sulfate (SDS)-resistant lysenin oligomers, indicating that NP s increases as the SDS-resistant oligomer fraction increases. On the other hand, lysenin formed pores in GUVs of SM/chol(60/40) membrane, which is in a homogeneous liquid-ordered phase, indicating that the phase boundary is not necessary for pore formation. The Ps values of SM/chol (60/40) GUVs were smaller than those of SM/DOPC/chol (42/30/28) GUVs even though the SDS-resistant oligomer fractions were similar for both membranes, suggesting that not all of the oligomers can convert into a pore. On the basis of these results, we discuss the elementary processes of lysenin-induced pore formation.

Spontaneous insertion of lipopolysaccharide into lipid membranes from aqueous solution.

Lipopolysaccharide (LPS), one of the main components of outer membranes of Gram-negative bacteria, consists of a hydrophobic lipid (lipid A) with six hydrocarbon chains and a large hydrophilic polysaccharide chain. LPS plays endotoxic roles and can stimulate macrophages and B cells. To elucidate the mechanism of the interaction of LPS with various cell membranes, it is important to investigate the interaction of wild type LPS in a buffer with lipid membranes. In this report we investigated the interaction of low concentrations of LPS in a buffer with giant unilamellar vesicles (GUVs) of dioleoylphosphatidylcholine (DOPC) membrane in the liquid-crystalline (L(α)) phase and sphingomyelin (SM)/cholesterol(chol) (molar ration; 6/4) membrane in the liquid-ordered (lo) phase. We found that low concentrations (less than critical micelle concentration) of LPS in aqueous solution induced the shape changes such as the transformation from a prolate to a two-spheres-connected by a very narrow neck in the DOPC-GUVs and also in the SM/chol (6/4)-GUVs above their threshold concentrations. The analysis of the shape changes of the GUVs indicates that the monomers of LPS can insert spontaneously into the external monolayer of the lipid membranes of these GUVs from the aqueous solution. Moreover, higher concentrations of LPS induced the vesicle fission of SM/chol(6/4)-GUVs above its higher threshold concentration. The vesicle fission of GUVs is similar to those induced by single long chain amphiphiles such as lysophosphatidylcholine. On the basis of these results, we discuss the interaction of wild type LPS with lipid membranes and cell membranes. These results suggest that LPS molecules can insert spontaneously into the external monolayer of the plasma membranes composed of the L(α) phase-membrane and the microdomain in the lo phase.