Lipids form membranes that delimit cellular compartments, but they also can act as signalling molecules that recruit proteins to membranes. The chemical and compositional diversity of lipids help to establish the identity of intracellular compartments and the nature of subdomains within these compartments. For example, the endoplasmic reticulum (ER) contains low levels of cholesterol and high levels of unsaturated glycerophospholipids, whereas the plasma membrane has an abundance of cholesterol and saturated sphingolipids and glycerophospholipids (Bigay and Antonny, 2012). Differences in lipid composition give membranes their physicochemical characteristics, which can control the activities of membrane-associated enzymes. The lipid phosphatidylinositol 3-phosphate (PI(3)P) plays important roles in the regulation of various intracellular events including autophagy, endocytosis, phagocytosis, and vacuolar protein sorting. In mammals, PI(3)P is generated mainly by the Class III phosphatidylinositol 3-kinase VPS34/PIK3C3, which phosphorylates the 3-OH position of phosphatidylinositol (PI). However, in cells, VPS34 is not found alone. Instead, it forms two principal complexes, complexes I and II. Both complexes have core components VPS34, VPS15 and Beclin 1 (Vps30 in yeast) bound to a specific component, which is either ATG14L (Atg14 in yeast) in complex I or UVRAG (Vps38 in yeast) in complex II. This single subunit difference between these two complexes localizes them to different membrane compartments. Complex I is active on autophagosomes (Matsunaga et al., 2009; Sun et al., 2008; Itakura et al., 2008) and some ER compartments (Hamasaki et al., 2013; Nascimbeni et al., 2017; Axe et al., 2008), in response to starvation. In contrast, complex II is mainly active in endocytic pathways by partially co-localizing in mammalian cells with Rab5- and Rab7-positive endolysosomal compartments as well as with Rab9-positive compartments important for endosome to TGN sorting (Itakura et al., 2008). As the membranes of these different compartments vary, we have undertaken to determine how lipid composition affects the lipid kinase activity of VPS34 complexes.

The PI(3)P generated by complex I appears in puncta at ER-mitochondria contact sites, at ER-plasma membrane contact sites and on ER sites known as omegasomes, which are recognized by PI(3)P-specific binding proteins such as DFCP (Nascimbeni et al., 2017; Prinz et al., 2020; Hamasaki et al., 2013; Axe et al., 2008). These findings indicate that complex I is recruited to specific sites on membranes for its activation. This recruitment involves an upstream complex, the ULK1 complex (Atg1 complex in yeast) (Russell et al., 2013). However, how complex I itself associates with membranes and is activated on them is unclear. Complex II recruitment to early endosomes and activation is aided by Rab5 (Christoforidis et al., 1999; Murray et al., 2002), although the mechanism by which this occurs is poorly characterized. PI(3)P produced by complex II, in turn, leads to recruitment of various PI(3)P binding proteins that function in endocytic pathways such as Hrs, EEA1, and sorting nexins (Gaullier et al., 1999; Xu et al., 2001). We have previously shown that yeast complex II has a Y-shaped structure, with the one arm bearing Vps34 and Vps15 acting as the catalytic arm, while the other arm bearing Vps30 and Vps38 is the adaptor arm (Rostislavleva et al., 2015). A similar structural organization has been observed for human complexes I and II (Young et al., 2019; Baskaran et al., 2014; Ma et al., 2017). The aromatic finger motif in the BARA domain of Vps30 is important for membrane binding and kinase activity of complex II, shedding light on the importance of the adaptor arm for membrane binding and kinase activity. Many previous in vitro studies of Class III PI3K have used pure PI (Fassy et al., 2017; Lu et al., 2014; Munson and Ganley, 2016; Russell et al., 2013; Zhong et al., 2009; Kim et al., 2013) or a lipid mixture of only PI and PS as the substrate (Brier et al., 2019). This approach underestimates the effects that membrane complexity could have on VPS34 activity. For this reason, we have devised alternative VPS34 assays to measure the activities of Class III PI3Ks on membranes whose character we can control.

Protein binding to membranes is affected by three important physicochemical parameters: membrane electrostatics, lipid packing, and membrane curvature (Bigay and Antonny, 2012). The negative charge of intracellular membranes arises mainly from phosphoinositides and phosphatidylserine (PS). PS is known to enhance membrane-protein interactions by binding to basic residues of proteins (Xu et al., 2013; Yeung et al., 2009; Yeung et al., 2006; Zhu et al., 1999), whereas protein affinity for a phosphoinositide may depend on both the net charge (the number of phosphates on the inositol ring) or stereochemistry of the phosphoinositide.

Lipid packing is determined by the shape of polar lipid headgroups and by saturation of acyl chains. Unsaturated acyl chains only loosely pack into membranes due to the kinks introduced by unsaturated bond(s), whereas completely saturated acyl chains provide a more uniform geometry and can therefore pack tightly to restrict membrane flexibility (Bigay and Antonny, 2012). Packing also depends on global membrane properties, such as stretching/compression, membrane curvature, or membrane stress (Johner et al., 2014; Marsh, 2007; McMahon and Boucrot, 2015; Bigay and Antonny, 2012). Membrane flexibility is important for processes such as vesiculation, endocytosis and phagocytosis (Calder et al., 1990; Degreif et al., 2019; Jackson, 2018; Schroit and Gallily, 1979; Manni et al., 2018). Phospholipids usually have a saturated fatty acyl chain at position sn1 of the glycerophospholipid, while the sn2 fatty acyl chain can be saturated, monounsaturated or polyunsaturated. Phospholipids with two polyunsaturated fatty acyl chains make membranes more flexible, but very permeable, while the more common sn1-saturated-sn2-polyunsaturated enable flexibility with low membrane permeability (Manni et al., 2018). Acyl chain saturation status also influences membrane curvature. Unsaturated lipids can tolerate high membrane curvature, whereas the shapes of saturated lipids are cylindrical, making membranes containing them flatter. ER membranes are enriched with unsaturated phospholipids, enabling high membrane curvature. This facilitates protein secretion from the ER. In contrast, PM membranes are more tightly packed, due to the abundance of saturated phospholipid, which creates a thick boundary between the cell and its environment (Holthuis and Menon, 2014; Bigay and Antonny, 2012).

Here, we have developed a purely in vitro system with purified VPS34 complexes I and II in which we can directly control membrane bilayer properties, including acyl chain saturation of both substrate and non-substrate lipids, membrane charge and membrane curvature. By combining HDX-MS with mutagenesis and lipid kinase assays, we have investigated which domains and motifs of complex I and complex II are important for membrane binding and kinase activity, and which physicochemical parameters of membranes have the greatest impact on complexes I and II.

To assess how membrane properties modify the activities of VPS34 complexes, we have developed a quantitative confocal microscopy assay to measure the activity of the human Class III PI3K complexes on giant unilamellar vesicles (GUVs), which have a radius of ~ 1–30 µm. Purified, recombinant, monodisperse VPS34 complexes I or II (Figure 1A) were incubated with Lissamine-Rhodamine labelled GUVs containing PI and an AlexaFluor647-labelled PX domain, which specifically binds PI(3)P (Ellson et al., 2001; Figure 1B). By following the rate of recruitment of the PX domain to the GUVs, we can estimate the PI(3)P production and compare the relative activities of the complexes on a range of membrane types.

Figure 1 with 3 supplements see all

Download asset Open asset

Both substrate and non-substrate lipid unsaturation as well as membrane curvature greatly increase VPS34 activities

We examined how lipid packing affects VPS34 activities by varying the saturation status of the acyl chains. For each complex, we compared VPS34 activity on GUVs composed of different percentages of phospholipids with either stearoyl (18:0) or oleoyl (18:1) acyl chains (Figure 2A). As stearoyl acyl chains are saturated, they cause a tighter lipid packing than the mono-unsaturated oleoyl chains. Whereas SO lipids are made with one stearoyl and one oleoyl acyl chain, DO lipids carry two oleoyl acyl chains. Three different lipid compositions were examined: 82% DO lipids (18:1-18:1), 82% SO lipids (18:0-18:1), and 55% SO with 27% DO lipids (55% SO+27% DO) (Figure 2B). In each case, we varied only the acyl chain saturation of the non-substrate lipids while we kept the source of the PI and the relative proportions of PI, PS, PE and PC the same. GUV assays revealed that complexes I and II were only active on 82% DO lipids and not on 55% SO+27% DO or 82% SO (Figure 2C and D), suggesting that loose lipid packing activates complexes I and II. However, Z-stack analysis of ten Z slices, which covers a range of vesicle sizes revealed some activity on smaller vesicles with 55% SO+27% DO, but not on vesicles with 82% SO (Figure 2E and F). To examine this further, the curvatures (1/radius) for each GUV mixture were plotted against their fluorescence intensities. GUVs with 55% SO+27% DO displayed a positive correlation between increasing membrane curvature and activity for both complexes I and II, but 82% SO GUVs did not exhibit this correlation (Figure 2G and H). This suggests that high membrane curvature can at least partially compensate for the negative effect of lipid saturation.

Figure 2

Download asset Open asset

We also examined the effect of lipid saturation of the substrate PI, while not changing the surrounding non-substrate lipids made of DO (Figure 3A). Four PI species were examined: DSPI (18:0-18:0), SAPI (18:0-20:4), DOPI (18:1-18:1) and the mixed chain PI (Figure 3A). GUV assays with complex I showed that DOPI was the best substrate, followed by the mixed chain PI and SAPI. However, with the more saturated acyl chains of DSPI, the activity of complex I was greatly reduced (Figure 3B).

Figure 3

Download asset Open asset

As a control, we tested whether lipid packing could affect binding of the AF647-PX reporter domain to PI(3)P. We found that GUVs with 3% DOPI(3)P in 82% SO lipids showed a similar PX binding as the GUVs with 3% DOPI(3)P in 82% DO lipids (Figure 3C), indicating that membrane packing had little effect on PX binding. The binding of the PX reporter to DSPI(3)P and SOPI(3)P containing GUVs was not tested, since there is no commercial source of these lipids.

Together, these results show that lipid packing affects the complexes I and II activities, with both complexes showing a dramatic preference for unsaturated lipids, in both the general background lipids of the membrane and in the PI substrate for the enzyme. They also show that the negative impact of tight lipid packing on activities of VPS34 complexes can be partially compensated by increased membrane curvature.

Increasing negative membrane charge activates complex I, complex II, and VPS34 alone

In order to examine the effect of negative membrane charge on VPS34 activity, we made membranes with increasing proportions of the negatively charged lipid PS. We compared activities on GUVs with 10% PS (DO base lipids) and 25% PS (DO high PS lipids) (Figure 4A and B). Using VPS34 alone, we could see no activity on GUVs with 10% PS. However, on GUVs with 25% PS, VPS34 activity was clearly detectable (Figure 4C). Also, complexes I and II showed substantial activation on 25% PS compared with 10% PS GUVs (Figure 4D and E). As shown before, the activity of complex II on 10% PS (DO base lipids) was lower than that of complex I (Figures 1D and 4F). However, when the two complexes were compared on 25% PS (DO high PS), complexes I and II had nearly the same activity, suggesting a higher preference of complex II for negative charge (Figure 4G).

Figure 4

Download asset Open asset

Membranes influence structures and dynamics of the VPS34 complexes

We used hydrogen-deuterium exchange mass-spectrometry (HDX-MS) to determine the regions in the complexes I and II involved in membrane interaction. The advantage of HDX-MS is that it can map direct interactions of the native, unlabelled protein with membranes as well as mapping conformational changes associated with membrane binding (Masson et al., 2016).

The HDX-MS experiment for complex I had a sequence coverage of 87% (Supplementary file 4). Several regions in complex I showed a reduction in solvent exchange in the presence of 100 nm LUVs (Figure 5). The most pronounced HDX reduction was in a region of the BATS domain of ATG14L (residues 478–492), with additional changes in Beclin 1 and VPS34. In Beclin 1, three regions show protection in the presence of membranes, one in coiled-coil II (residues 201–212), another in the CC2/BARA linker (residues 262–274) and the third in a region that has been referred to as the BARA domain aromatic finger (AF1, residues 343–359 Huang et al., 2012). In the VPS34 kinase domain, helices k α1 and k α2 and the loop connecting them (residues 558–573) had a reduction in HDX. When the protected areas were overlaid onto a model of human complex I, the AF1 of Beclin 1, and the k α1-k α2 of VPS34 delineate a putative membrane-binding surface (Figure 5B). Although there is no structure for the BATS domain, this region was shown to be involved in membrane binding (Fan et al., 2011).

Figure 5

Download asset Open asset

We also performed HDX-MS for complex II in the presence and absence of 100 nm LUVs (Figure 6 and Supplementary file 5). We see prominent protection in the presence of membranes in Beclin 1 and VPS34. The Beclin 1 CC2/BARA linker had decreased HDX in the presence of membranes in complex II similarly to complex I. Furthermore, as in complex I, the aromatic finger (AF1) of the Beclin 1 BARA domain showed decreased HDX in complex II (residues 358–359). Surprisingly, two additional regions, 288–300 and 409–419, were picked up in Beclin 1 at the putative membrane interface. We will refer to 288–300 as the hydrophobic loop (HL) and 409–419 as the aromatic finger 2 (AF2). In VPS34 (in the context of complex II), a peptide (874-884) from the C-terminal helix of VPS34 catalytic subunit showed protection. We had previously shown that the C-terminal helix is critical for catalysis on membranes (Miller et al., 2010). Although both complexes I and II show protection in the same domains, Beclin 1 BARA and VPS34 kinase, there are differences in the peptides protected that suggest that the two complexes assume similar but distinct orientations on membranes.

Figure 6

Download asset Open asset

The ATG14L BATS domain is critical for the activity of complex I on membranes

Although it has been shown that the BATS domain is important for membrane binding, it is not clear how membrane properties affect this interaction and complex I activity (Brier et al., 2019; Fan et al., 2011; Ma et al., 2017). Sequence analysis suggests that the BATS domain has an amphipathic α-helix referred to as the amphipathic lipid packing sensor (ALPS) (residues 471–492), and hydrophobic residues 484W, 485F, and 488Y within this motif are important for the binding of the BATS domain to SUVs and for the localization of ATG14L to autophagosomes (Fan et al., 2011). Using the GUV assay, we compared activities of complex I carrying a BATS mutation in which the ALPS motif was truncated (ΔALPS 471–492) with the wild-type complex I (WT) (Figure 7A). On DO base lipids (Figure 7B), the ΔALPS mutant showed no activity, suggesting that the motif is essential for the complex I activity on membranes. Even on GUVs made with DO high PS, the ΔALPS mutant of complex I showed hardly any detectable activity (Figure 7C). Surprisingly, complex I ΔALPS displayed significantly less activity than VPS34 alone on GUVs containing 25% PS (Figure 7C). This suggests that the non-catalytic subunits of complex I partially inhibit basal VPS34 activity in the absence of the ALPS motif. To determine whether the increase in activity conferred on complex I by the BATS domain is an intrinsic property of the domain, we fused the BATS domain to the C-terminus of the full-length UVRAG subunit and co-expressed it with other complex II subunits (Figure 7A). This fusion complex was shown previously to increase membrane binding of complex II (Ma et al., 2017). We found that the fusion complex II (UVRAG+BATS) showed increased activity, about 7-fold compared with the wild-type complex II (CII WT, Figure 7D). The human UVRAG has a 235 amino-acid long C-terminal extension compared to its yeast orthologue Vps38. When this UVRAG-specific extension was deleted and the BATS domain was fused to the truncated UVRAG, the resulting complex II variant (UVRAG ∆C+BATS, Figure 7A) showed an even higher activation of 11-fold over the WT complex II (Figure 7D). This suggests that the presence or absence of the UVRAG C-terminal region alters the orientation of the BATS domain with respect to the membrane. Interestingly, both UVRAG+BATS and UVRAG ∆C+BATS exhibited even higher activity than complex I WT (Figure 7D). Our activity results are consistent with the previous studies showing that the presence of the BATS domain increased membrane binding (Ma et al., 2017). Together with our HDX-MS results, this strongly indicates that the BATS domain activates complex I through a mechanism that involves an increase in membrane binding, and that the BATS domain can act as an autonomous membrane-interacting module that confers gain of function to VPS34 activity.

Figure 7

Download asset Open asset

The Beclin 1 BARA domain is more important for the activity of complex II than complex I

Three motifs, aromatic finger 1 (AF1, 359-FFW-361), aromatic finger 2 (AF2, 418-QF-419), and the hydrophobic loop motif (HL, 293-LPSVPV-298) found in the BARA domain of Beclin 1 are part of the putative membrane interface (Figure 8A). HDX-MS showed that membrane binding decreased HDX for only AF1 in complex I, whereas HL, AF1 and AF2 had reduced HDX in complex II (Figure 8A). This suggests that the BARA domain might bind membranes differently in the context of complex I and complex II (Figure 8A, middle panel). We measured the activities of both complexes I and II when carrying mutations in these motifs (AFM1 = F359D F360D W360D, AFM2 = Q418D F419D, and HLM = L293D V296D) in order to examine their importance. This revealed that complex I carrying AFM1 or AFM1+HLM (both AFM1 and HLM) had reduced but still significant activities (Figure 8B), while AFM2 did not affect the complex I activity (Figure 8C). In contrast, complex II activity depended on all three motifs (Figure 8D). This suggests that complex I activity is more dependent on the BATS domain than the BARA domain, with HL and AF2 dispensable for complex I. On the other hand, complex II activity is critically dependent on the BARA domain, where AF1, AF2 and HL are all essential for full activity.

Figure 8

Download asset Open asset

Phosphoinositides have modest effects on activities of complexes I and II

Phosphoinositides (PIPs) represent about 10% of total phospholipids, and less than 1% of total lipids in biological membranes (De Craene et al., 2017; Payrastre et al., 2001). However, specific PIPs are enriched in distinct organelle membranes, and this helps to selectively recruit proteins. Since previous work established the importance of kinases that generate phosphatidylinositol 4-phosphate (PI(4)P) and phosphatidylinositol 4,5-bisphosphate (PI(4,5)P₂) in autophagy (Wang et al., 2015; Tan et al., 2016), we examined the influence of PI(4)P and PI(4,5)P₂ on complexes I and II. For this, we added either 5% DOPI(4)P or 5% DOPI(4,5)P₂ in place of 5% DOPC to the DO base lipids (Figure 9A, B and C). We found that complex I was activated by PI(4)P (~1.8 fold) and by PI(4,5)P₂ (~1.2 fold) compared to DO base lipids (Figure 9D and E). In contrast, complex II was also modestly activated by PI(4)P (~1.7 fold) (Figure 9F) but unaffected by PI(4,5)P₂ (Figure 9G). The activation of complex I by PI(4,5)P₂ is consistent with the observation that the isolated BATS domain of ATG14L has a binding preference for PI(4,5)P₂, in addition to its ability to sense membrane curvature (Fan et al., 2011; Tan et al., 2016).

Figure 9

Download asset Open asset

In this study, we systematically examined the importance of three physicochemical membrane parameters on the lipid kinase activities, structures and dynamics of VPS34 complexes I and II. We found that all three parameters, packing defects caused by acyl chain unsaturation for both substrate and non-substrate lipids, membrane curvature and electrostatics, significantly affected the activity of both complexes. It was surprising that changing the saturation status on just one of the two acyl chains dramatically altered the activity of both complexes. Variations in acyl chain composition of glycerophospholipids can give rise to thousands of lipid species. Nevertheless, mammalian cells have some distinct patterns of acyl chain enrichments. One notable example of this is that 40–85% of PI is present as SAPI (18:0/20:4) (Barneda et al., 2019; Blunsom and Cockcroft, 2020; Bozelli and Epand, 2019). We showed that VPS34 kinase is activated by unsaturation of its substrate PI (Figure 3B). This is consistent with the observation in C. elegans that depletion of an enzyme that incorporates polyunsaturated fatty acids (PUFAs) in the sn-2 position of PI significantly decreased PI(3)P content and size of early endosomes, without reducing total PI (Lee et al., 2008; Lee et al., 2012). More surprising was our result that increasing unsaturation of the background lipids also dramatically increased activities of complexes I and II (Figure 2). The effect of background lipids unsaturation is likely to be mediated largely by the adaptor arm, since VPS34 alone shows no activity on DO base (10% PS) lipids (Figure 4C), whereas complexes I and II on the same lipid mixture are active (Figure 4D and E). Stearoyl-CoA desaturase activity is important for early steps in autophagy (Ogasawara et al., 2014; Köhler et al., 2009), and recent lipidomic analyses have shown that autophagic membranes of budding yeast contain as much as 60% lipids with two unsaturated fatty acyl moieties, and only 2% with no double bonds (Schütter et al., 2020). The potent effect of lipid unsaturation on VPS34 complexes combined with the compartment-specific lipid unsaturation suggests an important means of modulating VPS34 activity on cellular organelles. The importance of lipid unsaturation for VPS34 activity is analogous to what was reported for another lipid kinase involved in the generating polyphosphoinositides. Increasing acyl chain unsaturation for substrate and non-substrate lipids in detergent micelles activates phosphatidylinositol-4-phosphate 5-kinase isoforms (Shulga et al., 2012).

The human VPS34 complexes were activated by increased membrane curvature (Figure 2), which is consistent with results for the yeast complexes I and II (Rostislavleva et al., 2015) and human complexes I and II (Brier et al., 2019). An alternative possibility that cannot be excluded is that smaller GUVs incorporate more unsaturated lipid during their formation and hence are able to recruit the two complexes more efficiently. Membrane curvature causes tilting of lipids and decreased lipid packing, thereby opening hydrophobic cavities that can be sensed by protein motifs like the ALPS motif from complex I (Vanni et al., 2014) and motifs from the BARA domain of both complexes I and II. Although the VPS34 complexes make protein/protein interactions that help localize them to specific lipid compartments, the preference for high curvature and lipid unsaturation may contribute to the lack of activity of these PI3K complexes at the PM, which is enriched with saturated lipids and has a lower curvature (Bigay and Antonny, 2012). Rab5 is a primary determinant for recruiting complex II to early endosomes. Interestingly, like complex II, Rab5 has a preference for target membranes with high curvature and monounsaturated lipids, and this preference derives from the unsaturated prenyl lipids attached to Rab5 fitting into packing defects in the target membrane (Kulakowski et al., 2018). Curvature and lipid packing are closely linked, consequently, the loss of VPS34 activity caused by tight lipid packing of saturated lipids can be counteracted by an increase in membrane curvature (Figure 2).

As the third physicochemical parameter, we tested the effects of electrostatics on the activities of VPS34 complexes by increasing the concentration of phosphatidylserine (PS). PS is enriched on the cytosolic leaflet of the plasma membrane (PM) and endocytic vesicles mainly at recycling endosomes (RE) (Fairn et al., 2011). The PS on the cytosolic leaflet plays important roles in the regulation of endocytic pathways (Uchida et al., 2011). We found that PS activates complexes I and II, and also VPS34 alone (Figure 4C–E). Significantly, GUVs with 25% PS enabled complex II to achieve the same activity as complex I (Figure 4G), whereas on membranes with 10% PS, complex I was much more active than complex II (Figure 4F). The ability of high PS content to strongly activate complex II might contribute to complex II activity on early endosomes, which have more PS than the ER (Hullin-Matsuda et al., 2014). It is reasonable to hypothesize that PS, together with Rab5, may contribute to activation of complex II on early endosomes. Increased PS in combination with high curvature and loosely packed lipids may contribute also to activation of complex I during autophagy. In cells, PS is synthesized by PS synthases (PSS) I and II at the ER-mitochondria-associated membranes (MAM) (Lagace and Ridgway, 2013). Upon starvation, complex I is recruited to the MAM (Hamasaki et al., 2013). The PSS I co-localizes with FIP200, a member of autophagy-initiation complex, which is necessary for the recruitment of complex I (Nishimura et al., 2017). ER membranes are enriched in loosely packed lipids, and the site of initiation of the phagophore is a highly curved structure known as an omegasome (Axe et al., 2008). This would provide an ideal platform to activate complex I. However, membrane properties alone would not explain the proclivity of complex I to provide PI3K activity in autophagy compared with complex II. Additional interactions could contribute to enrichment of complex I on autophagosomes. Recently, it was shown that the ATG14L BATS domain may be acting as a coincidence detector, using its ALPS motif to interact with membranes and its LIR motif to interact with autophagic GAPARAPs (Birgisdottir et al., 2019). Along with the ULK1 complex, they may contribute to autophagosome-specific localization and activation.

Both the HDX-MS results and the mutational analyses suggest that complexes I and II have some complex-specific as well as common interactions with membranes. On GUVs made with DO base lipids, human complex I is considerably more active than complex II, which stands in contrast to our results for yeast VPS34 complexes, where complex II was more active than complex I (Rostislavleva et al., 2015). The contrasting behaviour of human and yeast VPS34 complexes likely arises from extensive sequence differences between human and yeast Atg14, particularly in the region corresponding to the C-terminal BATS of ATG14L. The complex I BATS domain interaction with membranes is the most striking of the complex-specific membrane interactions in the human VPS34 complexes. The BATS dominates the complex I membrane interaction. When the BATS domain is deleted, almost all of complex I activity is eliminated (Figure 7). Furthermore, complex I binds to membranes more strongly than complex II (Figure 1D and Ma et al., 2017). Adding the BATS domain to either full-length UVRAG or a UVRAG fragment lacking the C-terminus dramatically increased complex II activity (Figure 7D) and membrane binding in vitro and in cells (Ma et al., 2017). Our results suggest that the differences in activities for complexes I and II reflect a lower affinity of complex II for lipid vesicles (Figure 1C and D). In cells, this weak membrane affinity of complex II may be compensated by its binding partners, such as Rab5, which help to recruit complex II to membranes (Christoforidis et al., 1999). Our results also show that the activities of complexes I and II become equalized on lipid vesicles with increased PS, and this is consistent with a previous report that complexes I and II have identical activities on LUVs formed with 50%PS and 50%PI (Brier et al., 2019).

To our surprise, the ALPS deletion mutant showed even lower activity than VPS34 alone (Figure 7C), suggesting that the influence of the ATG14L subunit is intrinsically inhibitory, but that this inhibition is overcome by the much stronger membrane binding conferred by the BATS domain, resulting in an overall gain of function. A similar mechanism has been reported for the H-Ras-PI3Kα complex, where H-Ras binding is inhibitory for PI3Kα, but the inhibition is overcome by increased recruitment of PI3Kα to membranes (Buckles et al., 2017). The presence of the ATG14L subunit alters the interactions of the rest of the complex with membranes. This subunit appears to enhance the association of the VPS34 kinase domain with membranes (giving rise to protection of the kα1-kα2 region of VPS34; Figure 5). It also alters the interaction of Beclin 1 with membranes, so that while the BARA domain AF1 motif has a role in membrane binding for both complexes, AF2 and HL motifs are important only for complex II (Figures 5, 6 and 8). However, in the context of the dominant role of the BATS domain on complex I activity, the influence of the Beclin 1 AF1 on activity is minor, and there is no significant influence of AF2 or HL (Figure 8). This is consistent with the previous report that mutation of AF1 only mildly affects the autophagosome generation (Huang et al., 2012). It may be that the BATS domain affects the orientation of complex I, so that AF2 and HL no longer interact with membranes. In contrast to complex I, complex II depends critically on AF1, AF2 and HL to enable its maximal activity on membranes (Figure 8). The yeast complex II activity also depends on AF1 (Rostislavleva et al., 2015), indicating that the activation mechanisms of these two orthologous complexes are similar. The HL motif in the BARA domain (Figures 6 and 8) has S295 that can be phosphorylated by Akt (Wang et al., 2012), which enhances interactions of Beclin 1 with 14-3-3 and vimentin intermediate filament proteins. On the other hand, the S295A mutation enhances autophagy (Wang et al., 2012). In cells, the S295 phosphorylation by Akt and/or interaction with 14-3-3 may negatively affect membrane binding of complexes I and II.

In addition to the AF1 motif in the Beclin 1 BARA domain, another region showing decreased HDX in both complexes I and II is the CC2/BARA linker of Beclin 1 (residues 262–274 in complex I and residues 262–270 in complex II). It was proposed that hydrophobic residues Phe270 and Phe274 in this linker interact with lipid membranes (Chang et al., 2019). Consistent with this, mutation of Phe270 and Phe274 eliminated membrane binding and activity of complex II (Chang et al., 2019). The somewhat distant position of these two residues from the putative membrane-binding surface marked by AF1, prompted the proposal that the adjacent BARA β1 and β2 strands become unlatched to allow Phe270/Phe274 to reach the membrane. Although these two hydrophobic residues were protected from HDX in the presence of lipid membranes in our studies (Figures 5 and 6), the peptides covering the β1-β2 of the BARA domain β-sheet 1 (residues 275–287, 276–294, Supplementary files 4 and 5) did not show significant HDX changes. In addition, in yeast, membranes cause this region to increase in HDX (residues 312–327 in Vps30) (Rostislavleva et al., 2015). Consequently, the question whether the CC2/BARA linker becomes more protected due to direct interaction with membranes or due to an indirect ordering in the presence of membranes remains an open question.

Both complexes I and II are activated by PI(4)P, This phosphoinositide is enriched on the Golgi apparatus (Hammond and Balla, 2015), early endosomes (Henmi et al., 2016), and autophagosomes (Judith et al., 2019; Wang et al., 2015), where complexes I and II are activated. The coincidence of Rab5, PS, and PI(4)P at early endosomes might help to specifically enhance complex II activity on this compartment. (Judith et al., 2019; Munson et al., 2015; Sridhar et al., 2013; Wang et al., 2015) We also found that PI(4,5)P₂ modestly increases activity of complex I (Figure 9E). This could be mediated by binding of PI(4,5)P₂ to the ATG14L BATS domain (Wang et al., 2015; Tan et al., 2016). Interestingly, it was recently shown that PIPKIγi5, which phosphorylates PI(4)P to produce PI(4,5)P₂, binds to ATG14L and localizes to autophagosomes (Tan et al., 2016). Compared with the effects of the three membrane physicochemical parameters, phosphoinositides PI(4)P and PI(4,5)P₂ have only modest effects in vitro.

Overall, our results uncovered the importance of lipid packing dictated by acyl chain unsaturation of both substrate and background lipids, negative charge, and membrane curvature for the activation of both complexes I and II. However, these complexes have unique cellular localization and HDX-MS analysis revealed differences in membrane binding modes between complexes I and II that are likely to contribute to their different roles.

All data generated or analysed during this study are included in the manuscript and supporting files.

1. 1. Arumugam S
   2. Kaur A

   (2017) The lipids of the early endosomes: making multimodality work

   ChemBioChem 18:1053–1060.

   https://doi.org/10.1002/cbic.201700046

   • PubMed
   • Google Scholar
2. 1. Axe EL
   2. Walker SA
   3. Manifava M
   4. Chandra P
   5. Roderick HL
   6. Habermann A
   7. Griffiths G
   8. Ktistakis NT

   (2008) Autophagosome formation from membrane compartments enriched in phosphatidylinositol 3-phosphate and dynamically connected to the endoplasmic reticulum

   Journal of Cell Biology 182:685–701.

   https://doi.org/10.1083/jcb.200803137

   • PubMed
   • Google Scholar
3. 1. Barneda D
   2. Cosulich S
   3. Stephens L
   4. Hawkins P

   (2019) How is the acyl chain composition of phosphoinositides created and does it matter?

   Biochemical Society Transactions 47:1291–1305.

   https://doi.org/10.1042/BST20190205

   • PubMed
   • Google Scholar
4. 1. Baskaran S
   2. Carlson L-A
   3. Stjepanovic G
   4. Young LN
   5. Kim DJ
   6. Grob P
   7. Stanley RE
   8. Nogales E
   9. Hurley JH

   (2014) Architecture and dynamics of the autophagic phosphatidylinositol 3-kinase complex

   eLife 3:ee05115.

   https://doi.org/10.7554/eLife.05115

   • Google Scholar
5. 1. Bigay J
   2. Antonny B

   (2012) Curvature, lipid packing, and electrostatics of membrane organelles: defining cellular territories in determining specificity

   Developmental Cell 23:886–895.

   https://doi.org/10.1016/j.devcel.2012.10.009

   • PubMed
   • Google Scholar
6. 1. Birgisdottir ÅB
   2. Mouilleron S
   3. Bhujabal Z
   4. Wirth M
   5. Sjøttem E
   6. Evjen G
   7. Zhang W
   8. Lee R
   9. O'Reilly N
   10. Tooze SA
   11. Lamark T
   12. Johansen T

   (2019) Members of the autophagy class III phosphatidylinositol 3-kinase complex I interact with GABARAP and GABARAPL1 via LIR motifs

   Autophagy 15:1333–1355.

   https://doi.org/10.1080/15548627.2019.1581009

   • PubMed
   • Google Scholar
7. 1. Bissig C
   2. Gruenberg J

   (2013) Lipid sorting and multivesicular endosome biogenesis

   Cold Spring Harbor Perspectives in Biology 5:a016816.

   https://doi.org/10.1101/cshperspect.a016816

   • PubMed
   • Google Scholar
8. 1. Blunsom NJ
   2. Cockcroft S

   (2020) Phosphatidylinositol synthesis at the endoplasmic reticulum

   Biochimica Et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids 1865:158471.

   https://doi.org/10.1016/j.bbalip.2019.05.015

   • Google Scholar
9. 1. Bozelli JC
   2. Epand RM

   (2019) Specificity of acyl chain composition of phosphatidylinositols

   Proteomics 19:1900138.

   https://doi.org/10.1002/pmic.201900138

   • PubMed
   • Google Scholar
10. 1. Brier LW
    2. Ge L
    3. Stjepanovic G
    4. Thelen AM
    5. Hurley JH
    6. Schekman R

    (2019) Regulation of LC3 lipidation by the autophagy-specific class III phosphatidylinositol-3 kinase complex

    Molecular Biology of the Cell 30:1098–1107.

    https://doi.org/10.1091/mbc.E18-11-0743

    • PubMed
    • Google Scholar
11. 1. Buckles TC
    2. Ziemba BP
    3. Masson GR
    4. Williams RL
    5. Falke JJ

    (2017) Single-Molecule study reveals how receptor and ras synergistically activate PI3Kα and PIP3 signaling

    Biophysical Journal 113:2396–2405.

    https://doi.org/10.1016/j.bpj.2017.09.018

    • Google Scholar
12. 1. Calder PC
    2. Bond JA
    3. Harvey DJ
    4. Gordon S
    5. Newsholme EA

    (1990) Uptake and incorporation of saturated and unsaturated fatty acids into macrophage lipids and their effect upon macrophage adhesion and phagocytosis

    Biochemical Journal 269:807–814.

    https://doi.org/10.1042/bj2690807

    • PubMed
    • Google Scholar
13. 1. Chang C
    2. Young LN
    3. Morris KL
    4. von Bülow S
    5. Schöneberg J
    6. Yamamoto-Imoto H
    7. Oe Y
    8. Yamamoto K
    9. Nakamura S
    10. Stjepanovic G
    11. Hummer G
    12. Yoshimori T
    13. Hurley JH

    (2019) Bidirectional control of autophagy by BECN1 BARA domain dynamics

    Molecular Cell 73:339–353.

    https://doi.org/10.1016/j.molcel.2018.10.035

    • PubMed
    • Google Scholar
14. 1. Christoforidis S
    2. Miaczynska M
    3. Ashman K
    4. Wilm M
    5. Zhao L
    6. Yip SC
    7. Waterfield MD
    8. Backer JM
    9. Zerial M

    (1999) Phosphatidylinositol-3-OH kinases are Rab5 effectors

    Nature Cell Biology 1:249–252.

    https://doi.org/10.1038/12075

    • PubMed
    • Google Scholar
15. 1. De Craene J-O
    2. Bertazzi D
    3. Bär S
    4. Friant S

    (2017) Phosphoinositides, major actors in membrane trafficking and lipid signaling pathways

    International Journal of Molecular Sciences 18:634.

    https://doi.org/10.3390/ijms18030634

    • Google Scholar
16. 1. Degreif D
    2. Cucu B
    3. Budin I
    4. Thiel G
    5. Bertl A

    (2019) Lipid determinants of endocytosis and exocytosis in budding yeast

    Biochimica Et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids 1864:1005–1016.

    https://doi.org/10.1016/j.bbalip.2019.03.007

    • PubMed
    • Google Scholar
17. 1. Ellson CD
    2. Gobert-Gosse S
    3. Anderson KE
    4. Davidson K
    5. Erdjument-Bromage H
    6. Tempst P
    7. Thuring JW
    8. Cooper MA
    9. Lim ZY
    10. Holmes AB
    11. Gaffney PR
    12. Coadwell J
    13. Chilvers ER
    14. Hawkins PT
    15. Stephens LR

    (2001) PtdIns(3)P regulates the neutrophil oxidase complex by binding to the PX domain of p40(phox)

    Nature Cell Biology 3:679–682.

    https://doi.org/10.1038/35083076

    • PubMed
    • Google Scholar
18. 1. Fairn GD
    2. Schieber NL
    3. Ariotti N
    4. Murphy S
    5. Kuerschner L
    6. Webb RI
    7. Grinstein S
    8. Parton RG

    (2011) High-resolution mapping reveals topologically distinct cellular pools of phosphatidylserine

    The Journal of Cell Biology 194:257–275.

    https://doi.org/10.1083/jcb.201012028

    • PubMed
    • Google Scholar
19. 1. Fan W
    2. Nassiri A
    3. Zhong Q

    (2011) Autophagosome targeting and membrane curvature sensing by barkor/Atg14(L)

    PNAS 108:7769–7774.

    https://doi.org/10.1073/pnas.1016472108

    • PubMed
    • Google Scholar
20. 1. Fassy F
    2. Dureuil C
    3. Lamberton A
    4. Mathieu M
    5. Michot N
    6. Ronan B
    7. Pasquier B

    (2017) In vitro characterization of VPS34 lipid kinase inhibition by small molecules

    Methods in Enzymology 587:447–464.

    https://doi.org/10.1016/bs.mie.2016.09.070

    • Google Scholar
21. 1. Gaullier J-M
    2. Gillooly D
    3. Simonsen A
    4. Stenmark H

    (1999) Regulation of endocytic membrane traffic by phosphatidylinositol 3-phosphate

    Biochemical Society Transactions 27:666–670.

    https://doi.org/10.1042/bst0270666

    • Google Scholar
22. 1. Hamasaki M
    2. Furuta N
    3. Matsuda A
    4. Nezu A
    5. Yamamoto A
    6. Fujita N
    7. Oomori H
    8. Noda T
    9. Haraguchi T
    10. Hiraoka Y
    11. Amano A
    12. Yoshimori T

    (2013) Autophagosomes form at ER-mitochondria contact sites

    Nature 495:389–393.

    https://doi.org/10.1038/nature11910

    • PubMed
    • Google Scholar
23. 1. Hammond GR
    2. Balla T

    (2015) Polyphosphoinositide binding domains: key to inositol lipid biology

    Biochimica Et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids 1851:746–758.

    https://doi.org/10.1016/j.bbalip.2015.02.013

    • PubMed
    • Google Scholar
24. 1. Henmi Y
    2. Morikawa Y
    3. Oe N
    4. Ikeda N
    5. Fujita A
    6. Takei K
    7. Minogue S
    8. Tanabe K

    (2016) PtdIns4KIIα generates endosomal PtdIns(4)P and is required for receptor sorting at early endosomes

    Molecular Biology of the Cell 27:990–1001.

    https://doi.org/10.1091/mbc.E15-08-0564

    • Google Scholar
25. 1. Holthuis JC
    2. Menon AK

    (2014) Lipid landscapes and pipelines in membrane homeostasis

    Nature 510:48–57.

    https://doi.org/10.1038/nature13474

    • PubMed
    • Google Scholar
26. 1. Huang W
    2. Choi W
    3. Hu W
    4. Mi N
    5. Guo Q
    6. Ma M
    7. Liu M
    8. Tian Y
    9. Lu P
    10. Wang FL
    11. Deng H
    12. Liu L
    13. Gao N
    14. Yu L
    15. Shi Y

    (2012) Crystal structure and biochemical analyses reveal beclin 1 as a novel membrane binding protein

    Cell Research 22:473–489.

    https://doi.org/10.1038/cr.2012.24

    • PubMed
    • Google Scholar
27. 1. Hullin-Matsuda F
    2. Taguchi T
    3. Greimel P
    4. Kobayashi T

    (2014) Lipid compartmentalization in the endosome system

    Seminars in Cell & Developmental Biology 31:48–56.

    https://doi.org/10.1016/j.semcdb.2014.04.010

    • PubMed
    • Google Scholar
28. 1. Itakura E
    2. Kishi C
    3. Inoue K
    4. Mizushima N

    (2008) Beclin 1 forms two distinct phosphatidylinositol 3-kinase complexes with mammalian Atg14 and UVRAG

    Molecular Biology of the Cell 19:5360–5372.

    https://doi.org/10.1091/mbc.e08-01-0080

    • PubMed
    • Google Scholar
29. 1. Jackson CL

    (2018) Membrane trafficking: a little flexibility helps vesicles get into shape

    Current Biology 28:R706–R709.

    https://doi.org/10.1016/j.cub.2018.04.068

    • PubMed
    • Google Scholar
30. 1. Johner N
    2. Harries D
    3. Khelashvili G

    (2014) Curvature and lipid packing modulate the elastic properties of lipid assemblies: comparing HII and lamellar phases

    The Journal of Physical Chemistry Letters 5:4201–4206.

    https://doi.org/10.1021/jz5022284

    • PubMed
    • Google Scholar
31. 1. Judith D
    2. Jefferies HBJ
    3. Boeing S
    4. Frith D
    5. Snijders AP
    6. Tooze SA

    (2019) ATG9A shapes the forming autophagosome through arfaptin 2 and phosphatidylinositol 4-kinase IIIβ

    Journal of Cell Biology 218:1634–1652.

    https://doi.org/10.1083/jcb.201901115

    • Google Scholar
32. 1. Kim JY
    2. Zhao H
    3. Martinez J
    4. Doggett TA
    5. Kolesnikov AV
    6. Tang PH
    7. Ablonczy Z
    8. Chan CC
    9. Zhou Z
    10. Green DR
    11. Ferguson TA

    (2013) Noncanonical autophagy promotes the visual cycle

    Cell 154:365–376.

    https://doi.org/10.1016/j.cell.2013.06.012

    • PubMed
    • Google Scholar
33. 1. Kobayashi T
    2. Beuchat MH
    3. Chevallier J
    4. Makino A
    5. Mayran N
    6. Escola JM
    7. Lebrand C
    8. Cosson P
    9. Kobayashi T
    10. Gruenberg J

    (2002) Separation and characterization of late endosomal membrane domains

    Journal of Biological Chemistry 277:32157–32164.

    https://doi.org/10.1074/jbc.M202838200

    • PubMed
    • Google Scholar
34. 1. Köhler K
    2. Brunner E
    3. Guan XL
    4. Boucke K
    5. Greber UF
    6. Mohanty S
    7. Barth JM
    8. Wenk MR
    9. Hafen E

    (2009) A combined proteomic and genetic analysis identifies a role for the lipid desaturase Desat1 in starvation-induced autophagy in Drosophila

    Autophagy 5:980–990.

    https://doi.org/10.4161/auto.5.7.9325

    • PubMed
    • Google Scholar
35. 1. Kulakowski G
    2. Bousquet H
    3. Manneville J-B
    4. Bassereau P
    5. Goud B
    6. Oesterlin LK

    (2018) Lipid packing defects and membrane charge control RAB GTPase recruitment

    Traffic 19:536–545.

    https://doi.org/10.1111/tra.12568

    • Google Scholar
36. 1. Lagace TA
    2. Ridgway ND

    (2013) The role of phospholipids in the biological activity and structure of the endoplasmic reticulum

    Biochimica Et Biophysica Acta (BBA) - Molecular Cell Research 1833:2499–2510.

    https://doi.org/10.1016/j.bbamcr.2013.05.018

    • PubMed
    • Google Scholar
37. 1. Lee HC
    2. Inoue T
    3. Imae R
    4. Kono N
    5. Shirae S
    6. Matsuda S
    7. Gengyo-Ando K
    8. Mitani S
    9. Arai H

    (2008) Caenorhabditis elegans mboa-7, a member of the MBOAT family, is required for selective incorporation of polyunsaturated fatty acids into phosphatidylinositol

    Molecular Biology of the Cell 19:1174–1184.

    https://doi.org/10.1091/mbc.e07-09-0893

    • PubMed
    • Google Scholar
38. 1. Lee HC
    2. Kubo T
    3. Kono N
    4. Kage-Nakadai E
    5. Gengyo-Ando K
    6. Mitani S
    7. Inoue T
    8. Arai H

    (2012) Depletion of mboa-7, an enzyme that incorporates polyunsaturated fatty acids into phosphatidylinositol (PI), impairs PI 3-phosphate signaling in Caenorhabditis elegans

    Genes to Cells 17:748–757.

    https://doi.org/10.1111/j.1365-2443.2012.01624.x

    • PubMed
    • Google Scholar
39. 1. Lu J
    2. He L
    3. Behrends C
    4. Araki M
    5. Araki K
    6. Jun Wang Q
    7. Catanzaro JM
    8. Friedman SL
    9. Zong WX
    10. Fiel MI
    11. Li M
    12. Yue Z

    (2014) NRBF2 regulates autophagy and prevents liver injury by modulating Atg14L-linked phosphatidylinositol-3 kinase III activity

    Nature Communications 5:3920.

    https://doi.org/10.1038/ncomms4920

    • PubMed
    • Google Scholar
40. 1. Ma M
    2. Liu J-J
    3. Li Y
    4. Huang Y
    5. Ta N
    6. Chen Y
    7. Fu H
    8. Ye M-D
    9. Ding Y
    10. Huang W
    11. Wang J
    12. Dong M-Q
    13. Yu L
    14. Wang H-W

    (2017) Cryo-EM structure and biochemical analysis reveal the basis of the functional difference between human PI3KC3-C1 and -C2

    Cell Research 27:989–1001.

    https://doi.org/10.1038/cr.2017.94

    • Google Scholar
41. 1. Manni MM
    2. Tiberti ML
    3. Pagnotta S
    4. Barelli H
    5. Gautier R
    6. Antonny B

    (2018) Acyl chain asymmetry and polyunsaturation of brain phospholipids facilitate membrane vesiculation without leakage

    eLife 7:15114.

    https://doi.org/10.7554/eLife.34394

    • Google Scholar
42. 1. Marsh D

    (2007) Lateral pressure profile, spontaneous curvature frustration, and the incorporation and conformation of proteins in membranes

    Biophysical Journal 93:3884–3899.

    https://doi.org/10.1529/biophysj.107.107938

    • PubMed
    • Google Scholar
43. 1. Masson GR
    2. Burke JE
    3. Williams RL

    (2016) Methods in the study of PTEN structure: x-ray crystallography and hydrogen deuterium exchange mass spectrometry

    Methods in Molecular Biology 1388:215–230.

    https://doi.org/10.1007/978-1-4939-3299-3_14

    • PubMed
    • Google Scholar
44. 1. Masson GR
    2. Burke JE
    3. Ahn NG
    4. Anand GS
    5. Borchers C
    6. Brier S
    7. Bou-Assaf GM
    8. Engen JR
    9. Englander SW
    10. Faber J
    11. Garlish R
    12. Griffin PR
    13. Gross ML
    14. Guttman M
    15. Hamuro Y
    16. Heck AJR
    17. Houde D
    18. Iacob RE
    19. Jørgensen TJD
    20. Kaltashov IA
    21. Klinman JP
    22. Konermann L
    23. Man P
    24. Mayne L
    25. Pascal BD
    26. Reichmann D
    27. Skehel M
    28. Snijder J
    29. Strutzenberg TS
    30. Underbakke ES
    31. Wagner C
    32. Wales TE
    33. Walters BT
    34. Weis DD
    35. Wilson DJ
    36. Wintrode PL
    37. Zhang Z
    38. Zheng J
    39. Schriemer DC
    40. Rand KD

    (2019) Recommendations for performing, interpreting and reporting hydrogen deuterium exchange mass spectrometry (HDX-MS) experiments

    Nature Methods 16:595–602.

    https://doi.org/10.1038/s41592-019-0459-y

    • PubMed
    • Google Scholar
45. 1. Matsunaga K
    2. Saitoh T
    3. Tabata K
    4. Omori H
    5. Satoh T
    6. Kurotori N
    7. Maejima I
    8. Shirahama-Noda K
    9. Ichimura T
    10. Isobe T
    11. Akira S
    12. Noda T
    13. Yoshimori T

    (2009) Two beclin 1-binding proteins, Atg14L and rubicon, reciprocally regulate autophagy at different stages

    Nature Cell Biology 11:385–396.

    https://doi.org/10.1038/ncb1846

    • Google Scholar
46. 1. McMahon HT
    2. Boucrot E

    (2015) Membrane curvature at a glance

    Journal of Cell Science 128:1065–1070.

    https://doi.org/10.1242/jcs.114454

    • PubMed
    • Google Scholar
47. 1. Miller S
    2. Tavshanjian B
    3. Oleksy A
    4. Perisic O
    5. Houseman BT
    6. Shokat KM
    7. Williams RL

    (2010) Shaping development of autophagy inhibitors with the structure of the lipid kinase Vps34

    Science 327:1638–1642.

    https://doi.org/10.1126/science.1184429

    • Google Scholar
48. 1. Munson MJ
    2. Allen GF
    3. Toth R
    4. Campbell DG
    5. Lucocq JM
    6. Ganley IG

    (2015) mTOR activates the VPS34-UVRAG complex to regulate autolysosomal tubulation and cell survival

    The EMBO Journal 34:2272–2290.

    https://doi.org/10.15252/embj.201590992

    • PubMed
    • Google Scholar
49. 1. Munson MJ
    2. Ganley IG

    (2016) Determination of VPS34/PIK3C3 activity in vitro Utilising ³²P-γATP

    Bio-Protocol 6:e1904.

    https://doi.org/10.21769/BioProtoc.1904

    • PubMed
    • Google Scholar
50. 1. Murray JT
    2. Panaretou C
    3. Stenmark H
    4. Miaczynska M
    5. Backer JM

    (2002) Role of Rab5 in the recruitment of hVps34/p150 to the early endosome

    Traffic 3:416–427.

    https://doi.org/10.1034/j.1600-0854.2002.30605.x

    • PubMed
    • Google Scholar
51. 1. Nascimbeni AC
    2. Giordano F
    3. Dupont N
    4. Grasso D
    5. Vaccaro MI
    6. Codogno P
    7. Morel E

    (2017) ER-plasma membrane contact sites contribute to autophagosome biogenesis by regulation of local PI3P synthesis

    The EMBO Journal 36:2018–2033.

    https://doi.org/10.15252/embj.201797006

    • PubMed
    • Google Scholar
52. 1. Nishimura T
    2. Tamura N
    3. Kono N
    4. Shimanaka Y
    5. Arai H
    6. Yamamoto H
    7. Mizushima N

    (2017) Autophagosome formation is initiated at Phosphatidylinositol synthase-enriched ER subdomains

    The EMBO Journal 36:1719–1735.

    https://doi.org/10.15252/embj.201695189

    • PubMed
    • Google Scholar
53. 1. Ogasawara Y
    2. Itakura E
    3. Kono N
    4. Mizushima N
    5. Arai H
    6. Nara A
    7. Mizukami T
    8. Yamamoto A

    (2014) Stearoyl-CoA desaturase 1 activity is required for autophagosome formation

    Journal of Biological Chemistry 289:23938–23950.

    https://doi.org/10.1074/jbc.M114.591065

    • PubMed
    • Google Scholar
54. 1. Ohashi Y
    2. Soler N
    3. García Ortegón M
    4. Zhang L
    5. Kirsten ML
    6. Perisic O
    7. Masson GR
    8. Burke JE
    9. Jakobi AJ
    10. Apostolakis AA
    11. Johnson CM
    12. Ohashi M
    13. Ktistakis NT
    14. Sachse C
    15. Williams RL

    (2016) Characterization of Atg38 and NRBF2, a fifth subunit of the autophagic Vps34/PIK3C3 complex

    Autophagy 12:2129–2144.

    https://doi.org/10.1080/15548627.2016.1226736

    • PubMed
    • Google Scholar
55. 1. Payrastre B
    2. Missy K
    3. Giuriato S
    4. Bodin S
    5. Plantavid M
    6. Gratacap M

    (2001) Phosphoinositides: key players in cell signalling, in time and space

    Cellular Signalling 13:377–387.

    https://doi.org/10.1016/S0898-6568(01)00158-9

    • PubMed
    • Google Scholar
56. 1. Prinz WA
    2. Toulmay A
    3. Balla T

    (2020) The functional universe of membrane contact sites

    Nature Reviews Molecular Cell Biology 21:7–24.

    https://doi.org/10.1038/s41580-019-0180-9

    • PubMed
    • Google Scholar
57. 1. Rostislavleva K
    2. Soler N
    3. Ohashi Y
    4. Zhang L
    5. Pardon E
    6. Burke JE
    7. Masson GR
    8. Johnson C
    9. Steyaert J
    10. Ktistakis NT
    11. Williams RL

    (2015) Structure and flexibility of the endosomal Vps34 complex reveals the basis of its function on membranes

    Science 350:aac7365.

    https://doi.org/10.1126/science.aac7365

    • PubMed
    • Google Scholar
58. 1. Russell RC
    2. Tian Y
    3. Yuan H
    4. Park HW
    5. Chang YY
    6. Kim J
    7. Kim H
    8. Neufeld TP
    9. Dillin A
    10. Guan KL

    (2013) ULK1 induces autophagy by phosphorylating Beclin-1 and activating VPS34 lipid kinase

    Nature Cell Biology 15:741–750.

    https://doi.org/10.1038/ncb2757

    • PubMed
    • Google Scholar
59. 1. Schroit AJ
    2. Gallily R

    (1979)

    Macrophage fatty acid composition and phagocytosis: effect of unsaturation on cellular phagocytic activity

    Immunology 36:199–205.

    • PubMed
    • Google Scholar
60. 1. Schütter M
    2. Giavalisco P
    3. Brodesser S
    4. Graef M

    (2020) Local fatty acid channeling into phospholipid synthesis drives phagophore expansion during autophagy

    Cell 180:135–149.

    https://doi.org/10.1016/j.cell.2019.12.005

    • PubMed
    • Google Scholar
61. 1. Shulga YV
    2. Anderson RA
    3. Topham MK
    4. Epand RM

    (2012) Phosphatidylinositol-4-phosphate 5-kinase isoforms exhibit acyl chain selectivity for both substrate and lipid activator

    Journal of Biological Chemistry 287:35953–35963.

    https://doi.org/10.1074/jbc.M112.370155

    • PubMed
    • Google Scholar
62. 1. Sridhar S
    2. Patel B
    3. Aphkhazava D
    4. Macian F
    5. Santambrogio L
    6. Shields D
    7. Cuervo AM

    (2013) The lipid kinase PI4KIIIβ preserves lysosomal identity

    The EMBO Journal 32:324–339.

    https://doi.org/10.1038/emboj.2012.341

    • PubMed
    • Google Scholar
63. 1. Sun Q
    2. Fan W
    3. Chen K
    4. Ding X
    5. Chen S
    6. Zhong Q

    (2008) Identification of barkor as a mammalian autophagy-specific factor for beclin 1 and class III phosphatidylinositol 3-kinase

    PNAS 105:19211–19216.

    https://doi.org/10.1073/pnas.0810452105

    • PubMed
    • Google Scholar
64. 1. Tan X
    2. Thapa N
    3. Liao Y
    4. Choi S
    5. Anderson RA

    (2016) PtdIns(4,5)P2 signaling regulates ATG14 and autophagy

    PNAS 113:10896–10901.

    https://doi.org/10.1073/pnas.1523145113

    • PubMed
    • Google Scholar
65. 1. Uchida Y
    2. Hasegawa J
    3. Chinnapen D
    4. Inoue T
    5. Okazaki S
    6. Kato R
    7. Wakatsuki S
    8. Misaki R
    9. Koike M
    10. Uchiyama Y
    11. Iemura S
    12. Natsume T
    13. Kuwahara R
    14. Nakagawa T
    15. Nishikawa K
    16. Mukai K
    17. Miyoshi E
    18. Taniguchi N
    19. Sheff D
    20. Lencer WI
    21. Taguchi T
    22. Arai H

    (2011) Intracellular phosphatidylserine is essential for retrograde membrane traffic through endosomes

    PNAS 108:15846–15851.

    https://doi.org/10.1073/pnas.1109101108

    • PubMed
    • Google Scholar
66. 1. Vanni S
    2. Hirose H
    3. Barelli H
    4. Antonny B
    5. Gautier R

    (2014) A sub-nanometre view of how membrane curvature and composition modulate lipid packing and protein recruitment

    Nature Communications 5:4916.

    https://doi.org/10.1038/ncomms5916

    • PubMed
    • Google Scholar
67. 1. Wang RC
    2. Wei Y
    3. An Z
    4. Zou Z
    5. Xiao G
    6. Bhagat G
    7. White M
    8. Reichelt J
    9. Levine B

    (2012) Akt-mediated regulation of autophagy and tumorigenesis through beclin 1 phosphorylation

    Science 338:956–959.

    https://doi.org/10.1126/science.1225967

    • PubMed
    • Google Scholar
68. 1. Wang H
    2. Sun HQ
    3. Zhu X
    4. Zhang L
    5. Albanesi J
    6. Levine B
    7. Yin H

    (2015) GABARAPs regulate PI4P-dependent autophagosome:lysosome fusion

    PNAS 112:7015–7020.

    https://doi.org/10.1073/pnas.1507263112

    • PubMed
    • Google Scholar
69. 1. Xu Y
    2. Hortsman H
    3. Seet L
    4. Wong SH
    5. Hong W

    (2001) SNX3 regulates endosomal function through its PX-domain-mediated interaction with PtdIns(3)P

    Nature Cell Biology 3:658–666.

    https://doi.org/10.1038/35083051

    • PubMed
    • Google Scholar
70. 1. Xu P
    2. Baldridge RD
    3. Chi RJ
    4. Burd CG
    5. Graham TR

    (2013) Phosphatidylserine flipping enhances membrane curvature and negative charge required for vesicular transport

    The Journal of Cell Biology 202:875–886.

    https://doi.org/10.1083/jcb.201305094

    • PubMed
    • Google Scholar
71. 1. Yeung T
    2. Terebiznik M
    3. Yu L
    4. Silvius J
    5. Abidi WM
    6. Philips M
    7. Levine T
    8. Kapus A
    9. Grinstein S

    (2006) Receptor activation alters inner surface potential during phagocytosis

    Science 313:347–351.

    https://doi.org/10.1126/science.1129551

    • PubMed
    • Google Scholar
72. 1. Yeung T
    2. Heit B
    3. Dubuisson JF
    4. Fairn GD
    5. Chiu B
    6. Inman R
    7. Kapus A
    8. Swanson M
    9. Grinstein S

    (2009) Contribution of phosphatidylserine to membrane surface charge and protein targeting during phagosome maturation

    The Journal of Cell Biology 185:917–928.

    https://doi.org/10.1083/jcb.200903020

    • PubMed
    • Google Scholar
73. 1. Young LN
    2. Goerdeler F
    3. Hurley JH

    (2019) Structural pathway for allosteric activation of the autophagic PI 3-kinase complex I

    PNAS 116:21508–21513.

    https://doi.org/10.1073/pnas.1911612116

    • PubMed
    • Google Scholar
74. 1. Zhong Y
    2. Wang QJ
    3. Li X
    4. Yan Y
    5. Backer JM
    6. Chait BT
    7. Heintz N
    8. Yue Z

    (2009) Distinct regulation of autophagic activity by Atg14L and rubicon associated with beclin 1-phosphatidylinositol-3-kinase complex

    Nature Cell Biology 11:468–476.

    https://doi.org/10.1038/ncb1854

    • PubMed
    • Google Scholar
75. 1. Zhu Y
    2. Drake MT
    3. Kornfeld S

    (1999) ADP-ribosylation factor 1 dependent clathrin-coat assembly on synthetic liposomes

    PNAS 96:5013–5018.

    https://doi.org/10.1073/pnas.96.9.5013

    • PubMed
    • Google Scholar

Author details

1. Yohei Ohashi

   MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge, United Kingdom

   Contribution

   Conceptualization, Data curation, Formal analysis, Supervision, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Contributed equally with

   Shirley Tremel

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-2288-130X

2. Shirley Tremel

   MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge, United Kingdom

   Contribution

   Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

   Contributed equally with

   Yohei Ohashi

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-4077-0021

3. Glenn Robert Masson

   MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge, United Kingdom

   Contribution

   Data curation, Formal analysis, Supervision, Validation, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-1386-4719

4. Lauren McGinney

   MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge, United Kingdom

   Contribution

   Data curation, Formal analysis, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

5. Jerome Boulanger

   MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge, United Kingdom

   Contribution

   Software, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

6. Ksenia Rostislavleva

   MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge, United Kingdom

   Contribution

   Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

7. Christopher M Johnson

   MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge, United Kingdom

   Contribution

   Data curation, Formal analysis, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

8. Izabella Niewczas

   Babraham Institute, Cambridge, United Kingdom

   Contribution

   Data curation, Formal analysis, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

9. Jonathan Clark

   Babraham Institute, Cambridge, United Kingdom

   Contribution

   Data curation, Formal analysis, Supervision, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

10. Roger L Williams

    MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge, United Kingdom

    Contribution

    Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

    For correspondence

    rlw@mrc-lmb.cam.ac.uk

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0001-7754-4207

• 4,030

  views

• 473

  downloads

• 33

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.