Fig 2: Rab1a is a specific activator of VPS34 complex I.a Immunoblot of streptavidin precipitates from cells in which MitoID had been carried out for Rab1a WT, QL, and SN mutants (left) or Rab5 QL (right). Rab1a interacts specifically with components of complex I (VPS34, VPS15, Beclin 1, ATG14L) but not with the complex II-specific UVRAG. Rab5a shows only a weak interaction with the complex I-specific ATG14L. b Complex I was potently activated by membrane-attached Rab1a in a GTP-dependent manner. c No activation of complex II by either membrane-attached Rab1a–GTP or Rab1a–GDP was detected. b, c Micrographs: AF647-PX signals at the end of reactions. Scale bars: 5 µm. Bar graphs: initial rates of the reaction curves. d Lipid flotation assays showing complex I recruitment to Rab1a-decorated membranes in a GTP-dependent manner. Gel quantification is in Supplementary Fig. 2c. e Mapping Rab1a binding site on complex I by HDX-MS. Rab1a binding increases protection (coloured in cyan and blue) of the VPS34 C2 insertion (C2HH) and decreases protection (coloured yellow and red) of Beclin 1 CC2. Source data are provided as a Source Data file.

Fig 3: Role of BECN1 on PIK3C3 activity and PtdIns3K complex formation. (A) The endogenous protein-protein interaction between PIK3C3 and ATG14 or PIK3C3 and UVRAG in the indicated HeLa cells. Cells were lysed by NP40 lysis buffer. Cell lysates were immunoprecipitated with anti-PIK3C3 antibody or protein A beads only as indicated (WCL, whole cell lysate). (B) Detection of omegasomes in the indicated HeLa cells. ZFYVE1-GFP expression plasmid was introduced into cells by electroporation. The fluorescence microscopy images of cells subjected to amino acid starvation for the indicated time were collected using a live-cell imaging system. Scale bar: 10 µm. (C) The statistical results of omegasome number in the indicated HeLa cells. The data are the mean ± s.e.m.; **P = 0.0032; ***P < 0.0001; NS, not significant; the Student t test.

Fig 4: BECN1 is not required for LC3B lipidation during autophagy induction. (A) Strategy for knocking out BECN1 using transcription activator-like effector (TALE) nucleases. Exon 2 was targeted in the locus shown, with the 2 TALE nuclease-binding sequences shown by underlining. (B) Immunoblotting analysis of the indicated wild-type cells and different targeted clones for BECN1. The TUBB/ß-tubulin bands were used as internal controls in this and other figures. (C) Sequence analysis of the BECN1 gene in the genome of the selected BECN1 knockout HeLa clones (#1, #13, #14, #28, #40, and #45), the fraction number in the bracket indicates the number of genome PCR clones tested for a particular edition version versus the total PCR clones tested for the single cell clone. (D) Immunoblotting analysis of the levels of LC3B-I and LC3B-II of the indicated cells treated with amino acid starvation for 4 h. The GAPDH bands were used as internal protein loading controls in this and other figures. The total LC3B/GAPDH ratio is shown in numbers below each gel lane, and the quantification of relative levels of LC3B-II/LC3B-I is shown in the bar graph. The data were quantified using imageJ in this and other figures. This result was repeated more than 3 times as shown in Fig. S1. (E) Immunoblotting analysis of the indicated HeLa cell lysates. The monoclonal antibodies targeting N-terminal and C-terminal BECN1 were used. (F) Immunoblotting analysis of UVRAG expression in the indicated HeLa cells. (G) Immunoblotting analysis of the levels of LC3B-I and LC3B-II of the indicated cells treated by 0.5 mM H2O2 for 12 h.

Fig 5: The Beclin 1/UVRAG adaptor arm, together with Rab5a, mediates highly flexible membrane binding.a Subtomogram averaging density of complex II–BATS bound to Rab5a–GTP decorated LUVs. Complex II directly attaches to membranes via its Beclin 1/UVRAG adaptor arm. The catalytic arm hovers above the membrane with no direct contact. No density for the BATS domain is observed. b The distance between the Rab5a density and membrane is ~50 Å, which easily can be spanned by the 34 residues forming the Rab5a HVR (not ordered in the structure). c The adaptor arm shows density from the Beclin 1 BARA aromatic finger 1 (AF1) in contact with the outer membrane leaflet. Scale bar, 4 nm. d Three different classes from 3D classification of Rab5a–GTP/complex II–BATS show different orientations relative to the membrane, however, the same complex II model with two ordered arms and an active conformation of the VPS34 kinase domain can be fit to each class. While the adaptor arm stays bound to the membrane and serves as an anchor point, complex II can tilt both up/down as well as sideways. e A three-state model for VPS34 complex II activation on membranes by Rab5 GTPases. Off membranes, the VPS34 kinase domain of complex II is autoinhibited by VPS15 (left). Binding of Rab5a–GTP to the adaptor arm recruits the complex to membranes, releasing the auto-inhibitory interactions and enabling scooting via the adaptor arm state (middle). The complex is able to tilt up and down so that the catalytic arm transiently engages the membrane to phosphorylate PI to synthesise PI3P (right). States 2 and 3 are able to scoot on the membrane and thereby encounter new substrate.