(A) SEC-MALS analysis of purified human Class III PI3K complex I (left panel) and complex II (right panel). The protein samples were run on a Superose 6 10/300 column. Insets: SDS-PAGE of starting material stained with Coomassie staining. Both complex I and complex II are monodisperse, with an average mass for complex I consistent with a 1:1:1:1 Beclin 1/VPS15/VPS34/ATG14L complex (370 kDa, theoretical mass: 363 kDa) and an average mass for complex II consistent with a 1:1:1:1 Beclin 1/VPS15/VPS34/UVRAG complex (384 kDa, theoretical mass: 386 kDa). (B) Assay design for activities of human complexes I and II on GUVs, using confocal microscopy and a labelled PI(3)P-binding p40-PX domain (AF647-PX). Fluorescence from the Lissamine-Rhodamine (Liss-Rhod GUV) channel delineates the membrane, while the AF647 is indicative of VPS34 activity on the membrane. Scale bar: 5 μm. (C) Complex I is more active than complex II on GUVs with ‘DO base’ lipids (18% mixed chain PI, 10% DOPS, 17% DOPE and 55% DOPC). In the lower panels, the initial rates after a lag phase (AF647-PX fluorescence change/min in arbitrary units, AU) and confocal images corresponding to the AF647-PX and Liss-Rhod channels at the end of the reaction are shown. For clarity, only mean values of measurements for each time point in the reaction progress curves are plotted here and throughout all figures. Plots with SDs for each time point are shown in Supplementary file 3. All scale bars: 5 μm. ***: p<0.001 (p<0.0001). Figure 1—figure supplement 2 illustrates some of the raw images that were used for quantification. Scale bars: 15 μm. (D) Membrane binding of complexes I and II using a lipid flotation assay. Large unilamellar vesicles (LUVs, 100 nm) and proteins were mixed and pipetted on a sucrose gradient. The gradient was then centrifuged, fractionated and analysed by SDS-PAGE. Fractions 1–3 are the least dense fractions of a sucrose gradient, containing floating vesicles and membrane-bound proteins. Fractions 5–6 are the highest density sucrose fractions, containing the pelleted proteins. The complexes alone sediment into the denser portion of the gradients (left), but the presence of LUVs cause the complexes to associate with LUVs floating on the top of the gradient (right). Gel quantification can be seen in Figure 1—figure supplement 3.