Structure based biophysical characterization of the PROPPIN Atg18 shows Atg18 oligomerization upon membrane binding

Abstract

PROPPINs (β-propellers that bind polyphosphoinositides) are PtdIns3P and PtdIns(3,5)P₂ binding autophagy related proteins. They contain two phosphatidylinositolphosphate (PIP) binding sites and a conserved FRRG motif is essential for PIP binding. Here we present the 2.0 Å resolution crystal structure of the PROPPIN Atg18 from Pichia angusta. We designed cysteine mutants for labelling with the fluorescence dyes to probe the distances of the mutants to the membrane. These measurements support a model for PROPPIN-membrane binding, where the PROPPIN sits in a perpendicular or slightly tilted orientation on the membrane. Stopped-flow measurements suggest that initial PROPPIN-membrane binding is driven by non-specific PIP interactions. The FRRG motif then retains the protein in the membrane by binding two PIP molecules as evident by a lower dissociation rate for Atg18 in comparison with its PIP binding deficient FTTG mutant. We demonstrate that the amine-specific cross-linker Bis(sulfosuccinimidyl)suberate (BS3), which is used for protein-protein cross-linking can also be applied for cross-linking proteins and phosphatidylethanolamine (PE). Cross-linking experiments with liposome bound Atg18 yielded several PE cross-linked peptides. We also observed intermolecular cross-linked peptides, which indicated Atg18 oligomerization. FRET-based stopped-flow measurements revealed that Atg18 rapidly oligomerizes upon membrane binding while it is mainly monomeric in solution.

Figure 1

Pichia angusta Atg18 structure. (A) Cartoon representations of Atg18 rainbow-colored from blue at the N-terminus to red at the C-terminus. The ligands and FRRG arginine side chains are shown as ball and sticks. Loop 6CD (297–437) is missing and marked through the last ordered residues 296 and 438. Panels (A–C) show the protein in the same orientation. Surface representations of Atg18 showing (B) surface charges with blue depicting positive and red negative charges and (C) conservation of the molecule. Conservation analysis was done with Consurf. Proteins used for the sequence alignment are listed in the Methods section. (D) Close up on site 1 with bound citrate. The omit mFo-DFc difference map, where m is the figure of merit and D the sigma A weighting factor has a contour level of σ = +2.5. Scheme showing ligand protein interactions was adapted from LigPlot⁺. Hydrogen bonds and salt bridges between ligand and protein are shown with dashed lines and distances are given in Ångstrom. Water molecules are depicted as cyan spheres. (E) Site 1 with a bound phosphate ion. The 2.5 Å resolution 2mFo-DFc map is contoured at 1.0 σ. (F) PIP binding site 2 with a bound phosphate ion. The 2.0 Å resolution 2mFo-DFc electron density map is contoured at 1.0 σ. Residues with disordered side chains are marked with an asterisk.

Figure 2

Stopped-flow measurements with S448C Atg18 and S448C FTTG Atg18. Time courses of Texas Red (A and B) or Rhodamine B (E and F) fluorescence emission measurements at different accessible lipid concentrations are shown for S448C Atg18 (A and E) and S448C FTTG Atg18 (B and F). A scale-up of the the traces in Fig. 2B and fitting results are shown in Supplementary Figure 11 and Supplementary Tables 3–7, respectively. (C and G) Graphs showing the dependency of k _obs1 from the accessible lipid concentration. The y-intercept provides the dissociation rate constant k _off, and the slope yields the association rate constant k _on. Bars represent the standard errors of three to five technical repeats. The solid line shows a weighted linear fit (equation 2 in Methods). Amplitudes of the k _obs1 used for data analysis are shown in Supplementary Figure 5. (D and H) Increases of Texas Red (D) or Rhodamine B (H) fluorescence emission are shown at different accessible lipid concentrations for S448C Atg18 and S448C FTTG Atg18.

Figure 3

(A) Atg18 structure showing residues, which were mutated to cysteines for fluorescence labeling experiments. E170 is marked with an asterisk because its side chain is disordered in the structure and was modelled as an alanine. (B) Liposome flotation assays were performed with wild-type, cysteine-free and single cysteine Atg18 mutants using 100 nm LUVs composed of DOPC/DOPE/Texas Red-PE/PtdIns(3,5)P₂ (76:21:2:1, molar ratio). The top two fractions of the gradient show protein bound to liposomes and bottom fractions contain unbound protein. The S448C FTTG mutant served as a negative control. (C) Observed changes of the normalized fluorescence at 520 nm for the Oregon Green labelled mutants after Texas Red liposome addition. (D) Model for Atg18 binding at the membrane. The docking was manually performed taking data from (C) into account. Residues are color coded according to the observed change in normalized fluorescence. Color bar is shown.

Figure 4

(A) Intramolecular cross-links of Atg18 are shown with red lines. Cross-links that are not included in the crystal structure are indicated by dotted lines. The seven blades of the β-propeller are rainbow colored. Residues missing from the crystal structure are highlighted in grey and light orange (residues 297–437, blade 6). (B) Bar diagramme indicating intermolecular Atg18 interactions (blue Lines). (C) Native mass spectrum of Atg18 oligomers. Calculated masses of the oligomers are given. (D) Intramolecular cross-links identified in Atg18 in the presence of liposomes. (E) Blue lines indicate intermolecular cross-links between two Atg18 molecules in the presence of liposomes. Red lines show protein-PE cross-links. PE cross-linked residues are labelled.

Figure 5

Characterization of Atg18-Atg18 interactions with FRET based stopped-flow measurements. (A) Graph shows time courses of rapid mixing of 0.5 µM Atg18 C113 (labelled with Alexa Fluor 488) with 0.5 µM Atg18 C113 (labelled with Alexa Fluor 546) either in the presence of unlabeled liposomes (1 mM total lipid), Ins(1,3,5)P₃ (6 µM) or free in solution. (B) Experiments are similar to (A), but Oregon Green and Texas Red were used as donor and acceptor dyes instead. (C) Measurements with Texas Red labelled liposomes in comparison with measurements using labelled proteins for monitoring of the FRET signal. Unlabelled protein was mixed with unlabelled protein bound liposomes to monitor aggregation by measuring of the absorbance at 405 nm. Time course traces were normalized for comparison.

Figure 6

Model for Atg18 membrane binding. The protein oligomerizes upon membrane binding while it is mainly monomeric in solution. Higher order oligomers might be formed by membrane bound Atg18.