Identification of the GABARAP binding determinant in PI4K2A

Abstract

GABARAP is a member of the ATG8 family of ubiquitin-like autophagy related proteins. It was initially discovered as a facilitator of GABA-A receptor translocation to the plasma membrane and has since been shown to promote the intracellular transport of a variety of other proteins under non-autophagic conditions. We and others have shown that GABARAP interacts with the Type II phosphatidylinositol 4-kinase, PI4K2A, and that this interaction is important for autophagosome-lysosome fusion. Here, we identify a 7-amino acid segment within the PI4K2A catalytic domain that contains the GABARAP interaction motif (GIM). This segment resides in an exposed loop that is not conserved in the other mammalian Type II PI 4-kinase, PI4K2B, explaining the specificity of GABARAP binding to the PI4K2A isoform. Mutation of the PI4K2A GIM inhibits GABARAP binding and PI4K2A-mediated recruitment of cytosolic GABARAP to subcellular organelles. We further show that GABARAP binds to mono-phosphorylated phosphoinositides, PI3P, PI4P, and PI5P, raising the possibility that these lipids contribute to the binding energies that drive GABARAP-protein interactions on membranes.

Keywords: ATG8; GABARAP; LIR motif; fluorescence fluctuation spectroscopy; phosphatidylinositol 4-kinase 2A.

Figure 1. Identification of the GABARAP binding motif in PI4K2A

(A) Scheme of Type II PI 4-kinases highlighting the location of the diverse Loop 4 sequences within the highly conserved catalytic domains of PI4K2A and PI4K2B. Mutation of residues marked in red inhibit GABARAP binding. (B) Orientation of the GABARAP binding motif with respect to the ATP and membrane binding sites in PI4K2A. (C–E) Heterospecies partition (HSP) analysis showing comigration of GABARAP with PI 4-kinases. (C) HSP concept. (D) Comigration of GABARAP with WT PI4K2A (green dots) but not with the 7A mutant in which residues 330-336 were replaced with alanines (blue triangles). (E) Comigration of GABARAP with WT PI4K2A (green dots) but not with PI4K2B (red triangles). Note that 30 of the 47 data points for PI4K2A^WT in panels (D) and (E) were reproduced with permission from the ACS from Chen et al., Comobility of GABARAP and phosphatidylinositol 4-Kinase 2A on cytoplasmic vesicles (2018) Biochemistry 57: 3556-3559.

Figure 2. Lack of effect of mutating the GABARAP binding motif on the enzymatic activity of PI4K2A

(A) Immunoblots of membrane fractions from HeLa cells expressing WT EGFP-PI4K2A (EGFP-WT), EGFP-PI4K2A^7A (EGFP-7A) or untagged WT PI4K2A (WT (untagged)). Membranes were blotted with anti-PI4K2A. (B) Thin layer chromatography (TLC) showing PI4P production in membranes isolated from transfected HeLa cells. Assays were performed in the presence or absence of Triton X-100, which stimulates PI 4-kinases but inhibits PI 3-kinases. PI4P production is increased equally by expression of WT and mutant kinases, and EGFP-tagging does not diminish activity. (C) Quantification of TLC assays based on four separate experiments.

Figure 3. HSP analysis of cells expressing EGFP-tagged PI4K2A catalytic domains (residues 91-478) and mCherry-tagged GABARAP

Results show comigration of GABARAP with WT PI4K2A (91-478) (green dots) but not with (91-478) containing the 7A mutation (blue triangles). A PI4K2A (91-478) construct in which residues ₃₃₃MVRE₃₃₆ within the GABARAP binding motif are mutated to alanine shows reduced association with PI4K2A (purple dots).

Figure 4. HSP analysis of cells expressing EGFP-PI4K2A and mCherry-GABARAP mutants

(A) Analysis of the association of GABARAP truncation mutants lacking residues 110-117 (turquoise dots), residues 1-15 (red triangles), or residues 1-30 (purple dots) with PI4K2A. (B) Analysis of the effect of point mutations within a C-terminal segment of GABARAP on its association with PI4K2A. Data showing association of GABARAP (1-109) (turquoise dots) are reproduced from panel (A) for comparison.

Figure 5. Binding of GABARAP to lipids

(A) Flotation assay showing the binding of GST-GABARAP, but not GST, to liposomes that float to the 0–25% sucrose interface after centrifugation. Seven fractions were collected from the top to the bottom of the gradient, samples were electrophoresed, and gels were stained with Coomassie blue. The figure represents the result of one of three separate experiments. Percentages of liposome-bound GST-GABARAP in the three experiments were 100%, 88.3%, and 95.8%, as determined using the LI-COR imaging system. Uncropped gels are shown in Supplementary Figure S1. (B–D) Determination of the binding of GST (B) and GST-GABARAP (C) to different lipids and of GST-GABARAP to different amounts of phosphoinositides (D) by ‘PIP-Strip’ overlay. Membranes blotted with designated lipids were overlaid with ∼25 nM GST-GABARAP or ∼38 nM GST (as a negative control) and then immunoblotted with anti-GST antibody.

Figure 6. Hypothetical models showing association of PI4K2A with GABARAP

(A) Non-autophagic trafficking. Palmitoylated PI4K2A recruits unlipidated, cytosolic GABARAP (G) to Golgi and/or endosomal membranes. These membranes are the source of carrier vesicles that transport cargo (e.g., receptors) anterogradely from the trans-Golgi network (TGN) or recycling endosomes to the plasma membrane (PM). (B) Autophagic trafficking. ATG9A-containing vesicles dock on autophagosome initiation sites on the endoplasmic reticulum and expand into mature autophagosomes by ATG2-mediated lipid transport and ATG9A-mediated lipid scrambling. ULK1/PI3KC3-mediated PI3P production on the docked vesicles results in the recruitment of the ATG8 conjugation complex, which lipidates GABARAP, thereby anchoring it into the nascent autophagosome membrane. ATG9A vesicles carry both PI4K3B (PI4KIIIβ) and PI4K2A, which produce pools of PI4P utilized in autophagosome formation and autophagosome-lysosome-fusion, respectively. According to our model, lipidated PI4K2A and GABARAP are stably associated throughout the autophagosomal process. See text for more details.