Temporal Ordering in Endocytic Clathrin-Coated Vesicle Formation via AP2 Phosphorylation

Abstract

Clathrin-mediated endocytosis (CME) is key to maintaining the transmembrane protein composition of cells' limiting membranes. During mammalian CME, a reversible phosphorylation event occurs on Thr156 of the μ2 subunit of the main endocytic clathrin adaptor, AP2. We show that this phosphorylation event starts during clathrin-coated pit (CCP) initiation and increases throughout CCP lifetime. μ2Thr156 phosphorylation favors a new, cargo-bound conformation of AP2 and simultaneously creates a binding platform for the endocytic NECAP proteins but without significantly altering AP2's cargo affinity in vitro. We describe the structural bases of both. NECAP arrival at CCPs parallels that of clathrin and increases with μ2Thr156 phosphorylation. In turn, NECAP recruits drivers of late stages of CCP formation, including SNX9, via a site distinct from where NECAP binds AP2. Disruption of the different modules of this phosphorylation-based temporal regulatory system results in CCP maturation being delayed and/or stalled, hence impairing global rates of CME.

Keywords: AAK1; AP2 endocytic adaptor; NECAP; NMR; Numb-associated kinases (NAK); SNX9; TIRF; clathrin-mediated endocytosis; crystallography; regulation by phosphorylation.

Graphical abstract

Figure 1

μ2T156 Phosphorylation Increases as CCPs Mature and Is Important for Efficient CME (A) Representative images of control RPE cells and cells treated with LP. White arrows indicate the presence of a fluorescent signal in the green channel (EGFP-CLCa) and the same position in the AP2 (red) and P-AP2 channel (cyan) to compare presence or absence of the corresponding signal at the same position. Scale bar, 5 μm. (B) Time course of TfR internalization at 37°C. Data represent mean SD, n = 3. Two-tailed unpaired t tests were used to assess statistical significance. ^∗p ≤ 0.05. (C) Bar graphs represent the quantification of the membrane distribution of AP2 (red) or P-AP2 (blue) in CCPs. The bar height displays the proportion of CCPs (detected in the EGFP-CLCa channel) that contain significant AP2 or P-AP2 signal in the same position plotted as a function of EGFP-CLCa signal (x axis). Data represent mean values, n = 30. (D) Comparison of PCCs to assess the relationship of the paired intensity values of CCP components EGFP-CLCa, AP2, and P-AP2. Data represent mean SD, n = 30. Two-tailed unpaired t tests were used to assess statistical significance. ^∗∗∗∗p ≤ 0.0001; ns, p > 0.05. (E) Representative images of control cells and cells treated with LP, showing the distribution of EGFP-CLCa signal and TfR detected with anti-TfR mAb. White arrows indicate the presence of the signal in the green channel (EGFP-CLCa) and the same position in the TfR channel. Scale bar, 5 μm. (F) PCCs comparing the strength of linear relationship of the paired intensity values of EGFP-CLCa and TfR. Bars represent mean SD, n = 27. Two-tailed unpaired t tests were used to assess statistical significance. ^∗∗∗∗p ≤ 0.0001. (G) Representative TIRFM images of control cells and cells treated with LP. Red highlights indicate clathrin-coated structures categorized as persistent. Scale bar, 5 μm. (H and I) Scatterplots with the values of density of the persistent structures (H) and the average lifetime of the 95th percentile (I) for control cells, cells treated with LP, and μ2^NAK− RPE cells. Lines represent mean and SD for each group (n = 20). Two-tailed unpaired t tests were used to assess statistical significance. ^∗∗∗∗p ≤ 0.0001.

Figure 2

A New Open+ Conformation for Cargo-Bound AP2 in Both μ2t156 Phosphorylated and Unphosphorylated Forms (A) Structure of T156-phosphorylated AP2 core (Pcore) in open+ conformation with YxxΦ cargo in gold and dileucine cargo in orange. The refined 2mFo-DFc electron density map (gray) is contoured at 1 sigma. (B–E) Enlargements of functionally important regions with subunits and refined 2mFo-DFc electron density colored as above: (B) shows unambiguous Cμ2 positioning, (C) the missing portion of the phosphorylated μ2 linker, (D) YxxΦ cargo peptide (2mFo-DFc map clipped to 3Å around peptide, contoured at 1 sigma), and (E) dileucine cargo peptide (2mFo-DFc map clipped to 3Å around peptide, contoured at 0.5 sigma).

Figure 3

The Open+ Conformation Is Related to the Open Conformation by Movement of μ2 and Is Membrane- or Cargo-Binding Competent (A) Views “perpendicular to” (top row) and “through” (bottom row) the membrane of Pcore in “closed” or “locked” cytosolic conformation (left column) and membrane attached open (center) and open+ (right column) conformations. (B) Relative positions of Cμ2 in closed (orange), open (blue), and open+ (purple) conformations of Pcore aligned by the superposition of their α subunits. (C) Superposition of entire μ2 subunits in open (gray) and open+ conformations, superimposed on the basis of Nμ2 to highlight the change in orientation of the two subdomains relative to each other viewed “through the membrane.” Modeled fragments of linkers are shown as dashed.

Figure 4

NECAP Binds to AP2 Core in a μ2T156-Phosphorylation-Dependent Manner (A) Table showing the 10 proteins form pig brain cytosol as scored by their iBAQ values that are most enriched in their binding to GST-Pcore over GST-core. The top two “hits” are NECAP1 and NECAP2. (B) Myc-tagged Pcore preferentially binds to full-length recombinant NECAP in pull-downs. SDS- PAGE gels were blotted and stained with Ponceau red (upper) or developed with anti-myc antibody (lower). (C) Pcore binds more tightly to NECAP1 PHear than the phosphorylated μ2 linker. Representative ITC traces (top) and fitted curves with K_Ds of binding (bottom) of Pcore in red and to μ2-linker (residues 149–163)-derived peptide (cell) in blue to WT NECAP1 PHear (syringe). (D) SDS-PAGE gels of “pull-downs” using His₆NECAP1 constructs of various length immobilized on NiNTA beads were blotted and stained with Ponceau red (upper panels) and developed with anti-myc antibody (lower panel), showing that PHear and PHear-Ex but not Ex bind to Pcore. (E) SDS-PAGE gels of “pull-downs” using His₆NECAP1 PHear immobilized on NiNTA beads shows that NECAP PHear binds GSTμ2 linkers 122–171 and 149–163 only when T156 is phosphorylated by co-expression of linkers with the AAK1 catalytic subunit. (F) NECAP1 PHear only binds μ2 linker peptides when the peptides are phosphorylated: K_D of 55 μM. Example ITC traces (top) and fitted curves with K_Ds of binding (bottom) of WT NECAP1 PHear (cell) to μ2-linker- (residues 149–163) derived peptides (syringe): Q154S + G157A (kinase ablation) mutant, red; WT unphosphorylated, cyan; and T156-phosphorylated, dark blue. (G) Comparison of the quantity of endogenous NECAP that co-immunoprecipitated with AP2 from lysates obtained from control cells (first lane) and cells treated with LP and μ2^NAK− RPE cells (middle and last lane) by western blotting with anti NECAP polyclonal antibody. AP2 binds NECAPs only when AP2 is μ2T156 phosphorylated. (H) Representative immunofluorescence images showing co-localization of EGFP-CLCa, NECAP1, and AP2 (top row) and EGFP-CLCa, NECAP1, and P-AP2 when treated with DMSO (middle row) or LP (bottom row). White arrows indicate the presence of a fluorescent signal in the EGFP-CLCa channel and the same position in the NECAP (red channel) and AP2 or P-AP2 channel (cyan channel). Scale bar, 5 μm. (I) Comparison of PCCs to assess the relationship of the paired intensity values of CCP components EGFP-CLCa, NECAP, AP2, and P-AP2 from immunofluorescence images. Data represent the mean SD, n = 30. Two-tailed Student’s t tests were used to assess statistical significance. ^∗∗p ≤ 0.01, ^∗∗∗∗p ≤ 0.0001.

Figure 5

A μ2T156 Phosphorylation Drives NECAP Recruitment to CCPs (A and B) CCPs detected in control RPE cells expressing EGFP-CLCa and NECAP1-mRuby2 were grouped into cohorts with a specific range of lifetimes, and corresponding intensity traces of EGFP-CLCa and NECAP1-mRuby2 were plotted as a function of the time. CCPs were further subcategorized into NECAP-positive CCPs (A) and NECAP-negative CCPs (B). Averaged CCP intensity traces for EGFP-CLCa (green) and NECAP1 (red) from control cells are shown as mean ± SE (shaded areas) per lifetime cohort. (C) Lifetime distributions of all CCPs (gray) in control RPE cells expressing EGFP-CLCa and NECAP1-mRuby2. CCPs were further subcategorized as NECAP1-positive (red) or NECAP1-negative (blue) for comparison of lifetime distribution between the two populations of CCPs. (D) Number of CCPs identified as NECAP1-positive or NECAP1-negative in control cells, cells treated with LP inhibitor (see [E] and [F]), pan-phosphatase inhibitor calyculin A (see Figure S5), and μ2^NAK− RPE cells. Values are mean ± SD, n= 18. (E and F) CCPs detected in RPE cells expressing EGFP-CLCa and NECAP1-mRuby2 and treated with LP inhibitor were grouped into cohorts with a specific range of lifetimes, and corresponding intensity traces of EGFP-CLCa and NECAP1-mRuby2 were plotted as a function of the time as in panels (A) and (B). (G and H) Scatterplots with values of density of persistent structures (G) and average lifetime of the 95th percentile (H) for cells treated with control siRNA (blue) and NECAP1 + NECAP2 depleting siRNAs (green). Lines represent mean and SD for each group, n = 20.

Figure 6

Mechanism of NECAP1 PHear Domain Binding to T156 Phosphorylated μ2 Linker (A) Family of the 30 lowest energy structures of the PHear domain of NECAP1 (colored blue N to red C) in complex with the T156 phosphorylated linker peptide (residues 149–163 colored pale to dark purple) of Cμ2. (B and C) Enlargement of best 10 structures (B) and single best structure (C) of the linker binding site, showing side chains of the strongest binding-abolition mutants Arg90Ala and Arg113Ala with carbons in light and dark green, respectively. Phosphate group is shown in purple (phosphorous) and red (oxygens). (D, E, G, and H) PHear:phospho μ2 peptide complex showing residues whose mutation strongly (light green, dark green, and orange) or mildly (yellow) affects the interaction as determined by ITC in (J). Mutations that do not affect NECAP binding but do affect amphiphysin and CALM binding via FxDxF motifs are shown in red and μ2 peptide in purples. (F and I) Electrostatic potential of PHear to highlight basic binding pockets of both binding sites colored from +10eV (blue) to −10eV (red). μ2 peptide in purples placed on after potential calculation. (J) PHear:phospho μ2 peptide complex colored according to the ConSurf conservation score (bottom) based on alignment of 150 sequences of 35%–95% sequence identity using the ConSurf algorithm. (H) Example ITC traces (top) and fitted curves with K_Ds (bottom) of binding of WT (dark blue) and mutants of NECAP1 PHear (cell) to WT μ2 linker phosphopeptide (syringe).

Figure 7

NECAP Recruits BAR Domain Proteins, Leading to CCV Scission (A) Analysis of the P-AP2 interactome. Volcano plot obtained from SILAC-based quantitative proteomics analysis of proteins that co-immunoprecipitated with AP2 in lysates obtained from control RPE cells treated with DMSO or cells treated with LP. Red dots represent proteins exhibiting significant (p < 0.05) fold enrichment in the P-AP2 interactome. n = 6. (B) Averaged CCP intensity traces for EGFP-CLCa (green) and SNX9 (red) from control cells (top left) and cells treated with NECAP1/2 siRNA (top right), and LP (bottom) are shown as mean ± SE (shaded areas) per lifetime cohort. (C) Schematic model of CCP maturation leading to CCV scission. The evolving CCP structure and the phases it passes through are shown at the top. The middle represents the degree of μ2T156 phosphorylation and cartoons of the individual components and their recruitment is shown below.