Phosphatidylinositol 4,5-biphosphate (PIP(2)) lipids regulate the phosphorylation of syntaxin N-terminus by modulating both its position and local structure

Abstract

Syntaxin (STX) is a N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) protein that binds to the plasma membrane and regulates ion channels and neurotransmitter transporters. Experiments have established the involvement of the N-terminal segment of STX in direct protein-protein interactions and have suggested a critical role for the phosphorylation of serine 14 (S14) by casein kinase-2 (CK2). Because the organization of STX in the plasma membrane was shown to be regulated by phosphatidylinositol 4,5-biphosphate (PIP(2)) lipids, we investigated the mechanistic involvement of PIP(2) lipids in modulating both the membrane interaction and the phosphorylation of STX, using a computational strategy that integrates mesoscale continuum modeling of protein-membrane interactions, with all-atom molecular dynamics (MD) simulations. Iterative applications of this protocol produced quantitative evaluations of lipid-type demixing due to the protein and identified conformational differences between STX immersed in PIP(2)-containing and PIP(2)-depleted membranes. Specific sites in STX were identified to be important for the electrostatic interactions with the PIP(2) lipids attracted to the protein, and the segregation of PIP(2) lipids near the protein is shown to have a dramatic effect on the positioning of the STX N-terminal segment with respect to the membrane/water interface. This PIP(2)-dependent repositioning is shown to modulate the extent of exposure of S14 to large reagents representing the CK2 enzyme and hence the propensity for phosphorylation. The prediction of STX sites involved in such PIP(2)-dependent regulation of STX phosphorylation at S14 offers experimentally testable probes of the mechanisms and models presented in this study, through structural modifications that can modulate the effects.

Figure 1

Sequence alignment of two human syntaxin 1 isoforms: Syntaxin 1A (top row) and Syntaxin 1B (bottom row). Sequence identity of these two isoforms is 83%. For completeness, the alignment also shows the location of different domains of syntaxin.

Figure 2

The “closed” state model of syntaxin was built by combining two high resolution X-ray structures (3HD7 and 3C98) with the model of the N-terminus predicted with Rosetta. (A) In 3HD7, the stretch of residues 188–286 of syntaxin 1A (yellow) forms an extended intertwined helical bundle with other SNARE proteins, Snap-25 and synaptobrevin-2 (cyan). The structural information from 3HD7 was used to model residues 188–245 (shown in space fill rendering in yellow in panel A). (B) In 3C98, syntaxin 1A is in the “closed” state (yellow) and in complex with Munc18a protein (cyan). Residue segments 2–9 and 27–248 have been resolved in this structure and are shown without Munc18a. Panel B also illustrates how the “closed” state model of STX (pink) superimposes on the structural model in 3C98. (C) Fold prediction from Rosetta for the N-terminus region of STX. Upper and lower panels show alignments of the predicted structure for the N-terminal 1–26 residue stretch (pink) with X-ray structures of syntaxin 1 and syntaxin 4 N-peptides (yellow), respectively. Crystallographic models for syntaxin 1 (residues 2–9) and syntaxin 4 (residues 1–19) N-peptides were taken from PDBid 3C98 and 2PJX, respectively. The root-mean-square deviation (rmsd) for the aligned helical regions was 1 Å for syntaxin 1 and 2.3 Å for syntaxin 4. (D) Orientation of the “closed” state STX (cartoon) with respect to the lipid membrane (lines). Intracellular and extracellular leaflets are top and bottom leaflets respectively. The membranes consisted of either mixture of neuronal lipids (5:45:50 PIP₂/POPE/POPC on the intracellular leaflet and 30:70 sphingomyelin/POPC on the extracellular leaflet, referred to as the PM membrane in the text) or contained only POPC lipid, ∼800 lipids in total. (E) Electrostatic potential isosurfaces for STX at +25 eV (positive, blue) and −25 eV (negative, red) levels in the same orientation as in panel D. The electrostatic calculation was performed using APBS software version 4.0.

Figure 3

Multiscale modeling of the STX embedded in PM (neuronal plasma membrane model). Panels (A, B) and (C, D) depict coarse-grained and all-atom representations of the system, respectively. In all the panels STX is shown in cartoon with Lys and Arg residues highlighted in licorice. (A) Steady-state distribution of PIP₂ lipids (colored shades represent ratios of local φ and average φ₀ lipid fraction values) from the CGM minimization procedure. STX_1–246 was used for this calculation (see text), and the intracellular (IC) membrane leaflet facing STX started with a uniform composition of φ₀^PIP2 = 0.05 and φ₀^POPC+POPE = 0.95. For clarity, only the part of the bilayer leaflet near the protein is shown. (B) Similar to panel (A) but showing the segment of the IC leaflet for which the CGM calculation predicts an excess of PIP₂ lipids (φ/φ₀ >1) at equilibrium. Several key Lys/Arg residues are highlighted. (C) Initial placement of PIP₂ lipids (shown in space-fill) in the all-atom construct of STX in PM. On the basis of the findings from the CGM minimization, one PIP₂ (in green) was placed within 2 Å of the R261/K264 residue pair. Two PIP₂ lipids (in yellow) were located 6 Å away from the protein, and two other PIP₂ lipids (in pink) were positioned 10 Å away from STX. Bilayer leaflets are indicated by the phosphate atoms in their lipid head-groups (shown in gold), and the rest of the system is removed for clarity. (D) Zoom-in on the region where the strongest electrostatic interaction between STX and PIP₂ is expected. Color code is the same as in (C).

Figure 4

STX interacts simultaneously with five PIP₂ molecules. (A) Snapshot after 90 ns of atomistic MD simulations of STX/PM complex, showing five different PIP₂ lipids (in space fill representation) bound to STX (in cartoon). Key basic residues of STX are highlighted, and PIP₂ lipids are numbered for designation purposes. Membrane leaflets are traced by their lipid phosphate atoms (in gold) and for clarity, the rest of the simulated system is removed. (B) Time-evolution of the cumulative number of PIP₂ molecules within 3.5 Å of STX, depicted and numbered as in A. (C) Time-evolution of the minimum distances between STX and five PIP₂ lipids.

Figure 5

Structural elements of STX responsible for PIP₂ lipid sequestration. (A, B) Two PIP₂ molecules (“3” and “4”, see also Figure 4A) are held near STX by interactions with M1/K2/D3 residues in the N-terminus and R261/K264 in the linker region. (A) Time-evolution of the minimum distance between amino groups of M1 (red), K2 (green), and D3 (blue) and PIP₂ molecule “3”. (B) Snapshot after 90 ns of atomistic MD simulations illustrating electrostatic interactions with M1/K2/D3 (drawn in ball and stick) and R261/K264 (licorice). The STX fragment is depicted in cartoon, and PIP₂ lipids “3” and “4” are represented in space-fill. (C–G) STX TM/linker helical region undergoes conformational change upon binding of PIP₂ lipids. (C) Time-evolution of changes in the 246–288 helix around K264; the kink (upper panel) and face shift (lower panel) angles are calculated with ProKink. (D–F) Snapshots of STX TM/linker helix and neighboring PIP₂ lipids (“4” and “5” as designated in Figure 4A) taken at trajectory time-points t = 0 ns (D), t = 40 ns (E), and t = 90 ns (F). The helix is shown in cartoon, and K255, K259, and R262 are drawn in licorice. (G) Superposition of conformations of the STX TM/linker helical segment at t = 0 ns (cyan), t = 40 ns (blue), and t = 90 ns (orange) time-points highlighting the face shift in the helix. K255/K259/R262 residues are depicted in licorice.

Figure 6

Results from iteration 2 between CGM minimization and MD equilibration: The steady-state distribution of PIP₂ lipids around STX obtained from the CGM minimization scheme using the STX conformation after 90 ns of all-atom MD simulations (shown in Figure 4A). Calculated as in Figure 3, but with an initially uniform composition of the membrane characterized by φ₀^PIP2 = 0.05 and φ₀^POPC+POPE = 0.95. (A) Only part of the membrane leaflet near the protein is shown and the CGM solution (color shades) represents ratios of local φ and average φ₀ lipid fraction values. (B) Similar to panel A, but showing only regions on the membrane leaflet with excess PIP₂ lipids (φ/φ₀ > 1).

Figure 7

Different interactions of the N-terminus fragment in wild type (STX_WT) and K2A mutant (STX_K2A) STX with PIP₂ lipids result in distinct conformational changes. (A) Snapshot after 90 ns of MD trajectory of STX_K2A (cartoon) highlighting interactions between sequestered PIP₂ lipids and residues in the N-terminal and linker regions. Key positions (R4, R25, K255, K259, R262, K263, K264) are shown in licorice. PIP₂ molecules and the S14 residue in the STX_K2A N-terminus are shown in space-fill. The panel inset describes the evolution in time of minimum distances between PIP₂ and residues R4 and R25. (B, C) Differential positioning of S14 (space-fill) in STX_WT (in B) and STX_K2A (in C). The snapshots were taken after 90 ns of MD simulations in the PM bilayer. The proteins are drawn in cartoon, and the lipid membrane is represented by the location of the lipid headgroup phosphate atoms (gold). (D) Evolution of the secondary structure of N-terminus region (residues 1–30) in STX_WT (lower panel) and STX_K2A (upper panel) during the simulations. Colors indicate the type of secondary structure: blue for 3-helix, and purple for 4-helix. Changes in colors over trajectory time indicate that the corresponding residue is in one or the other secondary structure at a particular time-point; the red ovals highlight the region near S14 residue, which is seen to assume different secondary structure in the wild type and the mutant STX. The secondary structure assessment was carried out with the Simulaid software.

Figure 8

(A) Solvent exposure of the S14 residue in the N-terminus of STX. The profile of the solvent accessible surface area (SASA) of S14 (calculated as described in Methods) is shown as a function of the probe radius, r_p, in three MD trajectories: Wild type STX (STX_WT) in PM (open circles), STX_WT in POPC (filled squares), and K2A STX mutant (STX_K2A) in PM (open triangles). The error bars indicate the standard deviations. (B–D) Atomic detail SASA maps for STX_WT simulated in PM and measured with different probe radii: r_p = 1.4 Å (in B), r_p = 5.4 Å (C), and r_p = 9.4 Å (D). STX_WT in these representative snapshots is shown in space-fill, and each atom of the protein is colored according to its SASA value (see color code). The location of S14 is highlighted in each panel.

Figure 9

Residues S14 and D17 exhibit different interaction pattern in STX_WT and STX_K2A. (A, B) Views of the N-terminus region of STX_K2A (in A) and STX_WT (in B) interacting with the PM membrane. S14 and D17 are shown in red and yellow space-fill representations, respectively. (C) Minimum distance between S14 and D17 residues as a function of time in simulations of STX_WT (solid) and STX_K2A (dashed). (D) SASA calculated for D17 residue with probes of different radii, in STX_WT (circles) and STX_K2A (triangles) simulations.

Figure 10

The N-terminus of STX_WT in PM and POPC membranes assumes different positions relative to both the lipid bilayer and the linker region. (A) STX_WT in the POPC membrane. Representation is the same as in Figure 8; S14 is shown in space-fill. (B, C) Superposition of STX_WT (cartoon) structures in PM (cyan) and POPC (purple) shown in two different views. S14 and K263 residues are highlighted in space-fill representations (K263 is at the interface between the linker and the TM segment). The two structures were superimposed by the backbone atoms of the 246–288 residues. (D) Distance between C_α atoms of S14 and K263 as a function of time plotted for the later parts of the respective trajectories.