Membrane docking geometry of GRP1 PH domain bound to a target lipid bilayer: an EPR site-directed spin-labeling and relaxation study

Abstract

The second messenger lipid PIP(3) (phosphatidylinositol-3,4,5-trisphosphate) is generated by the lipid kinase PI3K (phosphoinositide-3-kinase) in the inner leaflet of the plasma membrane, where it regulates a broad array of cell processes by recruiting multiple signaling proteins containing PIP(3)-specific pleckstrin homology (PH) domains to the membrane surface. Despite the broad importance of PIP(3)-specific PH domains, the membrane docking geometry of a PH domain bound to its target PIP(3) lipid on a bilayer surface has not yet been experimentally determined. The present study employs EPR site-directed spin labeling and relaxation methods to elucidate the membrane docking geometry of GRP1 PH domain bound to bilayer-embedded PIP(3). The model target bilayer contains the neutral background lipid PC and both essential targeting lipids: (i) PIP(3) target lipid that provides specificity and affinity, and (ii) PS facilitator lipid that enhances the PIP(3) on-rate via an electrostatic search mechanism. The EPR approach measures membrane depth parameters for 18 function-retaining spin labels coupled to the PH domain, and for calibration spin labels coupled to phospholipids. The resulting depth parameters, together with the known high resolution structure of the co-complex between GRP1 PH domain and the PIP(3) headgroup, provide sufficient constraints to define an optimized, self-consistent membrane docking geometry. In this optimized geometry the PH domain engulfs the PIP(3) headgroup with minimal bilayer penetration, yielding the shallowest membrane position yet described for a lipid binding domain. This binding interaction displaces the PIP(3) headgroup from its lowest energy position and orientation in the bilayer, but the headgroup remains within its energetically accessible depth and angular ranges. Finally, the optimized docking geometry explains previous biophysical findings including mutations observed to disrupt membrane binding, and the rapid lateral diffusion observed for PIP(3)-bound GRP1 PH domain on supported lipid bilayers.

Figure 1. The GRP1 PH domain and positions chosen for spin labeling.

(A) Domain topology, illustrating the secondary structure of GRP1 PH domain and the location of the PIP₃ headgroup analogue (IP₄) in the crystal structure of the co-complex (1FGY [22]). (B) The 18 sites selected for spin-labeling (blue spheres indicate Cα atoms), showing the high density of probe positions on the membrane docking face to provide optimal EPR analysis of the docking geometry. Figures generated in PyMol (DeLano Scientific LLC).

Figure 2. Effect of spin labeling on target membrane binding.

Shown are representative competitive displacement curves for three GRP1 PH domains: Wild Type, Cysless and V278R1. Each PH domain was added to PC∶ PS∶ PIP₃∶ dansylPE (mole ratios 70∶ 23∶ 2∶ 5) target membrane and allowed to form the PIP₃-protein complex on the membrane surface. Subsequently, using a standard competition assay , , , the competitive inhibitor IP₆ was titrated into the sample, thereby displacing PH domain from the membrane as revealed by decreasing protein-to-membrane FRET. The resulting competition curve was best fit for a homogeneous population of PIP₃/IP₆ binding sites (solid curves) to determine the K_i for IP₆. Table 1 summarizes the measured K_i(IP₆) values, which are directly proportional to the affinity of each PH domain for membrane-embedded PIP₃. Experimental conditions: 0.2 µM PH domain and 200 µM total lipid in 25 mM HEPES, 140 mM KCl, 15 mM NaCl, 0.5 mM MgCl₂, pH 7.4, 25°C.

Figure 3. Control EPR spectra for a representative mutant.

Shown are reproducible EPR spectral overlays for the MTSSL spin-labeled GRP1 PH domain V278R1, illustrating the strategy employed to analyze the spectral effects of membrane docking. (A) V278R1 PH domain in the absence and presence of control PC∶ PS (3∶1) membranes lacking PIP₃, illustrating spectral broadening due to nonspecific membrane association. (B) V278R1 PH domain saturated with 200 µM IP₆, both in the absence and presence of control PC∶ PS (3∶1) membranes, showing that unlike the apo PH domain the IP₆-PH domain complex does not bind nonspecifically to membranes when PIP₃ is absent. (C) V278R1 PH domain saturated with 200 µM IP₆, both in the absence and presence of target PC∶ PS∶ PIP₃ (74∶ 24∶ 2) membranes, showing the spectral change upon docking of the IP₆-PH domain complex to membrane-bound PIP₃ (with release of IP₆). This is the standard comparison carried out for all spin-labeled PH domains (see Fig. 4), since the free IP₆-PH domain complex does not dock to background lipids and use of this complex as a reference point ensures that spectral changes are due to the environmental effects of membrane docking, rather than to the conformational effects of ligand binding cleft occupancy. (D) V278R1 PH domain binding to target PC∶ PS∶ PIP₃ (74∶ 24∶ 2) membranes in the absence and presence of saturating 200 µM IP₆, showing that the competitive inhibitor IP₆ does not perturb PH domain binding to target membrane PIP₃ under these conditions. Each pair of overlayed spectra were obtained for two samples made from the same protein stock to ensure nearly identical spin concentrations, for which the same number of scans were collected and plotted in absolute intensity mode. Double integrations confirmed that each pair of spectra represented virtually identical numbers of spins. Thus, the relative intensities of each spectral pair can be directly compared. Spectra were acquired at 23°C and samples contained 10–200 µM protein, 0 or 40 mM total lipid as SUVs, and 0 or 200 µM IP₆, in 25 mM HEPES, 140 mM KCl, 15 mM NaCl, 0.5 mM MgCl₂, pH 7.4.

Figure 4. Effect of Target Membrane Docking on EPR Spectra.

Each spectral overlay shows the effects of target membrane docking on the EPR spectrum of a given MTSSL spin-labeled GRP1 PH domain. The free PH domain was saturated with 200 µM IP₆ and spectra were acquired in the absence and presence of target PC∶ PS∶ PIP₃ (74∶ 24∶ 2) membranes. A spectral change is observed when the free IP₆-PH domain complex docks to a target PIP₃ headgroup on the membrane surface, releasing IP₆. Since the ligand binding cleft is occupied in both states the spectral changes are triggered primarly by membrane docking rather than by cleft occupancy (see Figure 3). Table 1 qualitatively ranks the magnitudes of the target membrane-induced spectral changes (++, +, −). Each pair of overlayed spectra were collected as described in the Figure 3 legend thus their relative intensities can be directly compared. Spectra were acquired at 23°C and samples contained 10–200 µM protein, 0 or 40 mM total lipid as SUVs, and 200 µM IP₆ in 25 mM HEPES, 140 mM KCl, 15 mM NaCl, 0.5 mM MgCl₂, pH 7.4.

Figure 5. Hyperbolic relationship between spin label depth parameters and membrane penetration depths in the optimized, self-consistent EPR docking model.

As described in Methods, the crystal structure of the GRP1 PH domain co-complex with IP₄ (1FGY [22]) was modeled with MTSSL spin labels at the 18 chosen positions, then docked to the target bilayer using an interactive procedure that optimizes the known hyperbolic relationship between the measured spin label EPR depth parameters and the calculated spin label membrane penetration depths. Shown are the measured depth parameters for the protein spin labels (filled symbols) and the calibration lipid spin labels (open symbols), as well as the calculated membrane depth for each spin label in the final optimized, self-consistent EPR membrane docking model (Figure 6). The excellent agreement with the best-fit hyperbola (solid curve) emphasizes the high quality of the docking model. Depth parameters were measured by EPR power saturation (Methods) at 23°C and samples contained 10–200 µM protein, 40 mM total lipid as SUVs, 25 mM HEPES, 140 mM KCl, 15 mM NaCl, 0.5 mM MgCl₂, pH 7.4. Except where otherwise indicated, errors are propagated from the errors of the accessibility parameters (Π(NiEDDA) and Π(O₂)) used to calculate the depth parameter (Eq. 1), n≥15 power settings were used for each accessibility parameter measurement.

Figure 6. Protein-membrane interactions in the optimized, self-consistent EPR docking model.

Shown is the optimized, self-consistent EPR docking model for GRP1 PH domain co-complexed with IP₄ (1FGY [22]) and docked to a target bilayer. The schematic target bilayer highlights transient positions of backbone phosphates (red-brown spheres) and headgroups (PC or PS, black spheres) from a snapshot of a simulated bilayer . (A) Views of the PIP₃ headgroup relative to the mean backbone phosphate plane in both its lowest energy conformation (left) and its PH domain-bound conformation (right), illustrating the effect of PH domain binding on the target headgroup depth and orientation. (B) The PH domain docked to the schematic target bilayer in the optimized geometry. (C) Basic residues of the PH domain (dark blue spheres for R277, K279, K282, R283, R322, K323, R349) that can contact the negatively charged target bilayer in the optimized docking geometry. In some cases, the indicated side chain rotomer was adjusted to enhance membrane contact. (D) Hydrophobic and polar residues (light blue spheres for V278, T280, W281, P321, A346) that can contact the bilayer. Y298 obstructs the view and is not shown; it also contacts the bilayer and, perhaps more importantly, contacts multiple side chains responsible for specific PIP₃ binding. (E) Acidic residues (red spheres for D320, E345, D347) that contact the anionic bilayer surface and are thus proposed to limit protein penetration into the target bilayer.