Domain organization and conformational plasticity of the G protein effector, PDE6

Abstract

The cGMP phosphodiesterase of rod photoreceptor cells, PDE6, is the key effector enzyme in phototransduction. Two large catalytic subunits, PDE6α and -β, each contain one catalytic domain and two non-catalytic GAF domains, whereas two small inhibitory PDE6γ subunits allow tight regulation by the G protein transducin. The structure of holo-PDE6 in complex with the ROS-1 antibody Fab fragment was determined by cryo-electron microscopy. The ∼11 Å map revealed previously unseen features of PDE6, and each domain was readily fit with high resolution structures. A structure of PDE6 in complex with prenyl-binding protein (PrBP/δ) indicated the location of the PDE6 C-terminal prenylations. Reconstructions of complexes with Fab fragments bound to N or C termini of PDE6γ revealed that PDE6γ stretches from the catalytic domain at one end of the holoenzyme to the GAF-A domain at the other. Removal of PDE6γ caused dramatic structural rearrangements, which were reversed upon its restoration.

Keywords: conformational change; cryo-electron microscopy; phosphodiesterases; phototransduction; retina; tertiary structure.

FIGURE 1.

Three-dimensional structure of PDE6. A, a field of PDE6 suspended in vitreous ice. Scale bar, 200 Å. B, comparison of model projections (left) and corresponding class averages (right). C, three-dimensional map of PDE6. The map is filtered to 22 Å and displayed at an isosurface threshold approximating the PDE6 molecular mass of 220 kDa. Scale bar, 40 Å.

FIGURE 2.

Three-dimensional structure of PDE6·ROS-1-Fab. A, a field of PDE6·ROS-1-Fab complexes suspended in vitreous ice. Examples of front, side, and top/bottom views are indicated with circles, hexagons, and squares, respectively. Scale bar, 200 Å. Inset, class averages of front and top/bottom views, with extra density indicated by arrows. B, three-dimensional maps with no symmetry (C₁) or 2-fold rotational symmetry (C₂) imposed during reconstruction. Three different views of the reconstruction are shown at 100% isosurface volume according to the molecular mass of the PDE6·ROS-1-Fab complex (320 kDa). Lines shown in the bottom view (iii) indicate the plane of the page in i and ii. Scale bar, 40 Å. C, enlarged view of the GAF domains in the C₁ and C₂ maps shown at a higher density threshold level. D, distribution of particle orientations. E, gold standard FSC curves for PDE6·ROS-1-Fab C₁ and C₂ maps indicate a resolution of ∼18.4 and ∼17.3 Å, respectively, at FSC = 0.5 and ∼11 Å at FSC = 0.143.

FIGURE 3.

Fab fitting in the PDE6·ROS-1-Fab structure. A, superposition of the PDE6·ROS-1-Fab C₁ map, filtered to 22 Å (gray), with the PDE6 map (cyan). B, crystal structure of an Fab fragment (PDB ID 12E8 (62)) with surface low-pass filtered to 11 Å (top). Fitting of the Fab structure into the PDE6·ROS-1-Fab C₁ and C₂ maps (middle, bottom). Complementarity-determining regions for 12E8 are shown in green.

FIGURE 4.

Orientation of the GAF domains. A, comparison of the PDE2A (PDB 3IBJ (11)) and PDE6·ROS-1-Fab structures. Left, the PDE6 C₂ model shown with fitting of the intact PDE2A GAF domains (purple). Right, x-ray structure of PDE2A at 11 Å with GAF domain sequences also shown as ribbons (purple). B, independent fitting of GAFa domains from chicken cone PDE6C (PDB 3DBA (9)) (orange, blue) and GAFb domains from PDE2A (cyan, green). C, alignment of GAFa domain structures from PDE2A (pink) and PDE6C (orange) crystals, showing the location of the non-catalytic cGMP binding site in PDE6C (blue).

FIGURE 5.

Orientation of the catalytic domain. Front (A and C) and bottom (B and D) views of the PDE6·ROS-1-Fab map with fitting of catalytic domain crystal structures. A and B, the PDE5/6 chimera structure (PDB 3JWR (10)) is shown as a pink ribbon, with the N terminus (yellow spheres), C terminus (red spheres), and the PDE6γ-(71–87) peptide (blue) indicated. The Fab structure is as in Fig. 3. C and D, enlarged view of the catalytic domain, with the crystal structures of the PDE5/6 chimera (pink) and PDE5A/GMP (PDB 1T9S (8)) (yellow) aligned. GMP is shown in cyan.

FIGURE 6.

Comparison of PDE6 and PDE2A domain organization. The PDE6αβ atomic model is composed of the GAFa domain from chicken cone PDE6C (PDB 3DBA (9)), GAFb domain from PDE2A (PDB 3IBJ (11)), and catalytic domain from PDE5/6 (PDB 3JWR (10)), fit as in Figs. 3 and 4. Selected homologous sequences are marked in matching colors to guide the eye. The approximate changes in relative orientation between the catalytic and GAFa domains, relative to GAFb, are indicated.

FIGURE 7.

Three-dimensional structure of PDE6·PrBP/δ. A, left, a typical field of PDE6·PrBP/δ complexes suspended in vitreous ice. Scale bar, 200 Å. Right, gallery of individual particles in top/bottom view orientation, with one (i) or two (ii) PrBP/δ bound (yellow arrows). B, comparison of class averages (right) and corresponding projections (left) of the final model. C, front and bottom views of the three-dimensional model of PDE6·PrBP/δ low-pass filtered to 30 Å. Scale bar, 40 Å. The crystal structure of PrBP/δ (purple; PDB ID 1KSH (46)) was fit into the extra mass at the base of the catalytic domain.

FIGURE 8.

Localization of the N- and C terminus of PDE6γ. A, reconstructions of tPDE6 reconstituted with HA-PDE6γ or PDE6γ-HA and complexed with HA-Fab. Ribbon diagrams of an Fab structure (PDB 12E8 (62)) are shown in purple. B, the distance between the HA epitopes in PDE6 reconstituted with HA-PDE6γ and PDE6γ-HA was estimated by superimposing the two maps.

FIGURE 9.

Reversible conformational change in PDE6 upon removal of PDE6γ. A, example field and single particle gallery of tPDE6 imaged in negative stain. Scale bar, 200 Å. Examples of bell- and donut-shaped particles are indicated by circles and squares, respectively. B, class averages of tPDE6 and PDE6 or tPDE6 in the presence of excess recombinant PDE6γ. C, Coomassie-stained gel of PDE6 before and after trypsin treatment and reconstituted with recombinant PDE6γ. D, single particle Multirefine procedure starting with ∼4100 particles and three initial models yielded three similarly populated subsets of particles. The models (with C₂ symmetry imposed) were low-pass filtered to 30 Å and are shown in three orthogonal views. E, C₁ and C₂ maps constructed by single particle tomography using 120 sub-volumes are shown in the same orientations as in D. F, comparison of PDE6 and tPDE6 maps from samples in negative stain. Three-dimensional maps (with C₂ symmetry) are shown in an isosurface representation with the contour level approximating the PDE6 and tPDE6 molecular mass of 220 and 200 kDa, respectively. The atomic model of PDE6αβ, as shown in Fig. 6, is docked into the PDE6 map. The tPDE6 map is shown with a speculative rearrangement of the domains.

FIGURE 10.

Model of PDE6 membrane interaction. A and B, PDE6 model as described in Fig. 6. The C-terminal residue in the catalytic domain structure is shown in red, and the PDE6γ-(71–87) peptide in blue. A, bottom view with PDE6γ rendered transparent to reveal the inhibitor 3-isobutyl-1-methylxanthine (yellow) bound in the active site. B, front view with diagram of the C-terminal residues not present in the catalytic domain crystal (thick black line) and attached farnesyl and geranylgeranyl groups. The remainder of PDE6γ (residues 1–70) is drawn as a blue line. C, proposed model of PDE6 activation. Catalytic domains are rotated 90 ° in opposite directions, about their N termini, to expose the catalytic sites. PDE6γ-(50–87) is shown in the crystal structure of the complex with G_αt (purple) and an RGS-9 fragment (brown) (PDB 1FQJ (5)).