Assembly of bacteriophage P22 procapsids requires the participation of approximately 300 molecules of scaffolding protein in addition to the 420 coat protein subunits. In the absence of the scaffolding, the P22 coat protein can assemble both wild-type-size and smaller size closed capsids. Both sizes of procapsid assembled in the absence of the scaffolding protein have been studied by electron cryomicroscopy. These structural studies show that the larger capsids have T = 7 icosahedral lattices and appear the same as wild-type procapsids. The smaller capsids possess T = 4 icosahedral symmetry. The two procapsids consist of very similar penton and hexon clusters, except for an increased curvature present in the T = 4 hexon. In particular, the pronounced skewing of the hexons is conserved in both sizes of capsid. The T = 7 procapsid has a local non-icosahedral twofold axis in the center of the hexon and thus contains four unique quasi-equivalent coat protein conformations that are the same as those in the T = 4 procapsid. Models of how the scaffolding protein may direct these four coat subunit types into a T = 7 rather than a T = 4 procapsid are presented.

Assembly of certain classes of bacterial and animal viruses requires the transient presence of molecules known as scaffolding proteins, which are essential for the assembly of the precursor procapsid. To assemble a procapsid of the proper size, each viral coat subunit must adopt the correct quasiequivalent conformation from several possible choices, depending upon the T number of the capsid. In the absence of scaffolding protein, the viral coat proteins form aberrantly shaped and incorrectly sized capsids that cannot package DNA. Although scaffolding proteins do not form icosahedral cores within procapsids, an icosahedrally ordered coat/scaffolding interaction could explain how scaffolding can cause conformational differences between coat subunits. To identify the interaction sites of scaffolding protein with the bacteriophage P22 coat protein lattice, we have determined electron cryomicroscopy structures of scaffolding-containing and scaffolding-lacking procapsids. The resulting difference maps suggest specific interactions of scaffolding protein with only four of the seven quasiequivalent coat protein conformations in the T = 7 P22 procapsid lattice, supporting the idea that the conformational switching of a coat subunit is regulated by the type of interactions it undergoes with the scaffolding protein. Based on these results, we propose a model for P22 procapsid assembly that involves alternating steps in which first coat, then scaffolding subunits form self-interactions that promote the addition of the other protein. Together, the coat and scaffolding provide overlapping sets of binding interactions that drive the formation of the procapsid.

To clarify the role of phospholipids in G protein-effector interactions of vertebrate phototransduction, transducin activation of cGMP phosphodiesterase (PDE) has been reconstituted on the surface of well-defined phosphatidylcholine (PC) vesicles, using purified proteins from bovine rod outer segments (ROS). PC vesicles enhanced PDE stimulation by the GTP-gamma S-bound transducin alpha subunit (T alpha-GTP gamma S) as much as 17-fold over activation in the absence of membranes. In the presence of 3.5 microM accessible PC in the form of large (100 nm) unilamellar vesicles, 500 nM T alpha-GTP gamma S stimulated PDE activity to more than 70% of the maximum activity induced by trypsin. Activation required PC, PDE, and T alpha-GTP gamma S, but did not require prior incubation of any of the components, and occurred within 4 s of mixing. The PC vesicles were somewhat more efficient than urea-washed ROS membranes in enhancing PDE activation. Half-maximal activation occurred at accessible phospholipid concentrations of 3.8 microM for PC vesicles, and 13 microM for ROS membranes. Titrations of PDE with T alpha-GTP gamma S in the presence of membranes indicated a high-affinity (Kact less than 250 pM) activation of PDE by a small fraction (0.5-5%) of active T alpha-GTP gamma S, as did titrations of ROS with GTP gamma S. When activation by PC vesicles was compared to PDE binding to membranes, the results were consistent with activation enhancement resulting from formation of a T alpha-GTP gamma S-dependent PDE-membrane complex with half-maximal binding at phospholipid concentrations in the micromolar range. The value of the apparent dissociation constant, KPL, associated with the activation enhancement was estimated to be in the range of 2.5 nM (assuming an upper limit value of 1600 phospholipids/site) to 80 nM (for a lower limit value of 50 phospholipids/site). Another component of membrane binding was more than 100-fold weaker and was not correlated with activation by T alpha-GTP gamma S. Low ionic strength disrupted the ability of ROS membranes, but not PC vesicles, to bind and activate PDE. Removal of PDE's membrane-binding domain by limited trypsin digestion eliminated both the binding of PDE to vesicles and the ability of PDE to be activated by T alpha-GTP gamma S and membranes. These results suggest that ROS membrane stimulation of PDE activation by T alpha-GTP gamma S is due almost exclusively to the phospholipids in the disk membrane.

The G protein cascade of vision depends on two peripheral membrane proteins: the G protein, transducin (G t ), and cGMP phosphodiesterase (PDE). Each has covalently attached lipids, and interacts with transduction components on the membrane surface. We have found that their surface interactions are critically dependent on the nature of the lipid. Membranes enhance their protein-protein interactions, especially if electrostatic attraction is introduced with positively charged lipids. These interactions are less enhanced on highly curved surfaces, but are most enhanced by unsaturated or bulky acyl chains. On positively charged membranes, G t assembles at a high enough density to form two-dimensional arrays with short-range crystalline order. Cationic membranes also support extremely efficient activation of PDE by the GTP␥S (guanosine 5-O-(thiotriphosphate)) form of G␣ t (G␣ t -GTP␥S), minimizing functional heterogeneity of transducin and allowing activation with nanomolar G␣ t -GTP␥S. Quantification of PDE activation and of the amount of G␣ t -GTP␥S bound to PDE indicated that G t activates PDE maximally when bound in a 1:1 molar ratio. No cooperativity was observed, even at nanomolar concentrations. Thus, under these conditions, the one binding site for G␣ t -GTP␥S on PDE that stimulates catalysis must be of higher affinity than one or more additional sites which are silent with respect to activation of PDE.Transduction of extracellular signals into intracellular responses by heterotrimeric G proteins (reviewed in Ref. 1) is dominated by protein-protein interactions occurring on the cytoplasmic surfaces of membranes containing heptahelical receptors. G protein subunits convey information between transmembrane receptors and effectors that are also typically either integral or peripheral membrane proteins. Regulators of transduction, such as receptor kinases, arrestin, and its homologues, and GTPase accelerating proteins, are also often localized to the signaling domains of cell membranes. Localization of multiple cascade components may be important for achieving specificity of interactions, and rapid tightly coupled responses.The G protein cascade of vision has served as a useful model system for studying the role of lipid membranes in signal transduction. Both the G protein transducin (G t ) 1 and its effector, cGMP phosphodiesterase (PDE) are normally associated with the surface of disc membranes in rod outer segments, and each contains two covalently attached lipids: farnesyl and geranylgeranyl groups on PDE␣ and PDE␤, respectively (2), and NH 2 -terminal fatty acid (3-6) and COOH-terminal farnesyl (7, 8) on G␣ t and G␥ t , respectively. However, these lipids confer only modest membrane affinity, and both PDE and G t can be selectively extracted without use of detergents, and purified by chromatography. Interactions between activated forms of the ␣ subunit of the G protein, e.g. G␣ t -GTP␥S, and PDE can be reconstituted using the purified proteins, and show a strong membrane dependence (9 -13).The utility of th...

A simple and well-defined system of purified phospholipids and human complement proteins was used to study membrane permeability to macromolecules mediated by the membrane attack complex (MAC) of complement. Large unilamellar vesicles (LUVs) of phosphatidylcholine (PC) or phosphatidylserine (PS) containing trapped macromolecules [bovine pancreatic trypsin inhibitor (BPTI), thrombin, glucose-6-phosphate dehydrogenase (G6PD), and larger molecules] were used to monitor permeability. Membrane permeability to macromolecules was measured by thrombin inhibition by an external inhibitor or by separation of released molecules by gel filtration. Membrane-bound intermediates (C5b-8 or C5b-93) were stable for hours, and macromolecular permeability occurred without fragmentation, fusion, or aggregation of the vesicles. Quantitative membrane binding by C5b-7 as well as essentially quantitative release of thrombin was obtained for PS vesicles. MAC binding to PS-LUVs approximated the theoretical Poisson distribution curve for full release of vesicle contents by one complex per vesicle. Reactions with PC-LUVs occurred with some fluid-phase MAC assembly. Therefore, results from experiments with these vesicles were interpreted in a relative manner. However, the values obtained closely corroborated those obtained with PS-LUVs. At low C9/C5b-8 ratios, the size of the lesion was proportional to the C9 content of the MAC. Half-maximum release of BPTI, thrombin, and G6PD, by a single MAC per vesicle, required approximately 3,5, and 7 C9/C5b-8 (mol/mol), respectively. Larger molecules (greater than or equal to 118-A diameter) were not released from the vesicles. Release of G6PD (95.4-A diameter) required 45% of saturating C9. Therefore, it appeared that the last half of the bound C9 molecules did not increase pore size and the pore which released G6PD approached the diameter of the closed circular lesion measured (by others) in electron micrographs (approximately 100 A). The results were consistent with the formation of a stable membrane pore by a single complex per vesicle in which C9 molecules line only one side of the pore at low C9/C5b-8 ratios and maximum pore size is attained by incomplete, noncircular polymers of C9.