Molecular mechanism for the crystallization of bacteriorhodopsin in lipidic cubic phases.

Crystals of transmembrane proteins may be grown from detergent solutions or in a matrix of membranous lipid bilayers existing in a liquid crystalline state and forming a cubic phase (in cubo). While crystallization in micellar solutions appears analogous to that for soluble proteins, crystallization in lipidic matrices is poorly understood. As this method was shown to be applicable to several membrane proteins, understanding its mechanism will facilitate a rational design of crystallization, minimizing the laborious screening of a large number of parameters. Using polarization microscopy and low-angle X-ray diffraction, experimental evidence is provided to support a mechanistic model for the in cubo crystallization of bacteriorhodopsin in a lipid matrix. Membrane proteins are thought to reside in curved lipid bilayers, to diffuse into patches of lower curvature and to incorporate into lattices which associate to form highly ordered three-dimensional crystals. Critical testing of this model is necessary to generalize it to other membrane proteins.

High-resolution structures and dynamics of membrane protein--lipid complexes: a critique.

Crystallographic analysis of lipidic components in complex with membrane proteins reveals molecules exhibiting well-ordered polar or hydrophobic moieties in contact with protein. As dynamic methods indicate high exchange rates, the interpretation of structural data requires a detailed knowledge of the specificity, affinity and cooperativity of lipid binding.

Role of charged residues at the OmpF porin channel constriction probed by mutagenesis and simulation.

The channel constriction of OmpF porin, a pore protein in the bacterial outer membrane, is highly charged due to the presence of three arginines (R42, R82, and R132) and two acidic residues (D113 and E117). The influence of these charges on ion conductance, ion selectivity, and voltage gating has been studied with mutants D113N/E117Q, R42A/R82A/R132A/D113N/E117Q, and V18K/G131K, which were designed to remove or add protein charge at the channel constriction. The crystal structures revealed no or only local changes compared to wild-type OmpF, thus allowing a comparative study. The single-channel conductance of the isosteric D113N/E117Q variant was found to be 2-fold reduced, and that of the pentuple mutant was 70% of the wild-type value, despite a considerably larger pore cross section. Ion selectivity was drastically altered by the mutations with cation/anion permeability ratios ranging from 1 to 12. Ion flow through these and eight other mutants, which have been characterized previously, was simulated by Brownian dynamics based on the detailed crystal structures. The calculated ion selectivity and relative channel conductance values agree well with the experimental data. This demonstrates that ion translocation through porin is mainly governed by pore geometry and charge, the two factors that are properly represented in the simulations.

Stability of membrane proteins: relevance for the selection of appropriate methods for high-resolution structure determinations.

High stability is a prominent characteristic of integral membrane proteins of known atomic structure. But rather than being an intrinsic property, it may be due to a selection exerted by biochemical procedures prior to structure determination, since solubilization results in the transient exposure of membrane proteins to solution conditions. This may cause structural perturbations that interfere with 3D crystallization and hence with X-ray analysis. This problem also affects the preparation of samples for electron crystallography and NMR studies and may account for the fact that high-resolution structures of representatives of whole groups, such as transport proteins and signal transducers, have not been elucidated so far by any method. A knowledge of the proportion of labile proteins among membrane proteins, and of the kinetics of their denaturation, is therefore necessary. Establishing stability profiles, developing methods to maintain lateral pressure, or preventing contact with water (or both) should prove significant in establishing the structures of conformationally flexible proteins.

Approaches to determining membrane protein structures to high resolution: do selections of subpopulations occur?

Three different methods are currently used for the study of high-resolution structures of membrane proteins: X-ray crystallography, electron crystallography, and nuclear magnetic resonance (NMR) spectroscopy. Thus far, all methods combined have yielded a rather modest number of crystal structures that have been solved at the atomic level. It is hypothesized here that different methods may select different populations of proteins on the basis of various properties. Thus, protein stability may be a significant factor in the formation of three-dimensional (3D) crystals from detergent solutions, since exposure of hydrophobic protein zones to water may cause structural perturbation or denaturation in conformationally labile proteins. This is different in the formation of two-dimensional (2D) crystals where a protein remains protected in its native membrane environment. A biological selection mechanism may therefore be operative in that highly ordered lattices may form only if strong protein-protein interactions are relevant in vivo, thereby limiting the number of proteins that are amenable to electron crystallography. Keeping a protein in a bilayer environment throughout 3D crystallization maintains the lateral pressure existing in native membranes. This can be accomplished by using lipidic cubic phases. Alternatively, the hydrophobic interface of a membrane protein may be spared from contact with water by crystallization from organic solvents where the polar caps are protected in reverse micelles by using appropriate detergents. Some of the criteria that are useful in optimizing the various approaches are given. While the usefulness of complementary methods seems obvious, the results presented may be particularly critical in recognizing key problems in other structural approaches.

Crystallization in cubo: general applicability to membrane proteins.

Obtaining well ordered crystals of membrane proteins is the single most serious stumbling block in the pursuit of their high-resolution structures. The applicability of lipidic cubic phase-mediated crystallization is demonstrated on a diverse set of bacterial membrane proteins: two photosynthetic reaction centres, a light-harvesting complex and two retinal proteins, halorhodopsin and bacteriorhodopsin. Despite marked differences in molecular dimensions, subunit composition and membrane origin, one single lipid, monoolein, is sufficient to form a crystallization matrix for all the aforementioned systems. Therefore, the lipidic cubic phase approach is proposed as a general method for crystallizing membrane proteins.

Oriented channels reveal asymmetric energy barriers for sugar translocation through maltoporin of Escherichia coli.

Sugar transport through maltoporin of Escherichia coli was investigated. This protein facilitates maltooligosaccharide translocation via a binding site in the channel. Because incorporation of the protein into the bilayer results in randomly orientated channels, we re-examined the postulated symmetric translocation model by reconstitution of maltoporin under an externally applied field. Upon binding of bacteriophage lambda, which exploit surface-exposed loops of maltoporin as the receptor, sugar permeation, but not the ion current, was blocked. Thus using the phage-to-probe orientation we were able to show that the channels were approximately 80% directionally inserted into the bilayer. Moreover, asymmetry of the channel was revealed because sugar entrance through the 'open' periplasmic side of maltoporin was similarly reduced. Here a new asymmetrical two-barrier model is presented. Based on liposome-swelling assays and current-fluctuation analysis we conclude that the periplasmic side of the porin shows a two- to threefold higher energy barrier than the extracellular loop-side of the channels.

Folding patterns of membrane proteins: diversity and the limitations of their prediction.

Significantly more high resolution structures of membrane proteins, obtained either by X-ray analysis, electron crystallographic methods or, in the future, by NMR spectroscopy, will be required for reliable structure predictions. Aberrations from the motifs of alpha-helical bundles and beta-barrels occur that are not easily identified by algorithms unless structural homologies exist in the data banks. The coexistence of secondary structure motifs, originally proposed for a neurotransmitter receptor, has now been confirmed for a bacterial iron-siderophore translocating protein (FhuA) by X-ray analysis to 2.7 A resolution. This protein contains a plug domain that has both alpha- and beta-structures.

Protein, lipid and water organization in bacteriorhodopsin crystals: a molecular view of the purple membrane at 1.9 A resolution.

BACKGROUND: Bacteriorhodopsin (bR) from Halobacterium salinarum is a proton pump that converts the energy of light into a proton gradient that drives ATP synthesis. The protein comprises seven transmembrane helices and in vivo is organized into purple patches, in which bR and lipids form a crystalline two-dimensional array. Upon absorption of a photon, retinal, which is covalently bound to Lys216 via a Schiff base, is isomerized to a 13-cis,15-anti configuration. This initiates a sequence of events - the photocycle - during which a proton is transferred from the Schiff base to Asp85, followed by proton release into the extracellular medium and reprotonation from the cytoplasmic side. RESULTS: The structure of bR in the ground state was solved to 1.9 A resolution from non-twinned crystals grown in a lipidic cubic phase. The structure reveals eight well-ordered water molecules in the extracellular half of the putative proton translocation pathway. The water molecules form a continuous hydrogen-bond network from the Schiff-base nitrogen (Lys216) to Glu194 and Glu204 and includes residues Asp85, Asp212 and Arg82. This network is involved both in proton translocation occurring during the photocycle, as well as in stabilizing the structure of the ground state. Nine lipid phytanyl moieties could be modeled into the electron-density maps. Matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) analysis of single crystals demonstrated the presence of four different charged lipid species. CONCLUSIONS: The structure of protein, lipid and water molecules in the crystals represents the functional entity of bR in the purple membrane of the bacteria at atomic resolution. Proton translocation from the Schiff base to the extracellular medium is mediated by a hydrogen-bond network that involves charged residues and water molecules.

Charting the surfaces of the purple membrane.

The preponderance of structural data of the purple membrane from X-ray diffraction (XRD), electron crystallography (EC), and atomic force microscopy (AFM) allows us to ask questions about the structure of bacteriorhodopsin itself, as well as about the information derived from the different techniques. The transmembrane helices of bacteriorhodopsin are quite similar in both EC and XRD models. In contrast, the loops at the surfaces of the purple membrane show the highest variability between the atomic models, comparable to the height variance measured by AFM. The excellent agreement of the AFM topographs with the atomic models from XRD builds confidence in the results. Small technical difficulties in EC lead to poorer resolution of the loop structures, although the combination of atomic models with AFM surfaces allows clear interpretation of the extent and flexibility of the loop structures. While XRD remains the premier technique to determine very-high-resolution structures, EC offers a method to determine loop structures unhindered by three-dimensional crystal contacts, and AFM provides information about surface structures and their flexibility under physiological conditions.

Transmembrane signaling across the ligand-gated FhuA receptor: crystal structures of free and ferrichrome-bound states reveal allosteric changes.

FhuA protein facilitates ligand-gated transport of ferrichrome-bound iron across Escherichia coli outer membranes. X-ray analysis at 2.7 A resolution reveals two distinct conformations in the presence and absence of ferrichrome. The monomeric protein consists of a hollow, 22-stranded, antiparallel beta barrel (residues 160-714), which is obstructed by a plug (residues 19-159). The binding site of ferrichrome, an aromatic pocket near the cell surface, undergoes minor changes upon association with the ligand. These are propagated and amplified across the plug, eventually resulting in substantially different protein conformations at the periplasmic face. Our findings reveal the mechanism of signal transmission and suggest how the energy-transducing TonB complex senses ligand binding.

Stability of trimeric OmpF porin: the contributions of the latching loop L2.

The channel-forming protein OmpF porin from Escherichia coli spans the bacterial outer membrane. Each of the three monomers comprises a hollow, 16-stranded beta-barrel. These are associated to homotrimers which are unusually stable, due mostly to hydrophobic interactions between the beta-barrels. In addition, a loop, L2 connects one subunit to its neighbor by latching into its channel. Residue E71 on loop 2 is integrated into an ionic network and forms salt bridges and hydrogen bonds with R100 and R132 on the channel wall in the adjacent subunit. To examine these contributions quantitatively, six single-site, two double, and one deletion mutant were constructed on the basis of the atomic coordinates of the protein. Differential scanning calorimetric analysis showed that the salt-bridge, E71-R100, contributes significantly to trimer stability: the substitution E71Q causes a decrease of the transition temperature from 72 to 48 degreesC, with DeltaHcal diminishing from 430 to 201 kcal mol-1. A nearby substitution in the loop, D74N, has lesser effects on thermal stability, while the deletion in L2 (Delta69-77) has an effect comparable to that of E71Q. X-ray structure analysis to 3.0 A resolution revealed only local structural differences in the mutants except for the substitution R100A, where another residue, R132, is found to fill the gap left by the truncated side chain of A100. Functional assays in planar lipid bilayers show significantly increased cation selectivities if the charge distribution was affected.

Overexpression of outer membrane porins in E. coli using pBluescript-derived vectors.

The genes coding for four major outer membrane porins of Escherichia coli, ompF, ompC, phoE, and lamB, have been cloned into pBluescript-derived vectors and overexpressed to very high level (approximately 80% of the total membrane protein) in widely used host strains lacking one or more porins. For OmpF, OmpC, and PhoE porins it is shown that, contrary to current dogma, the genes can be overexpressed without undue deleterious effects upon cell growth and are stable, even under conditions of continuous expression. In contrast, overexpression of LamB is toxic to cell growth, but can be performed using tightly regulated lac promotor-driven expression. The vectors described allow overexpression, sequencing, and mutagenesis to be performed using a single system, without the necessity of subcloning, thus simplifying genetic manipulation. A particular advantage of these new vectors (with the exception of the vector for LamB) is that they do not require a particular regime for inducing the recombinant protein. To our knowledge, this study is the only comparative study of widely used membrane porin expression systems and the first to show that several porins can be stably expressed individually and maintained on high copy number vectors.

Inactivation in vitro of the Escherichia coli outer membrane protein FhuA by a phage T5-encoded lipoprotein.

Bacteriophage T5-encoded lipoprotein, synthesized by infected Escherichia coli cells, prevents superinfection of the host cell by this virus. The molecular basis of its ability to inactivate the receptor of phage T5, the FhuA protein, was investigated in vitro. Fully competent T5 lipoprotein, with a His tag attached to the C-terminus, was purified in detergent solution. Coreconstitution with homogeneous FhuA protein into liposomes revealed that the lipoprotein inhibited the irreversible inactivation of phage T5 by FhuA protein. This phenomenon correlated with the inhibition of phage DNA ejection determined by fluorescence monitoring. Addition of detergent abolished the interaction between T5 lipoprotein and FhuA protein. When the signal sequence and N-terminal cysteinyl residue of the lipoprotein were removed by genetic truncation, the soluble polypeptide could be refolded and purified from inclusion bodies. The truncated lipoprotein interfered with infection of E. coli by phage T5, but only at very high concentrations. Circular dichroism spectra of both forms of T5 lipoprotein exhibited predominantly beta-structure. T5 lipoprotein is sufficient for inactivation of the FhuA protein, presumably by inserting the N-terminal acyl chains into the membrane, thus increasing its local concentration. An in vitro stoichiometry of 10:1 has been calculated for the phage-encoded T5 lipoprotein to FhuA protein complex.

Assessing the functionality of a membrane protein in a three-dimensional crystal.

Hexagonal microcrystals of bacteriorhodopsin embedded in a lipidic cubic phase have been investigated by time-resolved FT-IR and resonance Raman spectroscopy. Retinal isomerization, conformational changes in the protein backbone and proton translocation are virtually indistinguishable from those in the native membrane. The protein is thus fully active in three-dimensional crystals.

Coupling site-directed mutagenesis with high-level expression: large scale production of mutant porins from E. coli.

Combination of an origin repair mutagenesis system with a new mutS host strain increased the efficiency of mutagenesis from 46% to 75% mutant clones. Overexpression with the T7 expression system afforded large quantities of proteins from mutant strains. A series of E. coli BE host strains devoid of major outer membrane proteins was constructed, facilitating the purification of mutant porins to homogeneity. This allowed preparation of 149 porin mutants in E. coli used in detailed explorations of the structure and function of this membrane protein to high resolution.

Identification and characterization of two quiescent porin genes, nmpC and ompN, in Escherichia coli BE.

The genomic DNA of the BE strain of Escherichia coli has been scrutinized to detect porin genes that have not been identified so far. Southern blot analysis yielded two DNA segments which proved highly homologous to, yet distinct from, the ompC, ompF, and phoE porin genes. The two genes were cloned and sequenced. One of them, designated ompN, encodes a porin which, due to low levels of expression, has eluded prior identification. The functional properties (single-channel conductance) of the OmpN porin, purified to homogeneity, closely resemble those of the OmpC porin from E. coli K-12. The second DNA fragment detected corresponds to the nmpC gene, which, due to an insertion of an IS1 element in its coding region, is not expressed in E. coli BE.

X-ray structure of bacteriorhodopsin at 2.5 angstroms from microcrystals grown in lipidic cubic phases.

Lipidic cubic phases provide a continuous three-dimensional bilayer matrix that facilitates nucleation and growth of bacteriorhodopsin microcrystals. The crystals diffract x-rays isotropically to 2.0 angstroms. The structure of this light-driven proton pump was solved at a resolution of 2.5 angstroms by molecular replacement, using previous results from electron crystallographic studies as a model. The earlier structure was generally confirmed, but several differences were found, including loop conformations and side chain residues. Eight water molecules are now identified experimentally in the proton pathway. These findings reveal the constituents of the proton translocation pathway in the ground state.

Channel specificity: structural basis for sugar discrimination and differential flux rates in maltoporin.

Maltoporin (LamB) facilitates the diffusion of maltodextrins across the outer membrane of E. coli. The structural basis for the specificity of the channel is investigated by X-ray structure analysis of maltoporin in complex with the disaccharides sucrose, trehalose, and melibiose. The sucrose complex, determined to 2.4 A resolution, shows that the glucosyl moiety is partly inserted into the channel constriction, while the bulky fructosyl residue appears to be hindered to enter the constriction, thus interfering with its further translocation. One of the glucosyl moieties of trehalose is found in a similar position as the glucosyl moiety of sucrose, whereas melibiose appears disordered when bound to maltoporin. A comparison with the previously reported maltoporin-maltose complex sheds light on the basis for sugar discrimination, and explains the different permeation rates observed for the saccharides.

Oligomeric states and siderophore binding of the ligand-gated FhuA protein that forms channels across Escherichia coli outer membranes.

The channel-forming FhuA protein, which translocates ferrichrome across Escherichia coli outer membranes, binds 1 mol ligand/mol monomer in detergent solution. The protein is homogenous and migrates as a single band with a mobility corresponding to 77 kDa in SDS/PAGE electrophoresis. Analytical ultracentrifugation revealed a monodisperse species (s(20,w) = 3.8 S) with a mass of 77,800 +/- 3200 Da. The properties of ligand binding, determined by two independent methods, revealed one binding site/monomer, but are complicated by a pronounced convexity of the Scatchard plot and a Hill coefficient calculated to be 2.5. This strongly suggests that oligomeric species are present. Cross-linking agents revealed the existence of possibly transient, mostly dimeric and trimeric species. The difference between the FhuA protein in detergent solution and in its native membrane environment may be related to the removal of lateral pressure that exists in situ.