Expression, purification, and characterization of Thermotoga maritima membrane proteins for structure determination

Structural studies of integral membrane proteins typically rely upon detergent micelles as faithful mimics of the native lipid bilayer. Therefore, membrane protein structure determination would be greatly facilitated by biophysical techniques that are capable of evaluating and assessing the fold and oligomeric state of these proteins solubilized in detergent micelles. In this study, an approach to the characterization of detergent-solubilized integral membrane proteins is presented. Eight Thermotoga maritima membrane proteins were screened for solubility in 11 detergents, and the resulting soluble protein-detergent complexes were characterized with small angle X-ray scattering (SAXS), nuclear magnetic resonance (NMR) spectroscopy, circular dichroism (CD) spectroscopy, and chemical cross-linking to evaluate the homogeneity, oligomeric state, radius of gyration, and overall fold. A new application of SAXS is presented, which does not require density matching, and NMR methods, typically used to evaluate soluble proteins, are successfully applied to detergent-solubilized membrane proteins. Although detergents with longer alkyl chains solubilized the most proteins, further characterization indicates that some of these protein-detergent complexes are not well suited for NMR structure determination due to conformational exchange and protein oligomerization. These results emphasize the need to screen several different detergents and to characterize the protein-detergent complex in order to pursue structural studies. Finally, the physical characterization of the protein-detergent complexes indicates optimal solution conditions for further structural studies for three of the eight overexpressed membrane proteins.     

Ciliary membrane tubulin and associated proteins: a complex stable to Triton X-114 dissociation

When either membranes from scallop gill cilia or reconstituted membranes from the same source are solubilized with Triton X-114 and the detergent is condensed by warming, no significant fraction of any major membrane protein partitions into the micellar detergent. Rather, most of the membrane lipids condense with the detergent phase, forming mixed micelles from which nearly pure lipid vesicles may be produced by adsorption of detergent with polystyrene beads. One minor membrane protein, with a molecular weight of about 20 000, is associated consistently with these vesicles. The aqueous phase contains a fairly homogeneous protein-Triton X-114 micelle sedimenting at 2.6 S in the analytical ultracentrifuge. Sucrose gradient velocity analysis in a detergent-free gradient indicates moderate size polydispersity but constant polypeptide composition throughout the sedimenting protein zone. Sucrose gradient equilibrium analysis (also in a detergent-free gradient) results in a protein-detergent complex banding at a density of 1.245 g/cm3. Sedimentation of the protein-detergent complex in the ultracentrifuge, followed by fixation and normal processing for electron microscopy, reveals a fine, reticular material consisting of 5-10-nm granules. These data are consistent with previous evidence that membrane tubulin and most other membrane proteins exist together as a discrete lipid-protein complex in molluscan gill ciliary membranes.     

Ciliary membrane tubulin and associated proteins: a complex stable to Triton X-114 dissociation

When either membranes from scallop gill cilia or reconstituted membranes from the same source are solubilized with Triton X-114 and the detergent is condensed by warming, no significant fraction of any major membrane protein partitions into the micellar detergent. Rather, most of the membrane lipids condense with the detergent phase, forming mixed micelles from which nearly pure lipid vesicles may be produced by adsorption of detergent with polystyrene beads. One minor membrane protein, with a molecular weight of about 20000, is associated consistently with these vesicles. The aqueous phase contains a fairly homogeneous protein-Triton X-114 micelle sedimenting at 2.6 S in the analytical ultracentrifuge. Sucrose gradient velocity analysis in a detergent-free gradient indicates moderate size polydispersity but constant polypeptide composition throughout the sedimenting protein zone. Sucrose gradient equilibrium analysis (also in a detergent-free gradient) results in a protein-detergent complex banding at a density of 1.245 g/cm³. Sedimentation of the protein-detergent complex in the ultracentrifuge, followed by fixation and normal processing for electron microscopy, reveals a fine, reticular material consisting of 5–10-nm granules. These data are consistent with previous evidence that membrane tubulin and most other membrane proteins exist together as a discrete lipid-protein complex in molluscan gill ciliary membranes.     

Characterization of triton X-100-solubilized prostaglandin E binding protein of rat liver plasma membranes.

Rat liver plasma membranes bind prostaglandins E1 and E2 (PGE) with high affinity and specificity. We have solubilized plasma membranes, prelabeled with radioactive PGE1, in water solutions of Triton X-100. We sedimented this material into sucrose density gradient containing H2O and D2O. From numerical integration of the sedimentation equation, taking explicitly into account the density and viscosity gradients present during the centrifugation, we have determined a value of s20,w = 5.6 to 5.7 X 10(-13) s and a partial specific volume, v = 0.80 to 0.81 cm3/g, for the PGE binding protein-Triton X-100 composed of 60% (w/w) protein and 40% (w/w) detergent. Gel filtration in water solutions of Triton X-100 gives a Stokes radius of 53 A for the complex. These data imply a molecular weight of 105,000 for the detergent-free binding protein and a frictional ratio of 1.3 for the complex. If the detergent is bound to the protein in a monolayer, about 40% of the PGE binding protein's surface would be covered with detergent. The procedures used in the analysis of the sedimentation behavior of the PGE binding protein-detergent complex, when coupled with a gel filtration measurement of the Stokes radius, allow valid determination of the size, shape, and extent of detergent binding of a wide variety of membrane proteins, even when they are present as minor components of complex mixtures.     

Gel chromatography and analytical ultracentrifugation to determine the extent of detergent binding and aggregation, and Stokes radius of membrane proteins using sarcoplasmic reticulum Ca2+-ATPase as an example

For structural studies of integral membrane proteins, including their 3D crystallization, the judicious use of detergent for solubilization and purification is required. Detergent binding by the solubilized protein is an important parameter to determine the hydrodynamic properties in terms of size and aggregational (monomeric/oligo(proto)meric) state of the protein. Detergent binding can be measured by gel filtration chromatography under equilibrium conditions and after separation from mixed micelles of solubilized lipid and detergent. Using sarcoplasmic reticulum Ca(2+)-ATPase as an example, we demonstrate in this protocol complete procedures for measurement of detergent binding using (i) radiolabeled n-dodecyl-beta-D-maltoside (DM) or (ii) from measurements of the increase in refractive index due to the presence of bound detergent on the protein. The latter measurement can also be performed by sedimentation velocity (SV) analysis in the analytical ultracentrifuge which in addition allows determination of the sedimentation coefficient. In combination with estimation of Stokes radius by gel filtration calibration, the molecular mass and asymmetry of the solubilized protein can be calculated. In the proposed protocols, the gel chromatographic procedures require 1 d; SV experiments are performed just after size exclusion. The whole time for these experiments is 24 h. Data analysis of analytical ultracentrifugation requires a couple of days.     

Direct analysis of protein sedimentation equilibrium in detergent solutions without density matching

Characterizing membrane proteins by sedimentation equilibrium is challenging because detergents and/or lipid molecules, usually required for solubilization, form a complex with the protein. The most common way to overcome this problem is Tanford and Reynolds' density matching method, which eliminates the buoyant mass contributions of detergents/lipids by adjusting the solvent density with D2O/H2O mixtures to render either detergent or lipid molecules neutrally buoyant. Unfortunately, the method is practical only for detergent densities between 1.0 (H2O) and 1.1 (D2O) g ml(-1), excluding many of the more commonly used detergents for membrane protein studies. Here, we present a modern variant of Tanford and Reynolds' method that (1) is applicable to any detergent regardless of its specific density, (2) does not compromise accuracy and precision, and (3) provides additional information about the number of detergent molecules that are bound to each protein. The new method was applied successfully to Delta(1-43)A-I, an amino-terminal deletion mutant of human apolipoprotein A-I. Interestingly, we observed a significantly lower Delta(1-43)A-I/octyl-glucoside complex partial specific volume than that expected from volume additivity rules, indicative of specific protein-detergent interactions.     

Integral membrane protein interaction with Triton cytoskeletons of erythrocytes

The organization of erythrocyte membrane lipids and proteins has been studied following the release of cytoplasmic components with the non-ionic detergent Triton X-100. After detergent extraction, a detergent-resistant complex called the erythrocyte cytoskeleton is separated from detergent, solubilized lipid and protein by sucrose buoyant density sedimentation. In cytoskeletons prepared under isotonic conditions all of the major erythrocyte membrane proteins are retained except for the integral protein, glycophorin, which is quantitatively solubilized and another integral glycoprotein, band 3, which is only 60% removed. When cytoskeletons are prepared in hypertonic KCl solutions, band 3 is fully solubilized along with bands 2.1 and 4.2 and several minor components. The resulting cytoskeletons have the same morphology as those prepared in isotonic buffer but they are composed of only three major peripheral proteins, spectrin, actin and band 4.1. We have designated this peripheral protein complex the 'shell' of the erythrocyte membrane, and have shown that the attachment of band 3 to the shell satisfies the criteria for a specific interaction. Although Triton did affect erythrocyte shape, cytoskeleton lipid content and the activity of membrane proteases, there was no indication that Triton altered the attachment of band 3 to the shell. We suggest that band 3 attaches to the shell as part of a ternary complex of bands 2.1, 3 and 4.2.     

Short-chain phospholipids as detergents

The physico-chemical properties of short-chain phosphatidylcholine are reviewed to the extent that its biological activity as a mild detergent can be rationalized. Long-chain diacylphosphatidylcholines are typical membrane phospholipids that form preferentially smectic lamellar phases (bilayers) when dispersed in water. In contrast, the preferred phase of the short-chain analogues dispersed in excess water is the micellar phase. The preferred conformation and the dynamics of short-chain phosphatidylcholines in the monomeric and micellar state present in H(2)O are discussed. The motionally averaged conformation of short-chain phosphatidylcholines is then compared to the single-crystal structures of membrane lipids. The main conclusion emerging is that in terms of preferred conformation and motional averaging short-chain phosphatidylcholines closely resemble their long-chain analogues. The dispersing power of short-chain phospholipids is emphasized in the second part of the review. Evidence is presented to show that this class of compounds is superior to most other detergents used in the solubilization of membrane proteins and the reconstitution of the solubilized proteins to artificial membrane systems (proteoliposomes). The prominent feature of the solubilization/reconstitution of integral membrane proteins by short-chain PC is the retention of the native protein structure and hence the protein function. Due to their special detergent-like properties, short-chain PC lend themselves very well not only to membrane solubilization but also to the purification of integral membrane proteins. The retention of the native protein structure in the solubilized state, i.e. in mixed micelles consisting of the integral membrane protein, intrinsic membrane lipids and short-chain PC, is rationalized. It is hypothesized that short-chain PC interacts primarily with the lipid bilayer of a membrane and very little if at all with the membrane proteins. In this way, the membrane protein remains associated with its preferred intrinsic membrane lipids and retains its native structure and its function.     

Characterization of the nicotinic acetylcholine receptor isolated from goldfish brain.

We have studied the binding of alpha-bungarotoxin to a particulate fraction of goldfish brain enriched in synaptosomes. The binding is specific and saturable and exhibits the pharmacological properties of a nicotinic cholinergic receptor. Equilibrium binding measurements yield a single dissociation constant (KD) of 0.92 nM. Kinetic analysis revealed one association rate constant and two dissociation rate constants. Dissociation constants calculated from kinetic measurements were 1.9 nM and 12.5 pM. The toxin . receptor complex is readily solubilized in nonionic detergent. The isoelectric point of the toxin . receptor complex was found to be 5.00 +/- 0.01. Sedimentation velocity analysis in sucrose/H2O and sucrose/D2O gradients in conjunction with Sepharose 4B chromatography and diffusion experiments yielded a sedimentation constant of 11.45, a partial specific volume of 0.79 cm3/g for the toxin . receptor . detergent complex, and a molecular weight of approximately 340,000 for the toxin . receptor complex.     

A detergent-induced charge shift as a prerequisite for the electrochemical analysis of the adenine nucleotide translocator, a basic membrane protein

The adenine nucleotide translocator is a hydrophobic, basic protein of the inner mitochondrial membrane which is solubilized by the non-ionic detergent Triton X-100. For immunochemical characterization of this membrane-protein by crossed immunoelectrophoresis a charge shift of the protein-Triton X-100 micelle by the introduction of an ionic detergent (deoxycholate) was necessary as a prerequisite to avoid unspecific precipitation of the protein. Beside the charge shift of the protein-detergent micelle, the selection, concentration and ratio of the detergents used and the choice of the agarose with different degrees of electroendosmosis should be carefully considered. The principle derived from these results provides a new methodological possibility for the immunochemical characterization of hydrophobic, basic membrane proteins.     

Solubilization and hydrodynamic properties of pig atrial muscarinic acetylcholine receptor in dodecyl beta-D-maltoside.

The pig atrial muscarinic acetylcholine receptor (mAcChR) has been solubilized from the membrane-bound state in high yield and in stable conformation by the non-ionic detergent dodecyl beta-D-maltoside (DBM). The yield and selectivity for receptor solubilization is dependent on the detergent/protein ratio during extraction. Extraction at 2 mg of DBM/mg of protein gave a 75% yield of solubilized receptor with a 1.5-fold enrichment. A double-extraction procedure, in which non-receptor protein was first extracted at 0.4 mg of DBM/mg of protein and mAcChR was selectively solubilized by a second extraction at 0.35 mg of DBM/mg of protein, gave a 50% overall yield and a 2.8-fold enrichment. Both preparations had a half-life of about 20 days on ice without addition of muscarinic ligands. Receptor stability was decreased by the presence of cations, particularly bivalent cations, and enhanced by the agonist carbachol. Dissociation constants for the interaction of the DBM-solubilized receptor with the antagonist L-quinuclidinyl benzilate (Kd = 223 pM) and the agonist carbachol (Kd = 100 microM) were similar to those for the digitonin/cholate-solubilized receptor. Pig atrial mAcChR purified in digitonin/cholate and exchanged into DBM displayed reliable hydrodynamic behaviour during sucrose density sedimentation in gradients of 2H2O and H2O and during gel filtration in Sephacryl S-300. DBM is thus the first detergent which will solubilize a stable form of the ligand-free mAcChR in yields similar to those with digitonin, and is the only stabilizing detergent thus far suitable for hydrodynamic studies. DBM is also likely to be similarly useful in studying other membrane proteins for which digitonin has been the solubilizing detergent of choice.     

Reconstitution of liver endoplasmic reticulum membranes from solubilized proteins and lipids by removal of detergent]

A method for membrane reconstitution from cholate-solubilized microsomal proteins and lipids by a removal of the detergent on a column with charcoal has been developed. A comparative study showed that the membranes reconstituted by a dialysis or absorption do not differ from each other in terms of membrane proteins incorporation into lipid vesicles and cytochrome P-450 reconversion into cytochrome P-450. A possibility of biomembrane reconstitution from membrane proteins and lipids solubilized by a non-ionic detergent Triton X-100 was shown. A removal of the detergent results in a formation of membranes, which are chemically close to the original ones but ultrastructurally very different from the latter. On the other hand, absorption or dialysis of cholate-solubilized proteins and lipids results in reconstituted membranes with asymmetrically arranged intramembrane particles located on the hydrophobic surfaces of the membrane halves. The number and size of these particles are similar to those of the original microsomal membranes.     

A cost-effective method for simultaneous homo-oligomeric size determination and monodispersity conditions for membrane proteins

The use of blue native polyacrylamide gel electrophoresis (BN-PAGE) has been reported in the literature to retain both water-soluble and membrane protein complexes in their native hetero-oligomeric state and to determine the molecular weight of membrane proteins. However, membrane proteins show abnormal mobility when compared with water-soluble markers. Although one could use membrane proteins as markers or apply a conversion factor to the observed molecular weight to account for the bound Coomassie blue dye, when one just wants to assess homo-oligomeric size, these methods appear to be too time-consuming or might not be generally applicable. Here, during detergent screening studies to identify the best detergent for achieving a monodisperse sample, we observed that under certain conditions membrane proteins tend to form ladders of increasing oligomeric size. Although the ladders themselves contain no indication of which band represents the correct oligomeric size, they provide a scale that can be compared with a single band, representing the native homo-oligomeric size, obtained in other conditions of the screen. We show that this approach works for three membrane proteins: CorA (42 kDa), aquaporin Z (25 kDa), and small hydrophobic (SH) protein from respiratory syncytial virus (8 kDa). In addition, polydispersity results and identification of the most suitable detergent correlate optimally not only with size exclusion chromatography (SEC) but also with results from sedimentation velocity and equilibrium experiments. Because it involves minute quantities of sample and detergent, this method can be used in high-throughput approaches as a low-cost technique.     

Strategies for crystallizing membrane proteins

Crystallizing membrane proteins remains a challenging endeavor despite the increasing number of membrane protein structures solved by X-ray crystallography. The critical factors in determining the success of the crystallization experiments are the purification and preparation of membrane protein samples. Moreover, there is the added complication that the crystallization conditions must be optimized for use in the presence of detergents although the methods used to crystallize most membrane proteins are, in essence, straightforward applications of standard methodologies for soluble protein crystallization. The roles that detergents play in the stability and aggregation of membrane proteins as well as the colloidal properties of the protein-detergent complexes need to be appreciated and controlledbefore and during the crystallization trials. All X-ray quality crystals of membrane proteins were grown from preparations of detergent-solubilized protein, where the heterogeneous natural lipids from the membrane have been replaced by ahomogeneous detergent environment. It is the preparation of such monodisperse, isotropic solutions of membrane proteins that has allowed the successful application of the standard crystallization methods routinely used on soluble proteins. In this review, the issues of protein purification and sample preparation are addressed as well as the new refinements in crystallization methodologies for membrane proteins. How the physical behavior of the detergent, in the form of micelles or protein-detergent aggregates, affects crystallization and the adaptation of published protocols to new membrane protein systems are also addressed. The general conclusion is that many integral membrane proteins could be crystallized if pure and monodisperse preparations in a suitable detergent system can be prepared.In memory of Glenn D. Garavito.     

A high-throughput differential filtration assay to screen and select detergents for membrane proteins

Structural studies on integral membrane proteins are routinely performed on protein-detergent complexes (PDCs) consisting of purified protein solubilized in a particular detergent. Of all the membrane protein crystal structures solved to date, a subset of only four detergents has been used in more than half of these structures. Unfortunately, many membrane proteins are not well behaved in these four detergents and/or fail to yield well-diffracting crystals. Identification of detergents that maintain the solubility and stability of a membrane protein is a critical step and can be a lengthy and “protein-expensive” process. We have developed an assay that characterizes the stability and size of membrane proteins exchanged into a panel of 94 commercially available and chemically diverse detergents. This differential filtration assay (DFA), using a set of filtered microplates, requires sub-milligram quantities of purified protein and small quantities of detergents and other reagents and is performed in its entirety in several hours.     

Interaction of membrane proteins and lipids with solubilizing detergents

Detergents are indispensable in the isolation of integral membrane proteins from biological membranes to study their intrinsic structural and functional properties. Solubilization involves a number of intermediary states that can be studied by a variety of physicochemical and kinetic methods; it usually starts by destabilization of the lipid component of the membranes, a process that is accompanied by a transition of detergent binding by the membrane from a noncooperative to a cooperative interaction already below the critical micellar concentration (CMC). This leads to the formation of membrane fragments of proteins and lipids with detergent-shielded edges. In the final stage of solubilization membrane proteins are present as protomers, with the membrane inserted sectors covered by detergent. We consider in detail the nature of this interaction and conclude that in general binding as a monolayer ring, rather than as a micelle, is the most probable mechanism. This mode of interaction is supported by neutron diffraction investigations on the disposition of detergent in 3-D crystals of membrane proteins. Finally, we briefly discuss the use of techniques such as analytical ultracentrifugation, size exclusion chromatography, and mass spectrometry relevant for the structural investigation of detergent solubilized membrane proteins.     

Organic solvent extracted EmrE solubilized in dodecyl maltoside is monomeric and binds drug ligand

Ethidium multidrug resistance protein (EmrE) is a member of the small multidrug resistance family of proteins and is responsible for resistance to a diverse group of lipophilic cations. To examine the multimeric state(s), size-exclusion HPLC and sedimentation velocity experiments were performed with EmrE solubilized in N-dodecyl-beta-d-maltopyranoside (DM) detergent. EmrE was purified from Escherichia coli membranes using organic extraction with a 3:1 chloroform:methanol solvent followed by LH-20 chromatography and the recovered pure protein was re-solubilized in a buffer containing 2% DM. The purified protein was analyzed by SEC-HPLC to estimate the monodispersity and to determine the amount of bound detergent. The results show that EmrE is homogeneous in DM with a Stokes radius of 3.6nm compatible with that of a monomer. Sedimentation velocity experiments indicated that the EmrE preparation was monodisperse and supports the fact that the organic extracted protein solubilized in DM is monomeric. This monomeric form of the protein analyzed here is also shown to bind substrate in the micromolar range.     

Photoreaction center of photosynthetic bacteria. 2. Size and quaternary structure of the photoreaction centers from Rhodospirillum rubrum strain G9 and from Rhodopseudomonas sphaeroides strain 2.4.1.

The photoreaction center from Rhodospirillum rubrum strain G9 binds about 6 times as much sodium dodecyl sulfate as certain proteins commonly used as molecular weight markers for sodium dodecyl sulfate--polyacrylamide gel electrophoresis. This presumably explains the apparent discrepancy between the molecular weight of the photoreaction center determined by electrophoresis (76 000) and its minimal molecular weight (87 000). The molecular weight of the photoreaction center solubilized with Triton X-100 was determined by three different methods: conventional sedimentation equilibrium, a combination of sedimentation velocity and gel filtration measurements, and sedimentation equilibrium in H2O and in D2O. Each technique required a determination of the amount of bound detergent. All three methods gave molecular weight values close to 60 000. A similar molecular weight was found for the photoactive beta gamma dimer obtained from the photoreaction center of Rhodopseudomonas sphaeroides strain 2.4.1 which, as a whole, had a molecular weight of 87 000. These results indicate that the photoreaction center from Rp. sphaeroides is an oligomer of the type alpha 1 beta 1 gamma 1. In contrast, the photoreaction center from Rs. rubrum appears to be dissociated, in solution, into a photoactive beta gamma dimer and a free polypeptide alpha.     

Determination of the molecular mass and dimensions of membrane proteins by size exclusion chromatography

Size exclusion chromatography is an established technique for the determination of hydrodynamic volumes of proteins or protein complexes. When applied to membrane proteins, the contribution of the detergent micelle, which is required to keep the protein soluble in the aqueous phase, needs to be determined to obtain accurate measurements for the protein. In a detergent series, in which the detergents differ only by the length of the alkyl chain, the contribution of the detergent micelle to the hydrodynamic volume is variable, whereas the contribution of the protein is constant. By using this approach, several parameters of membrane proteins can be estimated by extrapolation, such as the radius at the midpoint of the membrane, the average radius, the Stokes radius, and the excluded volume. The molecular mass of the protein can be determined by two independent measurements that arise from the behaviour of the free detergent micelle and protein-detergent micelle during size exclusion chromatography and the determination of the detergent-protein ratio. Determining the dimensions of protein-detergent micelles may facilitate membrane protein purification and crystallization by defining the accessibility of the protein surface.     

Mitochondrial bovine ADP/ATP carrier in detergent is predominantly monomeric but also forms multimeric species

ADP/ATP carriers (AACs) are major and essential constituents of the inner mitochondrial membrane. They drive the import of ADP and the export of newly synthesized ATP. They were described as functional dimers from the 1980s until the structures of the AAC shed doubt on this consensus. We aimed to ascertain the published biophysical data claiming that AACs are dimers and to characterize the oligomeric state of the protein before crystallization. Analytical ultracentrifugation sedimentation velocity experiments clearly show that the bovine AAC is a monomer in 3-laurylamido-N,N'-dimethylpropylaminoxide (LAPAO), whereas in Triton X-100 and reduced Triton X-100, higher molecular mass species can also be identified. Neutron scattering data for monomeric bovine AAC in LAPAO does not give definite conclusions on the association state, because the large amount of detergent and lipids is imperfectly matched by contrast methods. We discuss a possible way to integrate previously published biochemical evidence in favor of assemblies, the lack of well-defined multimers that we observe, and the information from the high-resolution structures, considering supramolecular organizations of AACs within the mitochondrial membrane.