Solution NMR approaches for establishing specificity of weak heterodimerization of membrane proteins

Optimizing NMR fragment-based drug screening for membrane protein targets

NMR spectroscopy has played a pivotal role in fragment-based drug discovery by coupling detection of weak ligand-target binding with structural mapping of the binding site. Fragment-based screening by NMR has been successfully applied to many soluble protein targets, but only to a limited number of membrane proteins, despite the fact that many drug targets are membrane proteins. This is partly because of difficulties preparing membrane proteins for NMR-especially human membrane proteins-and because of the inherent complexity associated with solution NMR spectroscopy on membrane protein samples, which require the inclusion of membrane-mimetic agents such as micelles, nanodiscs, or bicelles. Here, we developed a generalizable protocol for fragment-based screening of membrane proteins using NMR. We employed two human membrane protein targets, both in fully protonated detergent micelles: the single-pass C-terminal domain of the amyloid precursor protein, C99, and the tetraspan peripheral myelin protein 22 (PMP22). For both we determined the optimal NMR acquisition parameters, protein concentration, protein-to-micelle ratio, and upper limit to the concentration of D6-DMSO in screening samples. Furthermore, we conducted preliminary screens of a plate-format molecular fragment mixture library using our optimized conditions and were able to identify hit compounds that selectively bound to the respective target proteins. It is hoped that the approaches presented here will be useful in complementing existing methods for discovering lead compounds that target membrane proteins.

Verteporfin is a Substrate-Selective γ-Secretase Inhibitor that Binds the Amyloid Precursor Protein Transmembrane Domain

This work reports substrate-selective inhibition of a protease with broad substrate specificity based on direct binding of a small molecule inhibitor to the substrate. The target for these studies was γ-secretase protease, which cleaves dozens of different single span membrane protein substrates, including both the C99 domain of the human amyloid precursor protein and the Notch receptor. Substrate-specific inhibition of C99 cleavage is desirable to reduce production of the amyloid-β polypeptide without inhibiting Notch cleavage, a major source of toxicity associated with broad specificity γ-secretase inhibitors. In order to identify a C99-selective inhibitors of the human γ-secretase, we conducted an NMR-based screen of FDA-approved drugs against C99 in model membranes. From this screen, we identified the small molecule verteporfin with these properties. We observed that verteporfin formed a direct 1:1 complex with C99, with a KD of 15-47 μM (depending on the membrane mimetic used), and that it did not bind the transmembrane domain of the Notch-1 receptor. Biochemical assays showed that direct binding of verteporfin to C99 inhibits γ-secretase cleavage of C99 with IC50 values in the range of 15- 164 μM, while Notch-1 cleavage was inhibited only at higher concentrations, and likely via a mechanism that does not involve binding to Notch-1. This work documents a robust NMR-based approach to discovery of small molecule binders to single-span membrane proteins and confirmed that it is possible to inhibit γ-secretase in a substrate-specific manner.

Bicelles Rich in both Sphingolipids and Cholesterol and Their Use in Studies of Membrane Proteins

How the distinctive lipid composition of mammalian plasma membranes impacts membrane protein structure is largely unexplored, partly because of the dearth of isotropic model membrane systems that contain abundant sphingolipids and cholesterol. This gap is addressed by showing that sphingomyelin and cholesterol-rich (SCOR) lipid mixtures with phosphatidylcholine can be cosolubilized by n-dodecyl-β-melibioside to form bicelles. Small-angle X-ray and neutron scattering, as well as cryo-electron microscopy, demonstrate that these assemblies are stable over a wide range of conditions and exhibit the bilayered-disc morphology of ideal bicelles even at low lipid-to-detergent mole ratios. SCOR bicelles are shown to be compatible with a wide array of experimental techniques, as applied to the transmembrane human amyloid precursor C99 protein in this medium. These studies reveal an equilibrium between low-order oligomer structures that differ significantly from previous experimental structures of C99, providing an example of how ordered membranes alter membrane protein structure.

Critical Effect of the Detergent:Protein Ratio on the Formation of the Hepatitis C Virus p7 Channel

The p7 protein encoded by the hepatitis C virus forms a cation-selective viroporin in the membrane. One of the most intriguing findings about the p7 viroporin is its unique hexameric structure in dodecylphosphocholine (DPC) micelles determined by nuclear magnetic resonance (NMR), but the hexameric structure was recently challenged by another NMR study of p7, also in DPC detergent, which claimed that the p7 in this detergent is monomeric. Here, we show that p7 oligomerization is highly sensitive to the detergent:protein ratio used in protein reconstitution and that the 40-fold difference in this ratio between the two studies was the cause of their different conclusions. In addition, we have performed extensive measurements of interchain paramagnetic relaxation enhancements (PREs) for p7 hexamers reconstituted in DPC micelles and in 1,2-dimyristoyl-sn-glycero-3-phosphocholine/1,2-dihexanoyl-sn-glycero-3-phosphocholine bicelles. In both cases, interchain PREs are overall consistent with the hexameric structure determined in micelles. Our data validate the overall architecture of the p7 hexamer while highlighting the importance of the detergent:protein ratio in membrane protein sample preparation.

Understanding single-pass transmembrane receptor signaling from a structural viewpoint-what are we missing?

Single-pass transmembrane receptors are involved in essential processes of both physiological and pathological nature and represent more than 1300 proteins in the human genome. Despite the high biological relevance of these receptors, the mechanisms of the signal transductions they facilitate are incompletely understood. One major obstacle is the lack of structures of the transmembrane domains that connect the extracellular ligand-binding domains to the intracellular signaling platforms. Over a period of almost 20 years since the first structure was reported, only 21 of these receptors have become represented by a transmembrane domain structure. This scarceness stands in strong contrast to the significance of these transmembrane α-helices for receptor functionality. In this review, we explore the properties and qualities of the current set of structures, as well as the methodological difficulties associated with their characterization and the challenges left to be overcome. Without an increased and focused effort to bring this class of proteins on par with the remaining membrane protein field, a serious lag in their biological understanding looms. Design of pharmaceutical agents, prediction of mutational affects in relation to disease, and deciphering of functional mechanisms require high-resolution structural information, especially when dealing with a domain carrying so much functionality in so few residues.

Structural insight into the transmembrane domain and the juxtamembrane region of the erythropoietin receptor in micelles

Erythropoietin receptor (EpoR) dimerization is an important step in erythrocyte formation. Its transmembrane domain (TMD) and juxtamembrane (JM) region are essential for signal transduction across the membrane. A construct compassing residues S212-P259 and containing the TMD and JM region of the human EpoR was purified and reconstituted in detergent micelles. The solution structure of the construct was determined in dodecylphosphocholine (DPC) micelles by solution NMR spectroscopy. Structural and dynamic studies demonstrated that the TMD and JM region are an ?-helix in DPC micelles, whereas residues S212-D224 at the N-terminus of the construct are not structured. The JM region is a helix that contains a hydrophobic patch formed by conserved hydrophobic residues (L253, I257, and W258). Nuclear Overhauser effect analysis, fluorescence spectroscopy, and paramagnetic relaxation enhancement experiments suggested that the JM region is exposed to the solvent. The structures of the TMD and JM region of the mouse EpoR were similar to those of the human EpoR.

Statistical analysis of crystallization database links protein physico-chemical features with crystallization mechanisms

X-ray crystallography is the predominant method for obtaining atomic-scale information about biological macromolecules. Despite the success of the technique, obtaining well diffracting crystals still critically limits going from protein to structure. In practice, the crystallization process proceeds through knowledge-informed empiricism. Better physico-chemical understanding remains elusive because of the large number of variables involved, hence little guidance is available to systematically identify solution conditions that promote crystallization. To help determine relationships between macromolecular properties and their crystallization propensity, we have trained statistical models on samples for 182 proteins supplied by the Northeast Structural Genomics consortium. Gaussian processes, which capture trends beyond the reach of linear statistical models, distinguish between two main physico-chemical mechanisms driving crystallization. One is characterized by low levels of side chain entropy and has been extensively reported in the literature. The other identifies specific electrostatic interactions not previously described in the crystallization context. Because evidence for two distinct mechanisms can be gleaned both from crystal contacts and from solution conditions leading to successful crystallization, the model offers future avenues for optimizing crystallization screens based on partial structural information. The availability of crystallization data coupled with structural outcomes analyzed through state-of-the-art statistical models may thus guide macromolecular crystallization toward a more rational basis.

EmrE dimerization depends on membrane environment

The small multi-drug resistant (SMR) transporter EmrE functions as a homodimer. Although the small size of EmrE would seem to make it an ideal model system, it can also make it challenging to work with. As a result, a great deal of controversy has surrounded even such basic questions as the oligomeric state. Here we show that the purified protein is a homodimer in isotropic bicelles with a monomer-dimer equilibrium constant (KMD(2D)) of 0.002-0.009mol% for both the substrate-free and substrate-bound states. Thus, the dimer is stabilized in bicelles relative to detergent micelles where the KMD(2D) is only 0.8-0.95mol% (Butler et al. 2004). In dilauroylphosphatidylcholine (DLPC) liposomes KMD(2D) is 0.0005-0.0008mol% based on Förster resonance energy transfer (FRET) measurements, slightly tighter than bicelles. These results emphasize the importance of the lipid membrane in influencing dimer affinity.

Characterizing protein crystal contacts and their role in crystallization: rubredoxin as a case study

The fields of structural biology and soft matter have independently sought out fundamental principles to rationalize protein crystallization. Yet the conceptual differences and the limited overlap between the two disciplines have thus far prevented a comprehensive understanding of the phenomenon to emerge. We conduct a computational study of proteins from the rubredoxin family that bridges the two fields. Using atomistic simulations, we characterize the protein crystal contacts, and accordingly parameterize patchy particle models. Comparing the phase diagrams of these schematic models with experimental results enables us to critically examine the assumptions behind the two approaches. The study also reveals features of protein–protein interactions that can be leveraged to crystallize proteins more generally.

Architecture of the hepatitis C virus E1 glycoprotein transmembrane domain studied by NMR

Oligomerization of hepatitis C viral envelope proteins E1 and E2 is essential to virus fusion and assembly. Although interactions within the transmembrane (TM) domains of these glycoproteins have proven contributions to the E1/E2 heterodimerization process and consequent infectivity, there is little structural information on this entry mechanism. Here, as a first step towards our long-term goal of understanding the interaction between E1 and E2 TM-domains, we have expressed, purified and characterized E1-TM using structural biomolecular NMR methods. An MBP-fusion expression system yielded sufficient quantities of pure E1-TM, which was solubilized in two membrane-mimicking environments, SDS- and LPPG-micelles, affording samples amenable to NMR studies. Triple resonance assignment experiments and relaxation measurements provided information on the secondary structure and global fold of E1-TM in these environments. In SDS micelles E1-TM adopts a helical conformation, with helical stretches at residues 354-363 and 371-379 separated by a more flexible segment of residues 364-370. In LPPG micelles a helical conformation was observed for residues 354-377 with greater flexibility in the 366-367 dyad, suggesting LPPG provides a more native environment for the peptide. Replacement of key positively charged residue K370 with an alanine did not affect the secondary structure of E1-TM but did change the relative positioning within the micelle of the two helices. These results lay the foundation for structure determination of E1-TM and a molecular understanding of how E1-TM flexibility enhances its interaction with E2-TM during heterodimerization and membrane fusion.

Design and synthesis of curcumin analogues for in vivo fluorescence imaging and inhibiting copper-induced cross-linking of amyloid beta species in Alzheimer's disease

In this article, we first designed and synthesized curcumin-based near-infrared (NIR) fluorescence imaging probes for detecting both soluble and insoluble amyloid beta (Aβ) species and then an inhibitor that could attenuate cross-linking of Aβ induced by copper. According to our previous results and the possible structural stereohindrance compatibility of the Aβ peptide and the hydrophobic/hydrophilic property of the Aβ13-20 (HHQKLVFF) fragment, NIR imaging probe CRANAD-58 was designed and synthesized. As expected CRANAD-58 showed significant fluorescence property changes upon mixing with both soluble and insoluble Aβ species in vitro. In vivo NIR imaging revealed that CRANAD-58 was capable of differentiating transgenic and wild-type mice as young as 4 months old, the age that lacks apparently visible Aβ plaques and Aβ is likely in its soluble forms. According to our limited studies on the interaction mechanism between CRANAD-58 and Aβ, we also designed CRANAD-17 to attenuate the cross-linking of Aβ42 induced by copper. It is well-known that the coordination of copper with imidazoles on Histidine-13 and 14 (H13, H14) of Aβ peptides could initialize covalent cross-linking of Aβ. In CRANAD-17, a curcumin scaffold was used as an anchoring moiety to usher the designed compound to the vicinity of H13 and H14 of Aβ, and imidazole rings were incorporated to compete with H13/H14 for copper binding. The results of SDS-PAGE gel and Western blot indicated that CRANAD-17 was capable of inhibiting Aβ42 cross-linking induced by copper. This raises a potential for CRANAD-17 to be considered for AD therapy.

NMR-based conformational ensembles explain pH-gated opening and closing of OmpG channel

The outer membrane protein G (OmpG) is a monomeric 33 kDa 14-stranded β-barrel membrane protein functioning as a nonspecific porin for the uptake of oligosaccharides in Escherichia coli. Two different crystal structures of OmpG obtained at different values of pH suggest a pH-gated pore opening mechanism. In these structures, extracellular loop 6 extends away from the barrel wall at neutral pH but is folded back into the pore lumen at low pH, blocking transport through the pore. Loop 6 was invisible in a previously published solution NMR structure of OmpG in n-dodecylphosphocholine micelles, presumably due to conformational exchange on an intermediate NMR time scale. Here we present an NMR paramagnetic relaxation enhancement (PRE)-based approach to visualize the conformational dynamics of loop 6 and to calculate conformational ensembles that explain the pH-gated opening and closing of the OmpG channel. The different loop conformers detected by the PRE ensemble calculations were validated by disulfide cross-linking of strategically engineered cysteines and electrophysiological single channel recordings. The results indicate a more dynamically regulated channel opening and closing than previously thought and reveal additional membrane-associated conformational ensembles at pH 6.3 and 7.0. We anticipate this approach to be generally applicable to detect and characterize functionally important conformational ensembles of membrane proteins.

Modeling helical proteins using residual dipolar couplings, sparse long-range distance constraints and a simple residue-based force field

We present a fast and simple protocol to obtain moderate-resolution backbone structures of helical proteins. This approach utilizes a combination of sparse backbone NMR data (residual dipolar couplings and paramagnetic relaxation enhancements) or EPR data with a residue-based force field and Monte Carlo/simulated annealing protocol to explore the folding energy landscape of helical proteins. By using only backbone NMR data, which are relatively easy to collect and analyze, and strategically placed spin relaxation probes, we show that it is possible to obtain protein structures with correct helical topology and backbone RMS deviations well below 4 Å. This approach offers promising alternatives for the structural determination of proteins in which nuclear Overha-user effect data are difficult or impossible to assign and produces initial models that will speed up the high-resolution structure determination by NMR spectroscopy.

Reversible folding of human peripheral myelin protein 22, a tetraspan membrane protein

Misfolding of the α-helical membrane protein peripheral myelin protein 22 (PMP22) has been implicated in the pathogenesis of the common neurodegenerative disease known as Charcot-Marie-Tooth disease (CMTD) and also several other related peripheral neuropathies. Emerging evidence suggests that the propensity of PMP22 to misfold in the cell may be due to an intrinsic lack of conformational stability. Therefore, quantitative studies of the conformational equilibrium of PMP22 are needed to gain insight into the molecular basis of CMTD. In this work, we have investigated the folding and unfolding of wild type (WT) human PMP22 in mixed micelles. Both kinetic and thermodynamic measurements demonstrate that the denaturation of PMP22 by n-lauroyl sarcosine (LS) in dodecylphosphocholine (DPC) micelles is reversible. Assessment of the conformational equilibrium indicates that a significant fraction of unfolded PMP22 persists even in the absence of the denaturing detergent. However, we find the stability of PMP22 is increased by glycerol, which facilitates quantitation of thermodynamic parameters. To our knowledge, this work represents the first report of reversible unfolding of a eukaryotic multispan membrane protein. The results indicate that WT PMP22 possesses minimal conformational stability in micelles, which parallels its poor folding efficiency in the endoplasmic reticulum. Folding equilibrium measurements for PMP22 in micelles may provide an approach to assess the effects of cellular metabolites or potential therapeutic agents on its stability. Furthermore, these results pave the way for future investigation of the effects of pathogenic mutations on the conformational equilibrium of PMP22.

The quiet renaissance of protein nuclear magnetic resonance

From roughly 1985 through the start of the new millennium, the cutting edge of solution protein nuclear magnetic resonance (NMR) spectroscopy was to a significant extent driven by the aspiration to determine structures. Here we survey recent advances in protein NMR that herald a renaissance in which a number of its most important applications reflect the broad problem-solving capability displayed by this method during its classical era during the 1970s and early 1980s.

The backbone dynamics of the amyloid precursor protein transmembrane helix provides a rationale for the sequential cleavage mechanism of γ-secretase

The etiology of Alzheimer's disease depends on the relative abundance of different amyloid-β (Aβ) peptide species. These peptides are produced by sequential proteolytic cleavage within the transmembrane helix of the 99 residue C-terminal fragment of the amyloid precursor protein (C99) by the intramembrane protease γ-secretase. Intramembrane proteolysis is thought to require local unfolding of the substrate helix, which has been proposed to be cleaved as a homodimer. Here, we investigated the backbone dynamics of the substrate helix. Amide exchange experiments of monomeric recombinant C99 and of synthetic transmembrane domain peptides reveal that the N-terminal Gly-rich homodimerization domain exchanges much faster than the C-terminal cleavage region. MD simulations corroborate the differential backbone dynamics, indicate a bending motion at a diglycine motif connecting dimerization and cleavage regions, and detect significantly different H-bond stabilities at the initial cleavage sites. Our results are consistent with the following hypotheses about cleavage of the substrate: First, the GlyGly hinge may precisely position the substrate within γ-secretase such that its catalytic center must start proteolysis at the known initial cleavage sites. Second, the ratio of cleavage products formed by subsequent sequential proteolysis could be influenced by differential extents of solvation and by the stabilities of H-bonds at alternate initial sites. Third, the flexibility of the Gly-rich domain may facilitate substrate movement within the enzyme during sequential proteolysis. Fourth, dimerization may affect substrate processing by decreasing the dynamics of the dimerization region and by increasing that of the C-terminal part of the cleavage region.