Journal of Magnetic Resonance 241 (2014) 86–96
Contents lists available at ScienceDirect
Journal of Magnetic Resonance journal homepage: www.elsevier.com/locate/jmr
Perspectives in Magnetic Resonance
Functional dynamics of cell surface membrane proteins Noritaka Nishida a, Masanori Osawa a, Koh Takeuchi b, Shunsuke Imai a, Pavlos Stampoulis a, Yutaka Kofuku a, Takumi Ueda a, Ichio Shimada a,⇑ a b
Graduate School of Pharmaceutical Sciences, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan Molecular Profiling Research Center, National Institute of Advanced Industrial Science and Technology, Aomi, Koto-ku, Tokyo 135-0064, Japan
a r t i c l e
i n f o
Article history: Received 2 September 2013 Revised 8 November 2013 Available online 22 November 2013 Keywords: NMR Membrane proteins Cell surface receptors Dynamics
a b s t r a c t Cell surface receptors are integral membrane proteins that receive external stimuli, and transmit signals across plasma membranes. In the conventional view of receptor activation, ligand binding to the extracellular side of the receptor induces conformational changes, which convert the structure of the receptor into an active conformation. However, recent NMR studies of cell surface membrane proteins have revealed that their structures are more dynamic than previously envisioned, and they fluctuate between multiple conformations in an equilibrium on various timescales. In addition, NMR analyses, along with biochemical and cell biological experiments indicated that such dynamical properties are critical for the proper functions of the receptors. In this review, we will describe several NMR studies that revealed direct linkage between the structural dynamics and the functions of the cell surface membrane proteins, such as G-protein coupled receptors (GPCRs), ion channels, membrane transporters, and cell adhesion molecules. Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction Cell surface receptors are integral membrane proteins that receive specific stimuli from the outside of cells, and transmit signals toward the inside of cells in well-controlled manners. The functions and structures of cell surface receptors are diverse. The major classes of the cell surface receptors are classified into 3 types: (1) seven membrane-spanning G-protein coupled receptors (GPCRs), (2) ion channel-linked receptors, and (3) catalytic receptors that are coupled with an intracellular enzymatic domain (e.g. kinase or phosphatase) [1]. Membrane transporters form another class of integral membrane proteins on the cell surface that actively or passively transport specific substances (e.g. ions, small molecules, and macromolecules) to maintain the homeostasis of the intracellular environments. Since many therapeutic reagents target cell surface receptors/membrane transporters, it is increasingly important to elucidate the structure–function relationships of these receptors at atomic resolution. Over the past decade, structural understanding of cell surface membrane proteins has dramatically progressed. Growing numbers of structures of integral membrane proteins, including GPCRs, have been solved by X-ray crystallography or NMR, thereby strongly promoting our understanding of how these integral mem⇑ Corresponding author. Address: Graduate School of Pharmaceutical Sciences, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Fax: +81 3 3815 6540. E-mail address: [email protected] (I. Shimada). 1090-7807/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jmr.2013.11.007
brane receptors transmit the signals into the cells [2]. However, the crystal structures represent static snapshot of the target protein in the crystal lattice, and the observed conformation may not be the major state in the lipid bilayer environment. Recently, accumulating evidences have suggested that cell surface membrane proteins are structurally more dynamic, and interconvert between multiple conformations in equilibrium, such as between active and inactive conformations [3,4]. In this respect, NMR methods are very useful for obtaining information about the dynamics of the target protein on a wide range of timescales (e.g. from ns to even days) under physiological conditions [5]. Recent advances in the NMR techniques describing conformational dynamics, such as relaxation dispersion, have revealed that the global conformational fluctuations of proteins are tightly related to their functions, such as enzymatic activities [6,7] or molecular recognition [8]. Solid-state NMR can also provide atomic-level structural and dynamical information for heterogeneous biological systems, even in cellular membranes or whole cells. The structures and conformational dynamics of membrane proteins studied by the solid state NMR techniques have been extensively reviewed elsewhere [9]. Here, we define ‘‘functional dynamics’’ as conformational dynamics that are directly related to their function. In order to discriminate the functionally relevant dynamics from random fluctuations of protein conformations, the timescale of the protein dynamics should be correlated with that of their biological functions. In this review, we will cover the four demonstrations of functional dynamics of functionally independent membrane proteins
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
by solution NMR. First, we will discuss how the conformational equilibrium of a GPCR determines the signal transduction across membranes. Secondly, we will describe detailed conformational dynamics of the K+ channel, and directly demonstrate the conformational equilibrium underlying the channel activity in detergent micelles and lipid bilayer mimicking conditions. Thirdly, we will discuss the conformational exchange of the multi-drug resistant transporter, which is crucial for transporting its ligand across the membrane. Finally, we will describe the cell adhesion molecule that mediates the transient adhesion of leukocytes under the influence of hydrodynamic forces, and how the conformational dynamics at the molecular level is related to the phenomena at the cellular level.
2. Conformational equilibrium determines transmembrane signal transduction by GPCR GPCRs are one of the largest membrane protein families in eukaryotes, and more than 25% of modern drugs target GPCRs. These drugs bind to GPCRs, leading to the induction or inhibition of signal transduction mediated by G-proteins, b-arrestins, and various other effectors via GPCRs. Each chemical ligand for a GPCR has a different level of ability to activate or inhibit its target, which is commonly referred to as efficacy, and the ligands are classified according to their efficacies, such as full agonists, partial agonists, neutral antagonists, and inverse agonists. These differences in the efficacies significantly affect the therapeutic properties of the GPCR ligands. In the case of drugs that target b2-adrenergic receptor (b2AR), a full agonist offers a clinical advantage over a partial agonist in acute severe asthma, although full agonists are capable of causing more adverse effects. However, the mechanism underlying the differences in the efficacy was not clear, although several structures of GPCRs complexed with ligands have been determined by X-ray crystallography [10,11]. Recently, several NMR studies have revealed that the conformational dynamics of b2AR provide an explanation for its signal transduction mechanism. Liu et al. incorporated CF3 probes, which can be observed with high sensitivity and selectivity, into b2AR. They demonstrated that the 19F-NMR signals from the CF3 probe at C265 exhibited two components, which are referred to as components I and A, and the population of component A correlated with the efficacy of the ligand [12]. They also characterized the structure of components I and A by paramagnetic relaxation enhancement (PRE) experiments, which suggested that C327 is more exposed to the solvent in I than in A, in an agonist-bound state. Nygaard et al. demonstrated that b2AR exhibits dynamical feature in the presence of an agonist, which are reduced upon binding with a G-protein-mimetic nanobody [13]. In addition, as described below, Kofuku et al. utilized NMR to clarify the conformational diversity of the transmembrane (TM) region of b2AR in the inverse agonist-bound, neutral antagonist-bound, weak partial agonist-bound, partial agonist-bound, and full agonist-bound states [14]. b2AR possesses nine methionine residues in extracellular loop 1 (ECL1), TM1, TM2, TM4, TM5, and TM6, and M82, M215, and M279 assume distinctly different conformations between the inverse agonist-bound and full agonist/G-protein-bound crystal structures (Fig. 1A) [10,11]. Therefore, these methionine methyl groups can be utilized to investigate the conformations of the TM region of b2AR in various ligand-bound states by NMR. In the spectra of the inverse agonist- and full agonist-bound forms, signals that apparently corresponded to the nine methionine residues were detected, suggesting that most of the methionine residues were observed, including those in the TM region (Fig. 1B and C). The spectra were significantly simplified by introducing mutations into the four solvent- or lipid-exposed
87
methionine residues of b2AR (Fig. 1D and E). In the spectrum of the M82V mutant, two signals were absent in the inverse agonist-bound state, revealing that both of these resonances are from M82 (Fig. 1F). The downfield and upfield resonances from M82 are referred to as M82D and M82U, respectively. In the full agonist-bound state, one signal was absent in the spectrum of the M82V mutant, revealing that this resonance is from M82 (Fig. 1G). The chemical shifts of the resonance from M82 in the full agonist-bound state were different from M82D and M82U in the inverse agonist-bound state. This resonance from M82 in the full agonist-bound state is referred to as M82A. The 13C and 1H chemical shifts of the methionine methyl signals are reportedly affected by its side-chain conformation and the local environments, including the ring current effects from the neighboring aromatic rings, respectively [15,16]. The side-chain conformation and the local environment that correspond to the chemical shifts of M82U and M82D are in good agreement with the crystal structures of b2AR in complex with inverse agonists, and those of M82A are in good agreement with the crystal structure with both a full agonist and a G-protein (Fig. 1H). Therefore, we proposed that the M82U and M82D signals correspond to the inactive states that cannot directly activate G-proteins, and the M82A signal corresponds to the active state that can interact with G-proteins. These distinct conformations corresponding to the M82U, M82D and M82A signals are referred to as the M82U, M82D, and M82A conformations, respectively. To examine the correlation between the ligand efficacy and the b2AR conformations, the spectra were acquired in the presence of compounds with different efficacies for b2AR (Fig. 2A and B). In the antagonist-bound state, major and minor resonances that slightly shifted from M82U and M82D, respectively, were observed. In both the weak partial agonist- and partial agonist-bound states, a signal was observed at a chemical shift between M82U and M82A, and the chemical shifts in the weak partial agonist-bound state were closer to those of M82U. To examine whether the resonances from M82 in the ligand-bound states undergo conformational exchange, the spectra were recorded at a lower temperature, 283 K (Fig. 2C). As a result, the resonances from M82 in the weak partial agonist- and partial agonist-bound states significantly shifted away from M82U, and the M82A resonance in the full agonistbound state slightly shifted away from M82U at 283 K. These temperature-dependent shifts, together with the chemical shift changes in an efficacy-dependent manner (Fig. 2B) and the good agreement between the observed spectra and simulated data of M82 resonances (Fig. 2D and E), suggested that b2AR exists in an equilibrium between the M82D, M82U, and M82A conformations. Based on the above structural interpretation of the resonances from M82, the following signal regulation mechanism was proposed (Fig. 2F). In the full agonist-bound state, most of the b2AR molecules assume the active conformation. In the partial agonists-bound states, b2AR exists in an equilibrium between the inactive and active conformations, and the populations of the two conformations determine the efficacies. In the neutral antagonist alprenolol-bound state, b2AR exists in an equilibrium between two major inactive conformations, which differ only in the region close to the ligand binding site, and one minor active conformation. The weak basal activity is due to the existence of the minor M82A conformation. In the inverse agonist-bound state, b2AR exists in an equilibrium between the two locally different inactive conformations. The resonances from M215 and M279, which are located on the cytoplasmic side of the TM region, were not observed in the full agonist-bound state (Fig. 1C and E). Nygaard et al. demonstrated that the intensity of these resonances is increased upon binding with a G-protein mimicking nanobody [13]. These results suggested that these signals were broadened due to the
88
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
A
B
D
F
C
E
G
H
Fig. 1. NMR spectra of b2AR in the inverse agonist- and full agonist-bound states. (A) Distribution of the methionine residues in the overlaid crystal structures of b2AR. The crystal structure of b2AR with an inverse agonist, carazolol (PDB code: 2RH1), is shown in grey ribbons. Methionine sidechains and carazolol are depicted by cyan and grey sticks, respectively. The crystal structure of b2AR with a full agonist, BI-167107, and a G-protein (PDB code: 3SN6) is shown in violet ribbons. Methionine sidechains and BI167107 are depicted by orange and violet sticks, respectively. These structures are overlaid at TM2, and are shown in a side view with the extracellular sides at the top. ICL3s, which were either substituted with T4 lysozyme or not observed, are shown with dotted lines. (B and C) 1H–13C SOFAST-HMQC spectra of [methyl-13C-Met] b2AR in the form bound with an inverse agonist, carazolol (B), and in the form bound with a full agonist, formoterol (C). (D and E) 1H–13C SOFAST-HMQC spectra of the [methyl-13C-Met] b2AR M96T/M98T/M156L/M171S mutant in the inverse agonist-bound state (D) and in the full agonist-bound state (E). (F and G) 1H–13C SOFAST-HMQC spectra of [methyl-13CMet] b2AR M82V/M96T/M98T/M156L/M171S mutant in the inverse agonist-bound state (F) and in the full agonist-bound state (G). The regions with methionine chemical shifts are shown, and the assigned resonances are indicated. The resonances from M40 in the inverse agonist-bound state, and M40, M215, and M279 in the full agonistbound state were not observed. Resonances indicated with asterisks are derived from minor impurity proteins from insect cell membranes, but not from b2AR. Double asterisks are t1 noises derived from the intense DDM signal with a 1H chemical shift of 1.6–1.7 p.p.m. (H) Summary of the 1H and 13C chemical shifts of the M82 resonances and the structure around M82 in the b2AR crystal structures. The characteristics of the crystal structure with an inverse agonist, carazolol (PDB code: 2RH1), and that with another inverse agonist, timolol (PDB code: 3D4S), are in good agreement with the side-chain conformations and the local environments that correspond to the chemical shifts of M82U and M82D respectively. The characteristics of the crystal structure with both a full-agonist and a G-protein (PDB code: 3SN6) are in good agreement with the side-chain conformation and the local environment that correspond to the chemical shift of M82A.
conformational exchange between the multiple conformations in the full agonist-bound state. The multiple conformations on the cytoplasmic side of the TM region in the full agonist-bound state may be suitable for interacting with various effectors in the conformational selection manners. A number of b2AR ligands impart different degrees of signaling in the G-protein and arrestin pathways, in a phenomenon called ‘‘functional selectivity’’ or ‘‘biased signaling’’. Biased signaling is important for understanding the functions of GPCRs and for drug development, and it is quite interesting to investigate the conformations of b2AR in the biased ligand-bound states. Liu et al. observed the resonances from 19F labels introduced at C265 and C327, which are at the cytoplasmic ends of TM6 and TM7 of b2AR, respectively, and demonstrated that b-arrestin-biased
ligands caused large shifts in the equilibration of C327 toward the active state, whereas the binding of full agonists caused shifts in the equilibration of both C265 and C327 [12]. These resulted suggest that the biased signaling is due to the diversity of the conformational dynamics on the cytoplasmic side of the TM region in each ligand-bound state.
3. Functional dynamics of K+ channel KcsA is a pH-dependent K+ channel from Streptomyces lividans that functions as a tetramer. The crystal structures of KcsA revealed the K+-permeation pathway on the four-fold symmetry axis of the tetramer, where the selectivity filter exists on the extracellular side
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
A
89
B
C D
F
E
Fig. 2. The different b2AR M82 resonances in the states with various efficacies. (A) 1H–13C SOFAST-HMQC spectra of [a,b,b-2H3-, methyl-13C-Met] b2AR/4Met at 298 K in complex with an inverse agonist, carazolol (black), a neutral antagonist, alprenolol (cyan), a weak partial agonist, tulobuterol (green), a partial agonist, clenbuterol (violet), and a full agonist, formoterol (red). Only the regions with M82 resonances are shown. (B) Overlay of the spectra shown in (A), with the same colors. The centers of the resonances from M82 are indicated with dots. (C) Overlay of the 1H–13C HMQC spectra of [a,b,b-2H3-, methyl-13C-Met] b2AR/4Met recorded at 298 K (black) and 283 K (magenta), in the inverse agonist-, neutral antagonist-, weak partial agonist-, partial agonist-, and full agonist-bound states. Only the regions with M82 methyl resonances are shown. The M82 resonances are indicated with the superscripts used in the main text, and the centers are indicated with dots. Resonances with asterisks are derived from impurities. (D and E) Simulation of the ligand-dependent shift of the M82 resonances. Overlay of the spectra of M82 resonances in complexes with an inverse agonist, carazolol (black), a neutral antagonist, alprenolol (cyan), a weak partial agonist, tulobuterol (green), a partial agonist, clenbuterol (violet), and a full agonist, formoterol (red) at 298 K. (D) and (E) represent the observed and simulated spectra, respectively. F. Proposed mechanism for the various efficacies of b2AR with different ligands. b2AR adopts three conformations with different M82 environments; the M82A conformation induces signaling, whereas the M82U and M82D conformations do not. The M82U and M82A conformations differ largely on TM5 and TM6, whereas the differences between the M82D and M82U conformations are localized in the region close to the ligand binding site. In the full agonist formoterol-bound state, b2AR primarily adopts mostly the M82A conformation, exhibiting almost the full efficacy of b2AR. In the partial agonist clenbuteroland tulobuterol-bound states, b2AR exists in equilibrium between the M82A and M82U conformations, exhibiting significant signaling with reduced efficacies. In the tulobuterol-bound state, where the efficacy is lower than that of the clenbuterol-bound state, the population of the M82U conformation is larger. In the neutral antagonist alprenolol-bound state, b2AR primarily adopts the M82U and M82D conformations, in equilibrium with a small population of the M82A conformation. The presence of the small population with the M82A conformation accounts for the basal activity of b2AR. In the inverse agonist carazolol-bound state, b2AR exists in equilibrium between the M82U and M82D conformations, exhibiting the inhibition of basal activities.
and a gate of the permeation pathway, the helix bundle crossing, is on the intracellular side (Fig. 3A) [17,18]. The former is responsible for the selective permeation of K+ against other ions, and the latter is a gate that opens and closes in response to the intracellular pH. Therefore, the conformational changes and the dynamics of the K+ channels are directly correlated to their functions, K+-permeation through lipid bilayers. The helix bundle crossing is open at acidic pH (pH 3–4), and closed at neutral pH. Thus, a constant K+ current is expected while the gate is open. In macroscopic current analyses, however, a pH drop from 7 to 3 caused a transient strong current, which decayed exponentially with a time constant of seconds (Fig. 3B) [19–22]. This process, referred to as activation-coupled inactivation, is also
observed for many voltage-dependent K+ (Kv) channels upon depolarization of the membrane as an activation stimulus [23], which is one of the important functions of Kv channels for cellular K+ homeostasis. To reveal the structural mechanism of the physiologically relevant function, the dynamics of KcsA in DDM micelles was investigated. The pH-dependent NMR spectral changes were investigated, by monitoring the Trp indole HN signals and the methyl signals of Ile, Leu, and Val of the wild type KcsA and its mutants, which indicated that protonation of His25 triggers the gate opening of KcsA [24,25]. The conformational changes occurred not only in the residues near the helix bundle crossing, but also those around the selectivity filter on the extracellular side [25].
90
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
A
D
B
C
E
Fig. 3. Functional dynamics of a K+ channel, KcsA. (A) Crystal structure of KcsA in the closed conformation at neutral pH (PDB ID: 3EFF), shown in a view parallel to the membrane. One of the subunits in the tetramer is colored green. V76 in the selectivity filter is colored red. (B) Macroscopic current of KcsA [22]. (C) Selected region of the methyl-TROSY spectra of KcsA in the closed (pH 6.7, 45 °C) (black), permeable (pH 3.2, 45 °C) (red), and impermeable (pH 3.2, 25 °C) (green) conformations, in the presence of 120 mM K+. (D) 1H–1H strips representing Val76c, derived from 13C-edited NOESY-HMQC observed in 90% H2O, in the closed (black), permeable (red), and impermeable (green) states, using a mixing time of 50 ms. (E) Proposed mechanism of the activation-coupled inactivation of KcsA.
At pH 3.2, a number of minor signals were found close to the major methyl TROSY signals arising from the residues around the selectivity filter. A 13C ZZ-exchange spectrum [26] indicated that a pair of major and minor signals arises from one methyl group undergoing slow conformational exchange on the NMR chemical shift timescale. A comparison with the methyl-TROSY spectra of two mutants of KcsA, in which one is constitutively K+-permeable and the other is predominantly K+-impermeable at acidic pH, where the helix bundle crossing is open, in the single channel K+ current analyses [22], allowed the major and minor signals to be assigned as those from the permeable and impermeable conformations, respectively. Thus, three conformations, corresponding to the three physiological states, the closed state at neutral pH, and the permeable and impermeable states at acidic pH, which were identified in the electrophysiological assays [20,22], were clearly observed as the three different methyl-TROSY signals (Fig. 3C) [25]. K+-titration experiments at pH 6.7 and 3.2 revealed that at a physiological K+ concentration, the selectivity filter is K+-bound in the closed state at neutral pH, while the K+-bound, permeable conformation and the K+-unbound, impermeable conformation exist in equilibrium at acidic pH. NOE analyses clearly demonstrated that water molecules are trapped in the selectivity filter only in the K+-unbound, impermeable conformation (Fig. 3D). Furthermore, the mixing time dependence of the 13C ZZ-exchange signal intensities revealed that the transition rate from the permeable to the impermeable conformation, kpi, was 0.5 s 1, and that in the opposite direction, kip, was 0.9 s 1 (Fig. 4B) [27]. These NMR results provided a mechanism for the activationcoupled inactivation of KcsA (Fig. 3E). Since the selectivity filter
binds to the K+ ions at neutral pH, it remains K+-bound upon the pH drop that opens the helix bundle crossing. Thus, the equilibrium of the selectivity filter between the permeable and impermeable conformations at acidic pH initiates from the permeable one, and then reaches the steady state of the equilibrium. This is the mechanism of the observed transient strong K+-current and the subsequent exponential decay. Indeed, the NMR-derived transition rates of kpi and kip well reproduced the decay profile of the macroscopic current at acidic pH. This mechanism is assumed to be applicable to the C-type inactivation of the eukaryotic Kv channels, because of the high similarity in both the amino acid sequences and structures between KcsA and the eukaryotic Kv channels. Since the electrophysiological profile of the activation-coupled inactivation is determined by the equilibrium between the permeable and impermeable conformations of the selectivity filter, it is important to identify the factor(s) that governs its conformational equilibrium. The deletion of the C-terminal intracellular region of KcsA reportedly increased both the rate and extent of inactivation [21], indicating that the C-terminal intracellular region indirectly affects the conformational equilibrium of the selectivity filter. Since the C-terminal region stabilizes the tetrameric assembly by forming a bundle in the closed conformation at neutral pH (Fig. 3A) [28], the structural equilibrium of the selectivity filter may be affected by the surrounding environment that modulates the tetrameric stability. NMR and SDS-PAGE analyses of a variety of C-terminally truncated KcsA mutants indicated a positive correlation between the population of the permeable conformation (pperm) and the tetrameric stability of the KcsA mutants [27]. The transition rates
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
91
Fig. 4. Quantification of the equilibrium between the permeable and impermeable states of the KcsA selective filter by NMR. (A) Methyl-TROSY (left) and 13C ZZ exchange (right) spectra of KcsA at pH 3.2 and 40.0 °C, in the presence of 50 mM K+. The 13C ZZ exchange spectrum was acquired with a mixing time of 250 ms. The regions involved in forming for the permeable and impermeable conformations of V76c are shown. The intensities of the peaks are designated as Ipp, Iii, Iip, and Ipi, for the auto peak of the permeable conformation, the auto peak of the impermeable conformation, the cross peak of the impermeable to the permeable conformation, and the cross peak of the permeable to the impermeable conformation, respectively. (B) Plots of the peak intensities of V76c versus the mixing time. The intensities of the peaks for V76c of KcsA (left) and KcsA125 (right) at pH 3.2 and 40.0 °C, in the presence of 50 mM K+, were plotted against the mixing time. Solid lines indicate the best fit of the data to the theoretical equation of the two-site exchange model [21,27].
between the permeable and impermeable conformations of the selectivity filter were investigated for full length KcsA and the 35-residue C-terminally truncated mutant, KcsA125, by 13C ZZ exchange experiments (Fig. 4B). The kpi value obtained for KcsA125 was more than 10 times higher than that for KcsA, whereas the kip values were comparable for the two mutants, indicating that the C-terminal intracellular region indirectly stabilizes the permeable conformation of the selectivity filter under acidic conditions. In lipid bilayers, the tetrameric stability of KcsA is reportedly perturbed by the surrounding membrane environments, such as lipid composition [29], and the partitioning of alcohol, such as 2,2,2-trifluoroethanol (TFE) [30,31], at neutral pH, where the helix bundle crossing is closed. Therefore, even at an acidic pH, where the helix bundle crossing is open, the tetrameric assembly, and thus the structural equilibrium of the selectivity filter, might be influenced by the surrounding environments. The solution NMR spectra of KcsA reconstituted in the lipid bilayer can be observed by using reconstituted high density lipoprotein (rHDL), a lipid bilayer surrounded by membrane scaffold proteins that forms a disc-shaped particle [32,33]. Compared with the spectra recorded in the DDM micelles, small but significant differences in the chemical shifts were observed not only for the residues exposed to the lipid bilayer in rHDL, but also for the residue of the selectivity filter, V76, and those peripheral to the selectivity filter, L59 and L81 (Fig. 5A and B) [27]. Furthermore, the signals from the K+-permeable and K+-impermeable conformations were simultaneously observed for V76 and L59. These results indicated that the overall folds and/or the tetrameric assemblies in rHDL and DDM micelles are slightly different but almost identical. The most striking differences are found in the pperm values: In rHDL, the pperm value is about 20%, whereas it is almost 100% in DDM micelles under the same conditions (Fig. 5A) [25]. These results clearly indicated that the surrounding environments affect the conformations of KcsA and their equilibrium. Furthermore, sequential additions of TFE to KcsA in rHDL affected the structural equilibrium of the selectivity filter (Fig. 5C). When TFE was added to final concentrations of 1% and 2%, the intensities of the K+-impermeable conformation signals that are predominant in the absence of TFE decreased, whereas those of the K+-permeable conformation increased, clearly indicating the shift of the structural equilibrium from the K+-impermeable to K+-permeable conformation. Surprisingly, upon the further addition of TFE to final concentrations of 3% and 4%, the signals of the K+-permeable conformation of the selectivity filter decreased, and the signals that corresponded to the closed conformation emerged, although the pH of the sample was maintained at pH
3.2. The other signals did not exhibit large chemical shift changes upon the addition of TFE, suggesting that TFE does not directly interact with KcsA, and that the global fold of KcsA is not affected by TFE at concentrations below 4%. This equilibrium shift upon TFE addition is reflected in the open probability of KcsA in the planar lipid bilayer, which is observed in the single channel recording of the K+ current (Fig. 5D). While the open probability (popen) at acidic pH was very low (1%) in the absence of TFE, popen dramatically increased as the TFE concentration increased, and reached 60% at 3% (v/v) TFE. Under the acidic pH conditions, the activation gate of the helix bundle crossing fully opens, and the low popen value in the absence of TFE represents the infrequent opening of the selectivity filter. Thus, these results revealed that the conformational equilibrium of the selectivity filter is shifted towards the conductive form by TFE. These results clearly indicated that the shift in the structural and functional equilibria of KcsA is most likely caused by the modulation of the surrounding membrane environments.
4. Membrane transporter EmrE, a small multi-drug resistance transporter in bacteria, transports a broad class of polyaromatic cation substrates, such as ethidium bromide and methyl viologen [34]. EmrE is classified as an antiporter, since the transport of the substrates against their concentration gradient is coupled to the import of two protons in the opposite direction across the inner membrane. Each monomer of EmrE is composed of 110 amino acids, which forms four transmembrane helices. Although the crystal structures and the electron crystallographic studies suggested that EmrE forms an asymmetric dimer [35,36], the topology of EmrE in lipid bilayers was under debate [37]. EmrE is considered to function through a single-site alternating access model (Fig. 6A). In this model, the transporters convert between the inward- and outward-facing conformations to transport substrates across membranes. To perform the coupled antiport, both substrates (i.e. polyaromatic cations and protons) should share a single binding site and conformational exchange only occurs when the substrate is bound. Thus, saturating EmrE with its polyaromatic cation substrate, tetraphenylphosphonium (TPP+), should drive EmrE into a two-state equilibrium (Fig. 6B). A TROSY spectrum of TPP+-bound 2H/15N-EmrE in isotropic bicelles exhibited twice as many peaks as for the monomer EmrE, with similar intensities (Fig. 6C) [38]. From this peak doubling, it was postulated that EmrE forms an asymmetric dimer as observed
92
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
Fig. 5. NMR and electrophysiological analyses of KcsA reconstituted in the lipid bilayer. (A) Methyl-TROSY spectra of KcsA in rHDL and DDM micelles. The letters P and I denote the permeable and impermeable conformations, respectively. Signals from the detergent or lipid molecules are shown by the green boxes. (B) Mapping of the residues perturbed by the different reconstitution media. The N-terminal region, which is missing in the structure, is shown as a cartoon (green). (C) TFE titration experiments in the methyl-TROSY spectra of KcsA in rHDL. The methyl-TROSY spectra were observed in the absence and presence of 1%, 2%, 3%, and 4% TFE. Regions for V76c and L59d are shown. The letters P, I, and C denote the signals from the permeable, impermeable, and closed conformations, respectively. (D) TFE effects on the single channel recordings of KcsA. (Left) Representative single channel current traces of KcsA obtained by the planar lipid bilayer system, in the absence and presence of 3% (v/v) TFE on both sides of the planar lipid bilayer. Currents were recorded at +200 mV with a symmetric K+ concentration of 200 mM. (Center) the expanded traces for the relevant parts indicated with the black bars under the current traces in the left panels. The blue dashed lines indicate the closed level, and the red ones indicate the open levels. (Right) open probability (popen) as a function of TFE concentration.
by cryoelectron crystallography and X-ray crystallography. In the asymmetric dimer, each monomer has a distinct structure with different chemical shifts, and interconversion between in the inwardfacing and outward-facing conformations could effectively transport the ligands (Fig. 6A). To detect the interconversion between the two states, TROSYselected ZZ-exchange experiments were performed. As a result, many exchange peaks were observed, suggesting that there is a widespread conformational exchange in the EmrE dimer (Fig. 6C). Rates of the conformational exchange were calculated from the peak intensity of the ZZ exchange spectra with different mixing
times. Exchange rates between two states with equal populations were kopen-in to open-out = kopen-out to open-in = 4.8 ± 0.5 s-1 for TPP+bound EmrE in isotropic bicelles (Fig. 6D). Thus, NMR data demonstrates that the entire protein is involved in the conformational exchange on the same timescale. To determine whether EmrE forms an asymmetric dimer as proposed by X-ray crystallography, the water accessibility of each residue was measured by recording 1H–15N TROSY spectra with increasing concentrations of a soluble, chelated gadolinium compound, which causes line broadenings of the nearby amide groups, due to the paramagnetic effect. As a result, some residues exhibited
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
93
Fig. 6. Conformational exchange of EmrE transporter. (A) Proposed mechanism of EmrE transporter function. Interconversion between the inward-open and outward-open conformations enables EmrE to transport its substrate from the cytoplasm to the periplasm, while transporting H+ in the opposite direction. (B) Conformational exchange proposed for antiparallel, asymmetric EmrE. The two states adopt identical structure in opposite orientations with respect to the membrane. (C) Overlay of TROSY ZZexchange (red, 100 ms mixing time) and 1H–15N TROSY heteronuclear single-quantum coherence (HSQC) spectra (black). Exchange peaks and auto-peaks are connected by blue lines, and residue numbers are indicated. (D) The best-fit of ZZ-exchange auto- (solid circles) and cross-peak (open circles) intensities as a function of mixing time yields a single global conformational exchange rate of k = 4.9 ± 0.5 s 1. (E) 1H–15N TROSY spectra of TPP+-bound EmrE in bicelles in the absence (black) and presence of 1 mM (blue) and 5 mM (red) paramagnetic gadobenate dimeglumine. Residues with different PRE effects (green circles) or the same PRE effects (dark blue) in the two states are indicated. (F) Residues with equal (dark blue) and differential (yellow) accessibility to the water in the two states are mapped on the EmrE structure. Residues protected from water are colored red. This figure was adopted from [38] with permission.
line broadening to different extents (Fig. 6E). Their amide residues are located exactly within the region expected for the interconversion of an antiparallel dimer, indicating that the conformation observed by NMR is open toward one side of the membrane (Fig. 6F). Single –molecule FRET analyses, which are complementary to NMR methods that observe the equilibrium as ensemble, also confirmed the asymmetric dimer of EmrE. Therefore, the NMR study directly observed the exchange between the inward-facing and outwardfacing conformations of EmrE. This two-state equilibrium is a key step in the transport cycle of the multi-drug transporter to export the substrate across the membrane.
5. Cell adhesion molecules Cell-cell contacts are mediated by cell adhesion molecules. In the immunological response, leukocytes adhere and transmigrate
across the endothelial barrier, and then migrate to the inflammatory site (Fig. 7A) [39]. The first step of this event is rolling adhesion, where cells undergo repetitive attachment and detachment on endothelial cells, and then gradually decrease their speed to achieve firm adhesion. Such rolling adhesion is mediated by specialized adhesion molecules, such as selectins, CD44, and integrins. It remains unclear how such transient adhesion is achieved between the cell adhesion molecules and their ligands under fluid shear stress conditions. CD44 is a major hyaluronan (HA) receptor that mediates cell adhesion in leukocyte rolling and tumor metastasis [40]. The HAbinding domain (HABD) of CD44 contains a Link module (residues 32–124) and N- and C-terminal extensions (residues 21–31 and 125–178, respectively) (Fig. 7B). In the absence of the ligand, CD44 HABD adopts a well folded structure, in which the N- and C-terminal extensions form an additional structural element that intimately contacts the Link module [41,42] (Fig. 7C, referred to
94
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
Fig. 7. Two-state equilibrium of CD44 HABD in the absence and presence of ligand. (A) Schematic drawing of leukocyte trafficking to the inflammation site. The interaction between CD44 and hyaluronan mediates the rolling adhesion. (B) The domain organization of CD44. (C) Crystal structure of the CD44 HABD in the HA-unbound state (PDB ID: 1UUH). (D) Crystal structure of CD44 HABD in complex with an HA octamer (HA8) (PDB ID: 2JCR). (E) NMR structure of CD44 HABD in complex with an HA hexamer (PDB ID: 2I83). The Link module is colored grey, and the extension segments are blue. HA is shown as a yellow stick model. (F and G) Selected regions of the 1H–15N HSQC spectra of the CD44 HABD in the absence (F) and presence (G) of HA. The relative intensities of the peaks originating from the O-state and the PD-state are shown in each spectrum. (H) ZZexchange spectra of the ligand-free HABD, acquired without (black) or with a 500 ms mixing period (red). (I) Schematic representations of the equilibrium of the CD44 HABD between the O-conformation and the PD-conformation. The sizes of the arrows drawn between the O-state and the PD-state represent the changes in the conformational equilibrium in the presence and absence of HA.
as the ordered (O)-state). The crystal structure of HABD in complex with an 8mer HA revealed that HABD still adopts the O-state with only a minor conformational change, restricted to the HA binding site (Fig. 7D) [43]. In contrast, an NMR structure of HABD in complex with a 6mer HA demonstrated that HA-binding induces a conformational rearrangement of HABD, in which the C-terminal segment (residues following T153) becomes unstructured [44] (Fig. 7E, referred to as the partially disordered (PD)-state). There-
fore, the question was raised as to whether HABD adopts the Ostate or the PD-state in the HA-bound state. In an NMR spectrum of HABD in the absence of the ligand, minor peaks corresponding to the PD-state were observed, in addition to the major peaks corresponding to the O-state (Fig. 7F) [45]. Based on the peak intensities, the relative populations of the Ostate and the PD-state in the absence of HA are 86% and 14%, respectively. Conversely, in the presence of HA, minor peaks were
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
observed at the position of the O-state with a reversed population (i.e. O-state: 8%, PD-state: 92%) (Fig. 7G). In addition, ZZ-exchange experiments [26] demonstrated that HABD exists in an equilibrium between the O-state and the PD-state, with a timescale of hundreds of milliseconds (Fig. 7H). These results indicated that CD44 HABD exists in an equilibrium between the O- and PD-states, and the HA-binding induces a shift of the equilibrium toward the PD-conformation (Fig. 7I). Therefore, the HA-bound crystal structure can be interpreted as representing the O-state in the HAbound state, which is less than 10% of the population, according to the NMR analyses (Fig. 7I). The functional relevance of the two-state equilibrium of the CD44 was examined by utilizing a mutant (Y161A) that constitutively adopts the PD-state, by destabilizing the interaction between the C-terminal segment and the Link module, which is an important interaction for the O-state (Fig. 8A). SPR experiments demonstrated that the PD-state mutant exhibited seven-fold higher affinity for HA than the wild type protein, which predominantly adopts the O-state. Therefore, CD44 HABD exchanges between the low affinity (O-state) and high affinity (PD-state) conformations. To investigate the role of the two state-equilibrium in the context of cellular function, the cell rolling behaviors were analyzed under various shear stresses conditions, using VMRC-LCD cells expressing the wild type CD44 and the PD-state mutant [45]. Although the wild type CD44-expressing cells showed smooth rolling on the HA-coated surface, the PD-state mutantexpressing cells only exhibited firm adhesion (Fig. 8B). In order to smoothly roll along the ligand-coated surfaces for cells, new bond formation at the edge of the rolling front and breakage at the rear edge need to occur continuously, on an appropriate timescale (Fig. 8C) [46]. The impaired rolling behavior of the PD-state mutant suggested that the transition between the O-state and the PD-state in the ligand-bound form is necessary for the proper detachment of the receptor-ligand bond on the substrate-coated surface. This hypothesis is also consistent with the fact that the timescale of the conformational exchange of HABD is comparable
95
to the dwell time of the ligand-receptor bond formed on the cell surface (hundred millisecond order). Therefore, the intrinsic conformational equilibrium of the HABD between the low- and high affinity states is crucial for the CD44-mediated cell rolling.
6. Conclusions and future perspectives As described, a number of NMR studies have demonstrated that the conformational dynamics of cell surface membrane proteins are directly related to their functions. NMR methods are suitable for detecting protein dynamics on various timescales. Although the slower dynamics on NMR timescale were mainly characterized in this review, the dynamics at the different timescale should be explored. For example, the CPMG relaxation dispersion analyses of M2 proton channel of influenza A detected the millisecond timescale conformational exchange that is related to opening of the channel [47]. In addition, the membrane proteins may undergo conformational exchange between the ground state and the excited state. The paramagnetic relaxation enhancement experiments or relaxation dispersion experiments would be useful for obtaining structural parameters of sparsely populated species (less than
Contents lists available at ScienceDirect
Journal of Magnetic Resonance journal homepage: www.elsevier.com/locate/jmr
Perspectives in Magnetic Resonance
Functional dynamics of cell surface membrane proteins Noritaka Nishida a, Masanori Osawa a, Koh Takeuchi b, Shunsuke Imai a, Pavlos Stampoulis a, Yutaka Kofuku a, Takumi Ueda a, Ichio Shimada a,⇑ a b
Graduate School of Pharmaceutical Sciences, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan Molecular Profiling Research Center, National Institute of Advanced Industrial Science and Technology, Aomi, Koto-ku, Tokyo 135-0064, Japan
a r t i c l e
i n f o
Article history: Received 2 September 2013 Revised 8 November 2013 Available online 22 November 2013 Keywords: NMR Membrane proteins Cell surface receptors Dynamics
a b s t r a c t Cell surface receptors are integral membrane proteins that receive external stimuli, and transmit signals across plasma membranes. In the conventional view of receptor activation, ligand binding to the extracellular side of the receptor induces conformational changes, which convert the structure of the receptor into an active conformation. However, recent NMR studies of cell surface membrane proteins have revealed that their structures are more dynamic than previously envisioned, and they fluctuate between multiple conformations in an equilibrium on various timescales. In addition, NMR analyses, along with biochemical and cell biological experiments indicated that such dynamical properties are critical for the proper functions of the receptors. In this review, we will describe several NMR studies that revealed direct linkage between the structural dynamics and the functions of the cell surface membrane proteins, such as G-protein coupled receptors (GPCRs), ion channels, membrane transporters, and cell adhesion molecules. Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction Cell surface receptors are integral membrane proteins that receive specific stimuli from the outside of cells, and transmit signals toward the inside of cells in well-controlled manners. The functions and structures of cell surface receptors are diverse. The major classes of the cell surface receptors are classified into 3 types: (1) seven membrane-spanning G-protein coupled receptors (GPCRs), (2) ion channel-linked receptors, and (3) catalytic receptors that are coupled with an intracellular enzymatic domain (e.g. kinase or phosphatase) [1]. Membrane transporters form another class of integral membrane proteins on the cell surface that actively or passively transport specific substances (e.g. ions, small molecules, and macromolecules) to maintain the homeostasis of the intracellular environments. Since many therapeutic reagents target cell surface receptors/membrane transporters, it is increasingly important to elucidate the structure–function relationships of these receptors at atomic resolution. Over the past decade, structural understanding of cell surface membrane proteins has dramatically progressed. Growing numbers of structures of integral membrane proteins, including GPCRs, have been solved by X-ray crystallography or NMR, thereby strongly promoting our understanding of how these integral mem⇑ Corresponding author. Address: Graduate School of Pharmaceutical Sciences, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Fax: +81 3 3815 6540. E-mail address: [email protected] (I. Shimada). 1090-7807/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jmr.2013.11.007
brane receptors transmit the signals into the cells [2]. However, the crystal structures represent static snapshot of the target protein in the crystal lattice, and the observed conformation may not be the major state in the lipid bilayer environment. Recently, accumulating evidences have suggested that cell surface membrane proteins are structurally more dynamic, and interconvert between multiple conformations in equilibrium, such as between active and inactive conformations [3,4]. In this respect, NMR methods are very useful for obtaining information about the dynamics of the target protein on a wide range of timescales (e.g. from ns to even days) under physiological conditions [5]. Recent advances in the NMR techniques describing conformational dynamics, such as relaxation dispersion, have revealed that the global conformational fluctuations of proteins are tightly related to their functions, such as enzymatic activities [6,7] or molecular recognition [8]. Solid-state NMR can also provide atomic-level structural and dynamical information for heterogeneous biological systems, even in cellular membranes or whole cells. The structures and conformational dynamics of membrane proteins studied by the solid state NMR techniques have been extensively reviewed elsewhere [9]. Here, we define ‘‘functional dynamics’’ as conformational dynamics that are directly related to their function. In order to discriminate the functionally relevant dynamics from random fluctuations of protein conformations, the timescale of the protein dynamics should be correlated with that of their biological functions. In this review, we will cover the four demonstrations of functional dynamics of functionally independent membrane proteins
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
by solution NMR. First, we will discuss how the conformational equilibrium of a GPCR determines the signal transduction across membranes. Secondly, we will describe detailed conformational dynamics of the K+ channel, and directly demonstrate the conformational equilibrium underlying the channel activity in detergent micelles and lipid bilayer mimicking conditions. Thirdly, we will discuss the conformational exchange of the multi-drug resistant transporter, which is crucial for transporting its ligand across the membrane. Finally, we will describe the cell adhesion molecule that mediates the transient adhesion of leukocytes under the influence of hydrodynamic forces, and how the conformational dynamics at the molecular level is related to the phenomena at the cellular level.
2. Conformational equilibrium determines transmembrane signal transduction by GPCR GPCRs are one of the largest membrane protein families in eukaryotes, and more than 25% of modern drugs target GPCRs. These drugs bind to GPCRs, leading to the induction or inhibition of signal transduction mediated by G-proteins, b-arrestins, and various other effectors via GPCRs. Each chemical ligand for a GPCR has a different level of ability to activate or inhibit its target, which is commonly referred to as efficacy, and the ligands are classified according to their efficacies, such as full agonists, partial agonists, neutral antagonists, and inverse agonists. These differences in the efficacies significantly affect the therapeutic properties of the GPCR ligands. In the case of drugs that target b2-adrenergic receptor (b2AR), a full agonist offers a clinical advantage over a partial agonist in acute severe asthma, although full agonists are capable of causing more adverse effects. However, the mechanism underlying the differences in the efficacy was not clear, although several structures of GPCRs complexed with ligands have been determined by X-ray crystallography [10,11]. Recently, several NMR studies have revealed that the conformational dynamics of b2AR provide an explanation for its signal transduction mechanism. Liu et al. incorporated CF3 probes, which can be observed with high sensitivity and selectivity, into b2AR. They demonstrated that the 19F-NMR signals from the CF3 probe at C265 exhibited two components, which are referred to as components I and A, and the population of component A correlated with the efficacy of the ligand [12]. They also characterized the structure of components I and A by paramagnetic relaxation enhancement (PRE) experiments, which suggested that C327 is more exposed to the solvent in I than in A, in an agonist-bound state. Nygaard et al. demonstrated that b2AR exhibits dynamical feature in the presence of an agonist, which are reduced upon binding with a G-protein-mimetic nanobody [13]. In addition, as described below, Kofuku et al. utilized NMR to clarify the conformational diversity of the transmembrane (TM) region of b2AR in the inverse agonist-bound, neutral antagonist-bound, weak partial agonist-bound, partial agonist-bound, and full agonist-bound states [14]. b2AR possesses nine methionine residues in extracellular loop 1 (ECL1), TM1, TM2, TM4, TM5, and TM6, and M82, M215, and M279 assume distinctly different conformations between the inverse agonist-bound and full agonist/G-protein-bound crystal structures (Fig. 1A) [10,11]. Therefore, these methionine methyl groups can be utilized to investigate the conformations of the TM region of b2AR in various ligand-bound states by NMR. In the spectra of the inverse agonist- and full agonist-bound forms, signals that apparently corresponded to the nine methionine residues were detected, suggesting that most of the methionine residues were observed, including those in the TM region (Fig. 1B and C). The spectra were significantly simplified by introducing mutations into the four solvent- or lipid-exposed
87
methionine residues of b2AR (Fig. 1D and E). In the spectrum of the M82V mutant, two signals were absent in the inverse agonist-bound state, revealing that both of these resonances are from M82 (Fig. 1F). The downfield and upfield resonances from M82 are referred to as M82D and M82U, respectively. In the full agonist-bound state, one signal was absent in the spectrum of the M82V mutant, revealing that this resonance is from M82 (Fig. 1G). The chemical shifts of the resonance from M82 in the full agonist-bound state were different from M82D and M82U in the inverse agonist-bound state. This resonance from M82 in the full agonist-bound state is referred to as M82A. The 13C and 1H chemical shifts of the methionine methyl signals are reportedly affected by its side-chain conformation and the local environments, including the ring current effects from the neighboring aromatic rings, respectively [15,16]. The side-chain conformation and the local environment that correspond to the chemical shifts of M82U and M82D are in good agreement with the crystal structures of b2AR in complex with inverse agonists, and those of M82A are in good agreement with the crystal structure with both a full agonist and a G-protein (Fig. 1H). Therefore, we proposed that the M82U and M82D signals correspond to the inactive states that cannot directly activate G-proteins, and the M82A signal corresponds to the active state that can interact with G-proteins. These distinct conformations corresponding to the M82U, M82D and M82A signals are referred to as the M82U, M82D, and M82A conformations, respectively. To examine the correlation between the ligand efficacy and the b2AR conformations, the spectra were acquired in the presence of compounds with different efficacies for b2AR (Fig. 2A and B). In the antagonist-bound state, major and minor resonances that slightly shifted from M82U and M82D, respectively, were observed. In both the weak partial agonist- and partial agonist-bound states, a signal was observed at a chemical shift between M82U and M82A, and the chemical shifts in the weak partial agonist-bound state were closer to those of M82U. To examine whether the resonances from M82 in the ligand-bound states undergo conformational exchange, the spectra were recorded at a lower temperature, 283 K (Fig. 2C). As a result, the resonances from M82 in the weak partial agonist- and partial agonist-bound states significantly shifted away from M82U, and the M82A resonance in the full agonistbound state slightly shifted away from M82U at 283 K. These temperature-dependent shifts, together with the chemical shift changes in an efficacy-dependent manner (Fig. 2B) and the good agreement between the observed spectra and simulated data of M82 resonances (Fig. 2D and E), suggested that b2AR exists in an equilibrium between the M82D, M82U, and M82A conformations. Based on the above structural interpretation of the resonances from M82, the following signal regulation mechanism was proposed (Fig. 2F). In the full agonist-bound state, most of the b2AR molecules assume the active conformation. In the partial agonists-bound states, b2AR exists in an equilibrium between the inactive and active conformations, and the populations of the two conformations determine the efficacies. In the neutral antagonist alprenolol-bound state, b2AR exists in an equilibrium between two major inactive conformations, which differ only in the region close to the ligand binding site, and one minor active conformation. The weak basal activity is due to the existence of the minor M82A conformation. In the inverse agonist-bound state, b2AR exists in an equilibrium between the two locally different inactive conformations. The resonances from M215 and M279, which are located on the cytoplasmic side of the TM region, were not observed in the full agonist-bound state (Fig. 1C and E). Nygaard et al. demonstrated that the intensity of these resonances is increased upon binding with a G-protein mimicking nanobody [13]. These results suggested that these signals were broadened due to the
88
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
A
B
D
F
C
E
G
H
Fig. 1. NMR spectra of b2AR in the inverse agonist- and full agonist-bound states. (A) Distribution of the methionine residues in the overlaid crystal structures of b2AR. The crystal structure of b2AR with an inverse agonist, carazolol (PDB code: 2RH1), is shown in grey ribbons. Methionine sidechains and carazolol are depicted by cyan and grey sticks, respectively. The crystal structure of b2AR with a full agonist, BI-167107, and a G-protein (PDB code: 3SN6) is shown in violet ribbons. Methionine sidechains and BI167107 are depicted by orange and violet sticks, respectively. These structures are overlaid at TM2, and are shown in a side view with the extracellular sides at the top. ICL3s, which were either substituted with T4 lysozyme or not observed, are shown with dotted lines. (B and C) 1H–13C SOFAST-HMQC spectra of [methyl-13C-Met] b2AR in the form bound with an inverse agonist, carazolol (B), and in the form bound with a full agonist, formoterol (C). (D and E) 1H–13C SOFAST-HMQC spectra of the [methyl-13C-Met] b2AR M96T/M98T/M156L/M171S mutant in the inverse agonist-bound state (D) and in the full agonist-bound state (E). (F and G) 1H–13C SOFAST-HMQC spectra of [methyl-13CMet] b2AR M82V/M96T/M98T/M156L/M171S mutant in the inverse agonist-bound state (F) and in the full agonist-bound state (G). The regions with methionine chemical shifts are shown, and the assigned resonances are indicated. The resonances from M40 in the inverse agonist-bound state, and M40, M215, and M279 in the full agonistbound state were not observed. Resonances indicated with asterisks are derived from minor impurity proteins from insect cell membranes, but not from b2AR. Double asterisks are t1 noises derived from the intense DDM signal with a 1H chemical shift of 1.6–1.7 p.p.m. (H) Summary of the 1H and 13C chemical shifts of the M82 resonances and the structure around M82 in the b2AR crystal structures. The characteristics of the crystal structure with an inverse agonist, carazolol (PDB code: 2RH1), and that with another inverse agonist, timolol (PDB code: 3D4S), are in good agreement with the side-chain conformations and the local environments that correspond to the chemical shifts of M82U and M82D respectively. The characteristics of the crystal structure with both a full-agonist and a G-protein (PDB code: 3SN6) are in good agreement with the side-chain conformation and the local environment that correspond to the chemical shift of M82A.
conformational exchange between the multiple conformations in the full agonist-bound state. The multiple conformations on the cytoplasmic side of the TM region in the full agonist-bound state may be suitable for interacting with various effectors in the conformational selection manners. A number of b2AR ligands impart different degrees of signaling in the G-protein and arrestin pathways, in a phenomenon called ‘‘functional selectivity’’ or ‘‘biased signaling’’. Biased signaling is important for understanding the functions of GPCRs and for drug development, and it is quite interesting to investigate the conformations of b2AR in the biased ligand-bound states. Liu et al. observed the resonances from 19F labels introduced at C265 and C327, which are at the cytoplasmic ends of TM6 and TM7 of b2AR, respectively, and demonstrated that b-arrestin-biased
ligands caused large shifts in the equilibration of C327 toward the active state, whereas the binding of full agonists caused shifts in the equilibration of both C265 and C327 [12]. These resulted suggest that the biased signaling is due to the diversity of the conformational dynamics on the cytoplasmic side of the TM region in each ligand-bound state.
3. Functional dynamics of K+ channel KcsA is a pH-dependent K+ channel from Streptomyces lividans that functions as a tetramer. The crystal structures of KcsA revealed the K+-permeation pathway on the four-fold symmetry axis of the tetramer, where the selectivity filter exists on the extracellular side
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
A
89
B
C D
F
E
Fig. 2. The different b2AR M82 resonances in the states with various efficacies. (A) 1H–13C SOFAST-HMQC spectra of [a,b,b-2H3-, methyl-13C-Met] b2AR/4Met at 298 K in complex with an inverse agonist, carazolol (black), a neutral antagonist, alprenolol (cyan), a weak partial agonist, tulobuterol (green), a partial agonist, clenbuterol (violet), and a full agonist, formoterol (red). Only the regions with M82 resonances are shown. (B) Overlay of the spectra shown in (A), with the same colors. The centers of the resonances from M82 are indicated with dots. (C) Overlay of the 1H–13C HMQC spectra of [a,b,b-2H3-, methyl-13C-Met] b2AR/4Met recorded at 298 K (black) and 283 K (magenta), in the inverse agonist-, neutral antagonist-, weak partial agonist-, partial agonist-, and full agonist-bound states. Only the regions with M82 methyl resonances are shown. The M82 resonances are indicated with the superscripts used in the main text, and the centers are indicated with dots. Resonances with asterisks are derived from impurities. (D and E) Simulation of the ligand-dependent shift of the M82 resonances. Overlay of the spectra of M82 resonances in complexes with an inverse agonist, carazolol (black), a neutral antagonist, alprenolol (cyan), a weak partial agonist, tulobuterol (green), a partial agonist, clenbuterol (violet), and a full agonist, formoterol (red) at 298 K. (D) and (E) represent the observed and simulated spectra, respectively. F. Proposed mechanism for the various efficacies of b2AR with different ligands. b2AR adopts three conformations with different M82 environments; the M82A conformation induces signaling, whereas the M82U and M82D conformations do not. The M82U and M82A conformations differ largely on TM5 and TM6, whereas the differences between the M82D and M82U conformations are localized in the region close to the ligand binding site. In the full agonist formoterol-bound state, b2AR primarily adopts mostly the M82A conformation, exhibiting almost the full efficacy of b2AR. In the partial agonist clenbuteroland tulobuterol-bound states, b2AR exists in equilibrium between the M82A and M82U conformations, exhibiting significant signaling with reduced efficacies. In the tulobuterol-bound state, where the efficacy is lower than that of the clenbuterol-bound state, the population of the M82U conformation is larger. In the neutral antagonist alprenolol-bound state, b2AR primarily adopts the M82U and M82D conformations, in equilibrium with a small population of the M82A conformation. The presence of the small population with the M82A conformation accounts for the basal activity of b2AR. In the inverse agonist carazolol-bound state, b2AR exists in equilibrium between the M82U and M82D conformations, exhibiting the inhibition of basal activities.
and a gate of the permeation pathway, the helix bundle crossing, is on the intracellular side (Fig. 3A) [17,18]. The former is responsible for the selective permeation of K+ against other ions, and the latter is a gate that opens and closes in response to the intracellular pH. Therefore, the conformational changes and the dynamics of the K+ channels are directly correlated to their functions, K+-permeation through lipid bilayers. The helix bundle crossing is open at acidic pH (pH 3–4), and closed at neutral pH. Thus, a constant K+ current is expected while the gate is open. In macroscopic current analyses, however, a pH drop from 7 to 3 caused a transient strong current, which decayed exponentially with a time constant of seconds (Fig. 3B) [19–22]. This process, referred to as activation-coupled inactivation, is also
observed for many voltage-dependent K+ (Kv) channels upon depolarization of the membrane as an activation stimulus [23], which is one of the important functions of Kv channels for cellular K+ homeostasis. To reveal the structural mechanism of the physiologically relevant function, the dynamics of KcsA in DDM micelles was investigated. The pH-dependent NMR spectral changes were investigated, by monitoring the Trp indole HN signals and the methyl signals of Ile, Leu, and Val of the wild type KcsA and its mutants, which indicated that protonation of His25 triggers the gate opening of KcsA [24,25]. The conformational changes occurred not only in the residues near the helix bundle crossing, but also those around the selectivity filter on the extracellular side [25].
90
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
A
D
B
C
E
Fig. 3. Functional dynamics of a K+ channel, KcsA. (A) Crystal structure of KcsA in the closed conformation at neutral pH (PDB ID: 3EFF), shown in a view parallel to the membrane. One of the subunits in the tetramer is colored green. V76 in the selectivity filter is colored red. (B) Macroscopic current of KcsA [22]. (C) Selected region of the methyl-TROSY spectra of KcsA in the closed (pH 6.7, 45 °C) (black), permeable (pH 3.2, 45 °C) (red), and impermeable (pH 3.2, 25 °C) (green) conformations, in the presence of 120 mM K+. (D) 1H–1H strips representing Val76c, derived from 13C-edited NOESY-HMQC observed in 90% H2O, in the closed (black), permeable (red), and impermeable (green) states, using a mixing time of 50 ms. (E) Proposed mechanism of the activation-coupled inactivation of KcsA.
At pH 3.2, a number of minor signals were found close to the major methyl TROSY signals arising from the residues around the selectivity filter. A 13C ZZ-exchange spectrum [26] indicated that a pair of major and minor signals arises from one methyl group undergoing slow conformational exchange on the NMR chemical shift timescale. A comparison with the methyl-TROSY spectra of two mutants of KcsA, in which one is constitutively K+-permeable and the other is predominantly K+-impermeable at acidic pH, where the helix bundle crossing is open, in the single channel K+ current analyses [22], allowed the major and minor signals to be assigned as those from the permeable and impermeable conformations, respectively. Thus, three conformations, corresponding to the three physiological states, the closed state at neutral pH, and the permeable and impermeable states at acidic pH, which were identified in the electrophysiological assays [20,22], were clearly observed as the three different methyl-TROSY signals (Fig. 3C) [25]. K+-titration experiments at pH 6.7 and 3.2 revealed that at a physiological K+ concentration, the selectivity filter is K+-bound in the closed state at neutral pH, while the K+-bound, permeable conformation and the K+-unbound, impermeable conformation exist in equilibrium at acidic pH. NOE analyses clearly demonstrated that water molecules are trapped in the selectivity filter only in the K+-unbound, impermeable conformation (Fig. 3D). Furthermore, the mixing time dependence of the 13C ZZ-exchange signal intensities revealed that the transition rate from the permeable to the impermeable conformation, kpi, was 0.5 s 1, and that in the opposite direction, kip, was 0.9 s 1 (Fig. 4B) [27]. These NMR results provided a mechanism for the activationcoupled inactivation of KcsA (Fig. 3E). Since the selectivity filter
binds to the K+ ions at neutral pH, it remains K+-bound upon the pH drop that opens the helix bundle crossing. Thus, the equilibrium of the selectivity filter between the permeable and impermeable conformations at acidic pH initiates from the permeable one, and then reaches the steady state of the equilibrium. This is the mechanism of the observed transient strong K+-current and the subsequent exponential decay. Indeed, the NMR-derived transition rates of kpi and kip well reproduced the decay profile of the macroscopic current at acidic pH. This mechanism is assumed to be applicable to the C-type inactivation of the eukaryotic Kv channels, because of the high similarity in both the amino acid sequences and structures between KcsA and the eukaryotic Kv channels. Since the electrophysiological profile of the activation-coupled inactivation is determined by the equilibrium between the permeable and impermeable conformations of the selectivity filter, it is important to identify the factor(s) that governs its conformational equilibrium. The deletion of the C-terminal intracellular region of KcsA reportedly increased both the rate and extent of inactivation [21], indicating that the C-terminal intracellular region indirectly affects the conformational equilibrium of the selectivity filter. Since the C-terminal region stabilizes the tetrameric assembly by forming a bundle in the closed conformation at neutral pH (Fig. 3A) [28], the structural equilibrium of the selectivity filter may be affected by the surrounding environment that modulates the tetrameric stability. NMR and SDS-PAGE analyses of a variety of C-terminally truncated KcsA mutants indicated a positive correlation between the population of the permeable conformation (pperm) and the tetrameric stability of the KcsA mutants [27]. The transition rates
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
91
Fig. 4. Quantification of the equilibrium between the permeable and impermeable states of the KcsA selective filter by NMR. (A) Methyl-TROSY (left) and 13C ZZ exchange (right) spectra of KcsA at pH 3.2 and 40.0 °C, in the presence of 50 mM K+. The 13C ZZ exchange spectrum was acquired with a mixing time of 250 ms. The regions involved in forming for the permeable and impermeable conformations of V76c are shown. The intensities of the peaks are designated as Ipp, Iii, Iip, and Ipi, for the auto peak of the permeable conformation, the auto peak of the impermeable conformation, the cross peak of the impermeable to the permeable conformation, and the cross peak of the permeable to the impermeable conformation, respectively. (B) Plots of the peak intensities of V76c versus the mixing time. The intensities of the peaks for V76c of KcsA (left) and KcsA125 (right) at pH 3.2 and 40.0 °C, in the presence of 50 mM K+, were plotted against the mixing time. Solid lines indicate the best fit of the data to the theoretical equation of the two-site exchange model [21,27].
between the permeable and impermeable conformations of the selectivity filter were investigated for full length KcsA and the 35-residue C-terminally truncated mutant, KcsA125, by 13C ZZ exchange experiments (Fig. 4B). The kpi value obtained for KcsA125 was more than 10 times higher than that for KcsA, whereas the kip values were comparable for the two mutants, indicating that the C-terminal intracellular region indirectly stabilizes the permeable conformation of the selectivity filter under acidic conditions. In lipid bilayers, the tetrameric stability of KcsA is reportedly perturbed by the surrounding membrane environments, such as lipid composition [29], and the partitioning of alcohol, such as 2,2,2-trifluoroethanol (TFE) [30,31], at neutral pH, where the helix bundle crossing is closed. Therefore, even at an acidic pH, where the helix bundle crossing is open, the tetrameric assembly, and thus the structural equilibrium of the selectivity filter, might be influenced by the surrounding environments. The solution NMR spectra of KcsA reconstituted in the lipid bilayer can be observed by using reconstituted high density lipoprotein (rHDL), a lipid bilayer surrounded by membrane scaffold proteins that forms a disc-shaped particle [32,33]. Compared with the spectra recorded in the DDM micelles, small but significant differences in the chemical shifts were observed not only for the residues exposed to the lipid bilayer in rHDL, but also for the residue of the selectivity filter, V76, and those peripheral to the selectivity filter, L59 and L81 (Fig. 5A and B) [27]. Furthermore, the signals from the K+-permeable and K+-impermeable conformations were simultaneously observed for V76 and L59. These results indicated that the overall folds and/or the tetrameric assemblies in rHDL and DDM micelles are slightly different but almost identical. The most striking differences are found in the pperm values: In rHDL, the pperm value is about 20%, whereas it is almost 100% in DDM micelles under the same conditions (Fig. 5A) [25]. These results clearly indicated that the surrounding environments affect the conformations of KcsA and their equilibrium. Furthermore, sequential additions of TFE to KcsA in rHDL affected the structural equilibrium of the selectivity filter (Fig. 5C). When TFE was added to final concentrations of 1% and 2%, the intensities of the K+-impermeable conformation signals that are predominant in the absence of TFE decreased, whereas those of the K+-permeable conformation increased, clearly indicating the shift of the structural equilibrium from the K+-impermeable to K+-permeable conformation. Surprisingly, upon the further addition of TFE to final concentrations of 3% and 4%, the signals of the K+-permeable conformation of the selectivity filter decreased, and the signals that corresponded to the closed conformation emerged, although the pH of the sample was maintained at pH
3.2. The other signals did not exhibit large chemical shift changes upon the addition of TFE, suggesting that TFE does not directly interact with KcsA, and that the global fold of KcsA is not affected by TFE at concentrations below 4%. This equilibrium shift upon TFE addition is reflected in the open probability of KcsA in the planar lipid bilayer, which is observed in the single channel recording of the K+ current (Fig. 5D). While the open probability (popen) at acidic pH was very low (1%) in the absence of TFE, popen dramatically increased as the TFE concentration increased, and reached 60% at 3% (v/v) TFE. Under the acidic pH conditions, the activation gate of the helix bundle crossing fully opens, and the low popen value in the absence of TFE represents the infrequent opening of the selectivity filter. Thus, these results revealed that the conformational equilibrium of the selectivity filter is shifted towards the conductive form by TFE. These results clearly indicated that the shift in the structural and functional equilibria of KcsA is most likely caused by the modulation of the surrounding membrane environments.
4. Membrane transporter EmrE, a small multi-drug resistance transporter in bacteria, transports a broad class of polyaromatic cation substrates, such as ethidium bromide and methyl viologen [34]. EmrE is classified as an antiporter, since the transport of the substrates against their concentration gradient is coupled to the import of two protons in the opposite direction across the inner membrane. Each monomer of EmrE is composed of 110 amino acids, which forms four transmembrane helices. Although the crystal structures and the electron crystallographic studies suggested that EmrE forms an asymmetric dimer [35,36], the topology of EmrE in lipid bilayers was under debate [37]. EmrE is considered to function through a single-site alternating access model (Fig. 6A). In this model, the transporters convert between the inward- and outward-facing conformations to transport substrates across membranes. To perform the coupled antiport, both substrates (i.e. polyaromatic cations and protons) should share a single binding site and conformational exchange only occurs when the substrate is bound. Thus, saturating EmrE with its polyaromatic cation substrate, tetraphenylphosphonium (TPP+), should drive EmrE into a two-state equilibrium (Fig. 6B). A TROSY spectrum of TPP+-bound 2H/15N-EmrE in isotropic bicelles exhibited twice as many peaks as for the monomer EmrE, with similar intensities (Fig. 6C) [38]. From this peak doubling, it was postulated that EmrE forms an asymmetric dimer as observed
92
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
Fig. 5. NMR and electrophysiological analyses of KcsA reconstituted in the lipid bilayer. (A) Methyl-TROSY spectra of KcsA in rHDL and DDM micelles. The letters P and I denote the permeable and impermeable conformations, respectively. Signals from the detergent or lipid molecules are shown by the green boxes. (B) Mapping of the residues perturbed by the different reconstitution media. The N-terminal region, which is missing in the structure, is shown as a cartoon (green). (C) TFE titration experiments in the methyl-TROSY spectra of KcsA in rHDL. The methyl-TROSY spectra were observed in the absence and presence of 1%, 2%, 3%, and 4% TFE. Regions for V76c and L59d are shown. The letters P, I, and C denote the signals from the permeable, impermeable, and closed conformations, respectively. (D) TFE effects on the single channel recordings of KcsA. (Left) Representative single channel current traces of KcsA obtained by the planar lipid bilayer system, in the absence and presence of 3% (v/v) TFE on both sides of the planar lipid bilayer. Currents were recorded at +200 mV with a symmetric K+ concentration of 200 mM. (Center) the expanded traces for the relevant parts indicated with the black bars under the current traces in the left panels. The blue dashed lines indicate the closed level, and the red ones indicate the open levels. (Right) open probability (popen) as a function of TFE concentration.
by cryoelectron crystallography and X-ray crystallography. In the asymmetric dimer, each monomer has a distinct structure with different chemical shifts, and interconversion between in the inwardfacing and outward-facing conformations could effectively transport the ligands (Fig. 6A). To detect the interconversion between the two states, TROSYselected ZZ-exchange experiments were performed. As a result, many exchange peaks were observed, suggesting that there is a widespread conformational exchange in the EmrE dimer (Fig. 6C). Rates of the conformational exchange were calculated from the peak intensity of the ZZ exchange spectra with different mixing
times. Exchange rates between two states with equal populations were kopen-in to open-out = kopen-out to open-in = 4.8 ± 0.5 s-1 for TPP+bound EmrE in isotropic bicelles (Fig. 6D). Thus, NMR data demonstrates that the entire protein is involved in the conformational exchange on the same timescale. To determine whether EmrE forms an asymmetric dimer as proposed by X-ray crystallography, the water accessibility of each residue was measured by recording 1H–15N TROSY spectra with increasing concentrations of a soluble, chelated gadolinium compound, which causes line broadenings of the nearby amide groups, due to the paramagnetic effect. As a result, some residues exhibited
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
93
Fig. 6. Conformational exchange of EmrE transporter. (A) Proposed mechanism of EmrE transporter function. Interconversion between the inward-open and outward-open conformations enables EmrE to transport its substrate from the cytoplasm to the periplasm, while transporting H+ in the opposite direction. (B) Conformational exchange proposed for antiparallel, asymmetric EmrE. The two states adopt identical structure in opposite orientations with respect to the membrane. (C) Overlay of TROSY ZZexchange (red, 100 ms mixing time) and 1H–15N TROSY heteronuclear single-quantum coherence (HSQC) spectra (black). Exchange peaks and auto-peaks are connected by blue lines, and residue numbers are indicated. (D) The best-fit of ZZ-exchange auto- (solid circles) and cross-peak (open circles) intensities as a function of mixing time yields a single global conformational exchange rate of k = 4.9 ± 0.5 s 1. (E) 1H–15N TROSY spectra of TPP+-bound EmrE in bicelles in the absence (black) and presence of 1 mM (blue) and 5 mM (red) paramagnetic gadobenate dimeglumine. Residues with different PRE effects (green circles) or the same PRE effects (dark blue) in the two states are indicated. (F) Residues with equal (dark blue) and differential (yellow) accessibility to the water in the two states are mapped on the EmrE structure. Residues protected from water are colored red. This figure was adopted from [38] with permission.
line broadening to different extents (Fig. 6E). Their amide residues are located exactly within the region expected for the interconversion of an antiparallel dimer, indicating that the conformation observed by NMR is open toward one side of the membrane (Fig. 6F). Single –molecule FRET analyses, which are complementary to NMR methods that observe the equilibrium as ensemble, also confirmed the asymmetric dimer of EmrE. Therefore, the NMR study directly observed the exchange between the inward-facing and outwardfacing conformations of EmrE. This two-state equilibrium is a key step in the transport cycle of the multi-drug transporter to export the substrate across the membrane.
5. Cell adhesion molecules Cell-cell contacts are mediated by cell adhesion molecules. In the immunological response, leukocytes adhere and transmigrate
across the endothelial barrier, and then migrate to the inflammatory site (Fig. 7A) [39]. The first step of this event is rolling adhesion, where cells undergo repetitive attachment and detachment on endothelial cells, and then gradually decrease their speed to achieve firm adhesion. Such rolling adhesion is mediated by specialized adhesion molecules, such as selectins, CD44, and integrins. It remains unclear how such transient adhesion is achieved between the cell adhesion molecules and their ligands under fluid shear stress conditions. CD44 is a major hyaluronan (HA) receptor that mediates cell adhesion in leukocyte rolling and tumor metastasis [40]. The HAbinding domain (HABD) of CD44 contains a Link module (residues 32–124) and N- and C-terminal extensions (residues 21–31 and 125–178, respectively) (Fig. 7B). In the absence of the ligand, CD44 HABD adopts a well folded structure, in which the N- and C-terminal extensions form an additional structural element that intimately contacts the Link module [41,42] (Fig. 7C, referred to
94
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
Fig. 7. Two-state equilibrium of CD44 HABD in the absence and presence of ligand. (A) Schematic drawing of leukocyte trafficking to the inflammation site. The interaction between CD44 and hyaluronan mediates the rolling adhesion. (B) The domain organization of CD44. (C) Crystal structure of the CD44 HABD in the HA-unbound state (PDB ID: 1UUH). (D) Crystal structure of CD44 HABD in complex with an HA octamer (HA8) (PDB ID: 2JCR). (E) NMR structure of CD44 HABD in complex with an HA hexamer (PDB ID: 2I83). The Link module is colored grey, and the extension segments are blue. HA is shown as a yellow stick model. (F and G) Selected regions of the 1H–15N HSQC spectra of the CD44 HABD in the absence (F) and presence (G) of HA. The relative intensities of the peaks originating from the O-state and the PD-state are shown in each spectrum. (H) ZZexchange spectra of the ligand-free HABD, acquired without (black) or with a 500 ms mixing period (red). (I) Schematic representations of the equilibrium of the CD44 HABD between the O-conformation and the PD-conformation. The sizes of the arrows drawn between the O-state and the PD-state represent the changes in the conformational equilibrium in the presence and absence of HA.
as the ordered (O)-state). The crystal structure of HABD in complex with an 8mer HA revealed that HABD still adopts the O-state with only a minor conformational change, restricted to the HA binding site (Fig. 7D) [43]. In contrast, an NMR structure of HABD in complex with a 6mer HA demonstrated that HA-binding induces a conformational rearrangement of HABD, in which the C-terminal segment (residues following T153) becomes unstructured [44] (Fig. 7E, referred to as the partially disordered (PD)-state). There-
fore, the question was raised as to whether HABD adopts the Ostate or the PD-state in the HA-bound state. In an NMR spectrum of HABD in the absence of the ligand, minor peaks corresponding to the PD-state were observed, in addition to the major peaks corresponding to the O-state (Fig. 7F) [45]. Based on the peak intensities, the relative populations of the Ostate and the PD-state in the absence of HA are 86% and 14%, respectively. Conversely, in the presence of HA, minor peaks were
N. Nishida et al. / Journal of Magnetic Resonance 241 (2014) 86–96
observed at the position of the O-state with a reversed population (i.e. O-state: 8%, PD-state: 92%) (Fig. 7G). In addition, ZZ-exchange experiments [26] demonstrated that HABD exists in an equilibrium between the O-state and the PD-state, with a timescale of hundreds of milliseconds (Fig. 7H). These results indicated that CD44 HABD exists in an equilibrium between the O- and PD-states, and the HA-binding induces a shift of the equilibrium toward the PD-conformation (Fig. 7I). Therefore, the HA-bound crystal structure can be interpreted as representing the O-state in the HAbound state, which is less than 10% of the population, according to the NMR analyses (Fig. 7I). The functional relevance of the two-state equilibrium of the CD44 was examined by utilizing a mutant (Y161A) that constitutively adopts the PD-state, by destabilizing the interaction between the C-terminal segment and the Link module, which is an important interaction for the O-state (Fig. 8A). SPR experiments demonstrated that the PD-state mutant exhibited seven-fold higher affinity for HA than the wild type protein, which predominantly adopts the O-state. Therefore, CD44 HABD exchanges between the low affinity (O-state) and high affinity (PD-state) conformations. To investigate the role of the two state-equilibrium in the context of cellular function, the cell rolling behaviors were analyzed under various shear stresses conditions, using VMRC-LCD cells expressing the wild type CD44 and the PD-state mutant [45]. Although the wild type CD44-expressing cells showed smooth rolling on the HA-coated surface, the PD-state mutantexpressing cells only exhibited firm adhesion (Fig. 8B). In order to smoothly roll along the ligand-coated surfaces for cells, new bond formation at the edge of the rolling front and breakage at the rear edge need to occur continuously, on an appropriate timescale (Fig. 8C) [46]. The impaired rolling behavior of the PD-state mutant suggested that the transition between the O-state and the PD-state in the ligand-bound form is necessary for the proper detachment of the receptor-ligand bond on the substrate-coated surface. This hypothesis is also consistent with the fact that the timescale of the conformational exchange of HABD is comparable
95
to the dwell time of the ligand-receptor bond formed on the cell surface (hundred millisecond order). Therefore, the intrinsic conformational equilibrium of the HABD between the low- and high affinity states is crucial for the CD44-mediated cell rolling.
6. Conclusions and future perspectives As described, a number of NMR studies have demonstrated that the conformational dynamics of cell surface membrane proteins are directly related to their functions. NMR methods are suitable for detecting protein dynamics on various timescales. Although the slower dynamics on NMR timescale were mainly characterized in this review, the dynamics at the different timescale should be explored. For example, the CPMG relaxation dispersion analyses of M2 proton channel of influenza A detected the millisecond timescale conformational exchange that is related to opening of the channel [47]. In addition, the membrane proteins may undergo conformational exchange between the ground state and the excited state. The paramagnetic relaxation enhancement experiments or relaxation dispersion experiments would be useful for obtaining structural parameters of sparsely populated species (less than
