Anesthetic Effects on the Structure and Dynamics of the Second Transmembrane Domains of nAChR α4β2

Abstract

Channel functions of the neuronal α4β2 nicotinic acetylcholine receptor (nAChR), one of the most widely expressed subtypes in the brain, can be inhibited by volatile anesthetics. Our Na⁺ flux experiments confirmed that the second transmembrane domains (TM2) of α4 and β2 in 2:3 stoichiometry, (α4)₂(β2)₃, could form pentameric channels, whereas the α4 TM2 alone could not. The structure, topology, and dynamics of the α4 TM2 and (α4)₂(β2)₃ TM2 in magnetically aligned phospholipid bicelles were investigated using solid-state NMR spectroscopy in the absence and presence of halothane and isoflurane, two clinically used volatile anesthetics. ²H NMR demonstrated that anesthetics increased lipid conformational heterogeneity. Such anesthetic effects on lipids became more profound in the presence of transmembrane proteins. PISEMA experiments on the selectively ¹⁵N-labeled α4 TM2 showed that the TM2 formed transmembrane helices with tilt angles of 12° ± 1° and 16° ± 1° relative to the bicelle normal for the α4 and (α4)₂(β2)₃ samples, respectively. Anesthetics changed the tilt angle of the α4 TM2 from 12° ± 1° to 14° ± 1°, but had only a subtle effect on the tilt angle of the (α4)₂(β2)₃ TM2. A small degree of wobbling motion of the helix axis occurred in the (α4)₂(β2)₃ TM2. In addition, a subset of the (α4)₂(β2)₃ TM2 exhibited counterclockwise rotational motion around the helix axis on a time scale slower than 10⁻⁴ s in the presence of anesthetics. Both helical tilting and rotational motions have been identified computationally as critical elements for ion channel functions. This study suggested that anesthetics could alter these motions to modulate channel functions.

1. Introduction

Nicotinic acetylcholine receptors (nAChRs) belong to a superfamily of ligand gated ion channels involving the rapid chemical transmission of nerve impulses at synapses. Previous studies found that some subtypes of nAChRs might be potential targets of general anesthetics and their normal channel functions could be inhibited by general anesthetics [1–6]. Neuronal α4β2 nAChR is one of the subtypes sensitive to general anesthetics [4, 5, 7]. It is also one of the most abundant nAChR subtypes in the brain. Despite ample evidence showing that general anesthetics could alter α4β2 nAChR functions, it remains largely unclear how anesthetics perturb the protein structures and dynamics that ultimately affect the protein functions. Therefore, the insights of anesthetic modulation on the structure and dynamics of the α4β2 nAChR are valuable to resolve a long time mystery of the molecular mechanisms of general anesthesia [8, 9].

A comprehensive understanding of anesthetic action on α4β2 nAChR or other nAChR subtypes has often been restricted by limited available structural information of nAChRs in the past. The structural models of the closed- and open-channel α4β2 nAChR have been generated [10] recently via computations using the known structure of the Torpedo nAChR as a template [11]. More recent X-ray structures of pentameric ion channels in closed- and open-channel forms from Erwinia chrysanthemi and Gloeobacter violaceus provide high-resolution structural information relevant to nAChRs [12–14]. These new developments certainly facilitate the understanding of anesthetic action on α4β2 nAChR, but experimental studies of structural and dynamic effects of anesthetics on α4β2 nAChR may lead directly to insights into how anesthetics act on α4β2 nAChR and alter the protein function.

Solid-state NMR spectroscopy is a powerful technique for the characterization of membrane protein structures and dynamics and for the investigation of ligand-protein interactions [15–23]. The polarization Inversion and Spin Exchange at the Magic Angle (PISEMA) experiment [24] is particularly useful for the determination of topological structures and dynamics of helical proteins in a well-oriented lipid environment [25–29]. Ligand binding to the Torpedo nAChR was comprehensively analyzed using static ²H and cross polarization magic angle spinning (CPMAS) ¹³C solid-state NMR experiments [30, 31]. The structures of the transmembrane domains of the Torpedo nAChR were also examined by various solid-state NMR methods [32–34], including PISEMA [33]. However, no solid-state NMR study on neuronal nAChR has been reported previously. In the present study, we embedded the second transmembrane domains (TM2) of α4β2 nAChR in lipid bicelles, which served as membrane mimetic media and magnetically aligned the protein. The structural and dynamic properties of the TM2 α4β2 nAChR in the absence and presence of anesthetics halothane or isoflurane were investigated using solid state NMR, especially PISEMA experiments. Anesthetics were found to affect both helical tilting and rotational motions that have been identified computationally as critical elements for ion channel functions. This study suggested that anesthetics could alter these motions to modulate channel functions.

2. Materials and Methods

2.1 Materials and Sample Preparation

The second transmembrane (TM2) domains of the human nAChR were obtained by solid phase synthesis [35, 36]. The α₄ and β₂ TM2 domains have the sequences of EKITLCISVLLSLTVFLLLITE and EKMTLCISVLLALTVFLLLISK, respectively. In order to simplify the studies, only seven leucine residues in α₄ TM2 domain, as indicated in bold letters, were ¹⁵N labeled.

Lipids were purchased from Avanti Polar Lipids (Alabaster, AL). A conventional protocol of bicelle preparation [37] was followed. 1,2-dimyristoyl-sn-glycero-3-phosphocoline (DMPC) and 1,2-dihexyl-sn-glycero-3-phosphocholine (DHPC) were mixed in a desired molar ratio (q=3.2). Lipid concentration was 28% of sample volume (typically 220 µL). To extend sample stability for samples containing the nAChR TM2 domains, 1,2-O-ditetradecyl-sn-glycero- 3-phosphocoline (14-O-PC)) and 1,2-O-dihexyl-sn-glycero-3-phosphocholine (6-O-PC) replaced DMPC and DHPC, respectively. A small amount of deuterated DMPC_d54 (~1 mg) was added to each sample for H² NMR. The α4 or β2 TM2 was dissolved in 100 µL trifluoroethanol and added to the 6-O-PC-chloroform solution. The organic solvents were removed under a stream of nitrogen gas, followed by further evaporation under high vacuum overnight. The aqueous solutions were prepared by adding 110 µL of deuterium-depleted water to dried 14-O-PC and 6-O-PC/peptide. The 14-O-PC suspension was vortexed extensively followed by three freeze/thaw cycles (liquid nitrogen/42°C). The 14-O-PC suspension at 42 °C was then added to the 6-O-PC/peptide, followed by vortexing and three freeze/thaw cycles. Slow-speed centrifugation was sometimes necessary to remove air bubbles in the sample. A transparent solution was obtained that was viscous at 38 °C and fluid at 4 °C. Parallel-oriented peptide-containing bicelles were prepared by adding 100 mM YbCl₃·6H₂O to the peptide-bicelle solution to reach a final lanthanide concentration of 3 mM. The sample was transferred to a 5 mm OD glass tube (New Era Enterprises, Newfield, NJ) using a pre-cooled pipet tip at 4 °C. The glass tube was sealed with a tight fitting rubber cap and further sealed with hard bees wax. The α4-TM2 concentration was 7.4 mM in the α4 samples and 3 mM in the α4β2 samples. The molar ratio of β2 to α4 was 1.5 in the α4β2 samples to ensure a formation of (α4)₂(β2)₃. The peptide to lipid ratio is ~1:60. Anesthetics halothane (2-bromo-2-chloro-1,1, 1-trifluoro-ethane) and isoflurane (2-chloro-2-(difluoromethoxy)-1,1,1-trifluoro-ethane) were added directly to the NMR tube (pre-cooled in an ice bath) and mixed thoroughly with the samples. The anesthetic concentrations were determined using ¹⁹F NMR with 5 mM trifluoroacetic acid (TFA) as a reference.

2.2 NMR Spectroscopy

All solid state NMR experiments were performed at 40 °C on a Bruker Avance 600 MHz NMR spectrometer equipped with a ¹H/³¹P/¹⁵N(²H) Bruker flat-coil probe. ¹⁹F NMR experiments were performed on a Bruker Avance 600 spectrometer with a Bruker TXO probe. The 1-D ¹⁵N cross-polarization experiments were conducted using a ¹H 90° pulse of 5.1 µs, a 49 kHz ¹H-decoupling field, 1 ms contact time, 3 s recycle delay, and 10,000 scans. The same cross-polarization parameters were used in the 2-D ¹⁵N-¹H PISEMA [38]. Other parameters included ±35 kHz frequency jumps for the Lee-Goldberg condition, 400–1000 scans, and 6–8 s recycle delay. The ¹H carrier frequency was set at ~4.5 ppm on parallel oriented bicelle samples. The ¹⁵N chemical shift frequencies were referenced to solid ammonium sulfate at 26.8 ppm (relative to liquid ammonia at 0 ppm). The ²H NMR experiments were performed using a solid echo sequence ((π/2) − τ − (π/2) − τ). The 90° pulse length was 2 µs and τ values were ~40 µs (perpendicular bicelle) and ~20 µs (parallel bicelle). The recycle delay of 0.5 s and the scans of 256 to 10,000 were used.

2.3 NMR Data Processing

All NMR spectra were processed using NMRPipe [39] and analyzed using Sparky [40]. The data from PISEMA experiments were fitted to the PISA (Polarity Index Slant Angle) [41, 42] wheels using the program developed by Veglia’s group [26]. The tilt angle (θ) of the helix axis with respect to the bicelle normal, rotation angle (ρ) of the helix around its axis, the angle between the N-H bond and helix axis (δ), and the dipolar coupling constant (K_DD) were determined via fittings. The principal values of the ¹⁵N chemical shift tensors [24] for non-glycine residues, σ11 = 64 ppm, σ22 = 77 ppm, and σ33 = 217 ppm, were adopted for the fittings. These values were multiplied uniformly by a factor of 0.95 to accommodate reduced ¹⁵N chemical shift anisotropy of proteins in bicelles, as reflected in the reduction of spectral span of the same proteins in bicelles in comparison to those in mechanically aligned lipids.

The TM2 chemical shift assignments were accomplished by combining several sources of information, including the previous assignments of δ–TM2 nAChR [33], the solution NMR structure of β2-TM2 nAChR [36], and the best fitting of PISA wheel. L19 is distinctly separated from the rest of leucine residues in a top view of the NMR structure (pdb code: 2K59) [36]. L19 also has weaker NMR signal than other leucines, presumably because it is located at the lipid-water interface and has less protection from solvent exchange. The distinct location and intensity of L19 convinced us to assign L19 to the most downfield peak in the α4-TM2 PISEMA spectra. Once L19 is defined, the rest of peaks were assigned automatically by PISA wheel fitting. The final assignment our α4-TM2 PISEMA spectra matches very well with a 2-D projection of the leucine distribution along helical axis from N- to C-terminus in the solution NMR structure [36]. The assignment also agrees well with that for δ–TM2 nAChR [33].

The orientational order parameter, $S_i^{\mathit{\text{CD}}}$ , of the ith CD bond vector can relate to the residual quadrupolar splitting, $\mathrm{\Delta }{\nu }_Q^i$ , in a ²H NMR spectrum using the equation [43]

$$\mathrm{\Delta }{\nu }_Q^i=\frac{3}{2}{A}_Q\frac{3{\text{cos}}^2\theta -1}{2}{S}_i^{\mathit{\text{CD}}}$$ (1)

where A_Q = e²qQ/h =167 kHz is the static deuterium quadrupolar coupling constant for aliphatic CD bonds, θ is the angle between the bilayer normal and the magnetic field.

2.4 The Na⁺ flux assay and Confocal Fluorescence Microscopy

The Na⁺ flux assay, as measured by the enhancement of Sodium Green™ dye (Invitrogen, Carlsbad, CA) fluorescence due to Na⁺ entry into vesicles through open nAChR channels, is an effective way to assess nAChR activity macroscopically. We prepared 30 mM large lamellar vesicles with 250 µM α4β2. The vesicles contain phosphatidylcholine (PC) and phosphatidylglycerol (PG) in a 4:1 molar ratio [44]. PC and PG were mixed in chloroform. α4 and β2 were mixed in a molar ratio of 2:3 and dissolved in TFE. The dissolved lipids and α4β2 were mixed and dried to a thin film. Residual organic solvent was removed by vacuum overnight. The lipid-protein mixture was hydrated with a buffer solution at pH 7.5 containing 20mM Tris, 50mM CaCl₂, and 6uM membrane impermeable Sodium Green™. The vesicles were obtained by incubating the lipid-protein solution at 42°C overnight and subsequent multiple cycles of freeze and thaw. An extensive dialysis was performed to remove Sodium Green™ dye outside the vesicles.

The Na⁺ flux assay was performed using an Olympus Fluoview300 Confocal Laser Scanning head with an Olympus IX70 inverted microscope (Olympus, Melville, NY). Sodium Green™ was excited by the 488nm argon laser line and the emission was detected using sharp cutoff 510IF long-pass and BA530RIF short-pass filters. For each measurement, 50 µL vesicles containing α4β2 were added into a tray coated with poly-lysine that facilitated vesicle adhesion due to the interactions between negatively charged PG and positively charged poly-lysine. The image acquisition began before the addition of an isotonic 100 mM NaCl solution to media outside the vesicles and continued for 200 s after the exposure to Na⁺. The program MetaMorph was used for analyzing the image data.

3. Results and Discussion

3.1 Anesthetic Effects on Bicelles

Solid-state ²H NMR spectra of bicelles with deuterated DMPC_d54, as shown in Figure 1, provided insight of lipid alignment and dynamics. The high-resolution ²H NMR spectra with distinguished quadrupolar splittings in Figures 1A and 1B attested that the bicelles, without incorporating proteins, were well oriented. The quadrupolar splitting of individual ²H peaks corresponding to carbon positions close to the tail of the aliphatic chains showed little difference in Figures 1A and 1B, suggesting that anesthetic halothane molecules had no access to the hydrophobic tail region. However, halothane diffused to the carbon positions close to the glycerol group, as evidenced by significantly broadened ²H signals of the region in Figure 1B. The order parameters of individual C-²H bonds of DMPC_d54, S_CD, were calculated using Eq. 1 in the presence and absence of 6 mM halothane. S_CD normally contains information on (a) collective motions of lipids, including bicelle wobbling, (b) trans-gauche isomerizations around C-C bonds of lipids, and (c) the anisotropic reorientation of a whole lipid molecule. Collective motions of lipids were small in our systems, in which a 72% hydration level was remained and bicelles were well magnetically oriented. A recent comprehensive NMR analysis also suggested a good bicelle alignment at such a hydration level [45]. The contribution of anisotropic lipid reorientation to S_CD was expected to be uniform for all C–²H bonds linked to a rigid structure. Therefore, a small increase in S_CD values in the presence of halothane in Figure 2 predominately resulted from orientation of the C–²H bonds with respect to the principal axis of motion of individual lipid molecule. However, since only the readable quadrupolar splitting was used for S_CD calculations, the disordered component reflected in the broader peaks might be underestimated in Figure 2.

Figure 1.

²H NMR spectra of perdeuterated DPMC_d54 phospholipids embedded in (A) a pure DMPC/DHPC (q=3.2) bicelle system; (B) after adding 6 mM halothane to the system in (A); (C) the 4OPC/6OPC bicelle system incorporated with (α4)₂(β2)₃ nAChR TM2 domains; (D) after adding 6 mM halothane to the system in (C). The bicelle normal was perpendicular to the magnetic field in (A) and (B), parallel to the magnetic field in (C) and (D) due to the addition of 3 mM lanthanide. The spectra were acquired with the solid echo pulse sequence at 40 °C. The spectrum center was arbitrarily set to 0 Hz.

Figure 2.

C-²H order parameter S_CD as a function of DPMC_d54 carbon positions in the absence (Δ) and presence (■) of 6 mM anesthetic halothane.

Figures 1C and 1D show the solid-state ²H NMR spectra of the bicelles incorporated with the (α4)₂(β2)₃ TM2 in the absence and presence of 6 mM anesthetic isoflurane, respectively. The bicelles were flipped 90° by YbCl₃ so that their quadropolar splittings were double of that in Figures 1A and 1B. The presence of the (α4)₂(β2)₃ TM2 made the ²H peaks much broader, presumably because the lipid bilayer was severely disturbed by the insertion of the TM2 domains. A small amount of lipids were in isotropic phase after incorporating TM2 domains into the bicelles and the amount increased upon addition of 6 mM isoflurane, as reflected in the peak at zero frequency. The degree of anesthetics-induced ²H peak broadening in Figure 1D seemed to be more severe than that in Figure 1B, especially to the carbon positions at the middle of DMPC aliphatic chains. Formation of the (α4)₂(β2)₃ channels might facilitate anesthetic diffusion into the deep bilayer where anesthetics normally have less access. Direct interactions between anesthetics and transmembrane proteins were observed previously [35, 46–48]. Anesthetics could modulate protein dynamics directly and allosterically [46]. Such changes in protein motions could affect lipids surrounding proteins and contribute to the observed lipid peak broadening in Figure 1D.

3.2 TM2 Domains in Channel and Non-Channel Forms

The α4 and β2 subunits of nAChR need to be in 2:3 stoichiometry to form pentameric channels [49]. Our Na⁺ flux experiments confirmed that ion channels could be formed by combining the α4 and β2 TM2 domains in a 2:3 molar ratio, but not by the α4 TM2 domain alone. As shown in Figure 3, large vesicles made of phosphatidylcholine (PC) and phosphatidylglycerol (PG) with (α4)₂(β2)₃ TM2 could be easily identified. A time-dependent increase of sodium green fluorescence intensity upon injection of 10 µL 100mM NaCl signaled channel formation. Control vesicles without channels showed no increase in fluorescence intensity, nor did the channel-containing vesicles when isotonic CaCl₂ was added instead of the NaCl solution. The same experiment was also performed on the PC-PG vesicles containing the α4 TM2 domain alone and found no indication of channel formation, suggesting different topology and dynamics of the α4 and (α4)₂(β2)₃ TM2 assemblies.

Figure 3.

(A) Confocal fluorescence images of large unilamellar vesicles (LUV) made of phosphatidylcholine and phosphatidylglycerol containing α4β2 nAChR channels after exposure to ~100 mM NaCl solution at different time points. Membrane-impermeable Sodium Green fluorescent dyes were enclosed and trapped inside the LUVs to indicate intra-vesicle Na+ concentration. The fluorescence intensity of the vesicles with α4β2 channels increased significantly within a short period of time after exposure to extra-vesicle NaCl, indicating an influx of Na+ through the channels. (B) The intensity changes in the circled vesicles over time after the exposure to extra-vesicle Na⁺. The background refers to the region without vesicles. The control refers to vesicles without channels, whose fluorescence intensity remained constant before and after exposure to extra-vesicle Na⁺.

The solid state ¹H-¹⁵N dipolar /¹⁵N chemical shift correlation PISEMA spectra of the ¹⁵NLeu labeled α4 TM2 in Figure 4 demonstrated distinct differences in the absence and presence of the unlabeled β2 TM2. Wheel-like patterns in the PISEMA spectra of parallel 14-O-PC/6-O-PC bicelles confirmed the transmembrane helical structure of the TM2 domains [36]. Six out of seven total leucine residues of the α4 TM2 showed amide resonances between ~180 and 200 ppm of ¹⁵N chemical shift. Their resonance patterns are similar to those composed by the TM2 of the δ nAChR [33]. Leu 5 at the N-terminus of the TM2 did not show up in this region of PISEMA spectra, confirming our early solution NMR finding that Leu 5 was not part of the TM2 helix [36]. Leu 19 resides close to the C terminus of the TM2. Its relatively weak intensity in the PISEMA spectra is probably due to a conformational exchange and less efficient cross-polarization in the dynamic C-terminal region. The most downfield ¹⁵N chemical shift in the PISEMA spectra is ~12 ppm smaller than the conventional value of σ33 (217 ppm), suggesting a more dynamical environment in bicelles.

Figure 4.

PISEMA spectra of the ¹⁵N-Leu labeled α4 TM2 in the absence (black) and presence (green) of unlabeled β2 TM2. Best fit simulations to the PISEMA data reveal the helical tilt angles of 12 ± 1° and 16 ±1° of the α4 TM2 and the (α4)₂(β2)₃ TM2, respectively. The notable difference between two PISEMA spectra suggests that the α4 TM2 must have experienced the existence of the β2 TM2 in the (α4)₂(β2)₃ TM2 sample.

Although the α4 TM2 had a single set of ¹⁵N-Leu resonances in the PISEMA spectrum in Figure 4, the presence of unlabeled β2 TM2 brought up several minor resonances for the (α4)₂(β2)₃ TM2. Leu10, Leu17, and Leu13 showed almost the same ¹⁵N chemical shifts in both major and minor resonances, but their dipolar couplings were about 1 kHz greater in the minor resonances. Although it was almost impossible to accurately define topology of the minor conformation due to too few numbers of resonances, the chemical shift and dipolar coupling data suggested that the helix tilt angle relative to the bicelle normal must be smaller in the minor conformation than in the major population [41, 42]. The minor peaks might also result from a sub-population α4 TM2 in a slightly different motional environment.

A more noticeable difference between the α4 and (α4)₂(β2)₃ spectra in Figure 4 is the shift of the amide resonances along both ¹⁵N chemical shift and ¹H-¹⁵N dipolar coupling axis. Leu11 and Leu18 of the α4 TM2 experienced relatively small shifts before and after mixing with the β2 TM2, but Leu10, Leu13, Leu17 and Leu19 had more profound changes. It is plausible that Leu11 and Leu18 have experienced little interaction with residues in other helices. The best fitting of all downfield resonances to ideal PISA wheels revealed the helical tilt angles of 12 ± 1° and 16 ± 1° of the α4 TM2 and the (α4)₂(β2)₃ TM2, respectively. The rotation angle, ρ, was 65° for both α4 and (α4)₂(β2)₃. The tilt angle of 12° for the α4 TM2 in bicelles agrees with the result of the δ nAChR [33], where the δ TM2 helix was also tilted 12° relative to the normal of mechanically oriented DMPC bilayers. Neither α4 nor δ subunit could form nAChR channels without partitioning of other subunit types [49]. The tilt angle of 16 ± 1° of the (α4)₂(β2)₃ TM2 is comparable to the tilt angle of 15 ± 2° found for the GABA_A receptor TM2 domain, an anion channel-forming peptide, by a recent solid-state NMR study [18].

It is worth mentioning that all the resonances in our PISEMA spectra resulted solely from the ¹⁵N-Leu residues of the α4 TM2. The unlabeled δ2 TM2 did not generate ¹⁵N NMR signals. If the α4 TM2 was isolated from the β2 TM2, the (α4)₂(β2)₃ TM2 would give the same PISEMA spectrum as the α4 TM2. The notable difference between the two spectra in Figure 4 prove that the α4 TM2 must have interacted with the β2 TM2 in the (α4)₂(β2)₃ TM2 sample. A thorough NMR characterization of oligomerization states of ²H selective labeled transmembrane peptides in oriented lipid bilayers was demonstrated previously [50]. The same method can also be applied to the α4 and β2 TM2 if proper labeled samples become available. Nevertheless, the data in Figure 4 indicate that interaction between α4 and β2 subunits that might be the driving force for assembling a functional channel. The larger helix tilt angle found in the (α4)₂(β2)₃ TM2 supported the previous prediction that channel opening might involve tilting of pore lining helices [10, 12, 51, 52].

3.3 Anesthetic Effect on the TM2 Domains

Figure 5 displays an overlay of the PISEMA spectra of the α4 TM2 in the absence and presence of 6 mM halothane. Halothane lowered ¹H-¹⁵N dipolar coupling and ¹⁵N chemical shift of six leucine residues noticeably. The data fitting into PISA wheels provided a helical tilt angle of 14° in the presence of halothane. The dipolar coupling constant, K_DD, decreased from 8700 to 8100 KHz, but the rotational angle (ρ = 65°) and the angle between N-H bone and helix axis (δ = 13°) remained the same in the absence and presence of 6 mM halothane.

Figure 5.

An overlay of PISEMA spectra of the α4 TM2 in the absence (black) and presence (red) of 6 mM halothane. PISA wheel fitting of the α4 TM2 in the presence of 6 mM halothane reveals an increase in tilt angle from 12° ± 1° to 14° ± 1°.

In comparison to the α4 TM2, the (α4)₂(β2)₃ TM2 seemed less susceptible to anesthetics. Figure 6 shows the PISEMA spectra of the (α4)₂(β2)₃ TM2 before and after adding 12 mM anesthetic isoflurane. The helix tilt angle of the (α4)₂(β2)₃ TM2 with respect to the bicelle normal changed less than 1° (θ = 16.5°). The K_DD value changed from 8400 to 7900 KHz. The other two parameters, ρ = 65° and δ = 13°, remain unchanged. To majority of the (α4)₂(β2)₃ TM2 in the sample, anesthetic isoflurane had a subtle but real impact on their helical orientation in bicelles. The anesthetic-induced changes in the major resonance pattern also implicate the possibility of a wobbling motion of the helix axis with respect to the bicelle normal [53]. Those aforementioned minor resonances of Leu10, Leu13, and Leu17 shown in Figure 4 disappeared after adding 6 mM isoflurane to the sample (see the on-line supporting information). However, another subset of minor resonances appeared upon further increasing the isoflurane concentration to 12 mM, as shown in Figure 6. The new minor peaks, labeled as L10’, L11’, L13’ L17’, and L19’ in Figure 6, appeared as if they resulted from corresponding major resonances rotating ~50° counterclockwise around the helical axis. Distinct major and minor resonances signify that the helix rotational motion was on a time scale slower than 10⁻⁴_s [54]. Both tilting and rotational motions of the TM2 helices are critical elements for channel functions [10]. Our data reveal that anesthetic molecules are able to alter motions of the TM2 helices that could account, at least partially, for anesthetic inhibition effects on the α4β2 nAChR [7].

Figure 6.

An overlay of PISEMA spectra of the (α4)₂(β2)₃ TM2 in the absence (black) and presence (red) of 12 mM anesthetic isoflurane. Notice the appearance of subset resonance peaks (a, b, c in black and L19’, L11’, L18’, L10’, L13’, L17’ in red). The minor resonance peaks in red seem to result from rotating corresponding major resonances ~50° counterclockwise around the helical axis.

It is not surprising to observe different responses of the α4 TM2 and the (α4)₂(β2)₃ to anesthetics, considering that one might exist as monomer but the other could form channels. Without the presence of the β2 TM2, the α4 TM2 interacted loosely with other α4 TM2 helices so that individual helix orientation was affected by anesthetics more severely. In a pentameric (α4)₂(β2)₃ TM2, the helix tilting in respect to the bicelle normal became less sensitive to the addition of anesthetics, presumably due to stronger interaction between the α4 TM2 and its adjacent β2 TM2. One may wonder if small changes in helix orientation induced by anesthetics, either through tilting or rotational motion, are significant enough to alter channel functions. The X-ray structures of pentameric ion channels from bacterial demonstrated that the pore-lining TM2 domains in closed- and open-channel conformations differ by merely 9° rotation around an axis that is parallel to the membrane normal [12–14], suggesting that a small change in the TM2 orientation could elicit a sizable change in channel functions. Although further structural and dynamical investigations on integral α4β2 nAChR are necessary to define a final answer, the present study highlights the possibility that anesthetics may modulate the channel function via altering the motion as well as orientation of the pore-ling domain.

4. Conclusions

Several key findings emerged from the current study. First, transmembrane proteins could facilitate anesthetic diffusion into deep membrane bilayers even though anesthetics normally prefer amphiphilic lipid-water interface region [55]. Consequently, anesthetics affected lipid alignment and conformation more severely in bicelles containing proteins. Secondly, the α4 TM2 could not form channels unless it had been mixed with the β2 TM2. The interactions between the α4 and β2 TM2 subunits do exist and such interactions may be essential to drive channel formation. Thirdly, general anesthetics could perturb orientations of the transmembrane helices in lipid bilayers and introduce changes in helical motions. The observed anesthetic effects on the tilt and rotational angles of pore lining TM2 helices reveal a potential pathway for anesthetic inhibition of channel functions. A more challenging question has arisen based on the current study: how do anesthetics make changes in transmembrane protein orientations and motions that are related to protein functions? This is certainly a question worth further investigation.