Diastereomeric Cyclopentane-Based Maltosides (CPMs) as Tools for Membrane Protein Study

Abstract

Amphiphilic agents, called detergents, are invaluable tools for studying membrane proteins. However, membrane proteins encapsulated by conventional head-to-tail detergents tend to denature or aggregate, necessitating the development of structurally distinct molecules with improved efficacy. Here, a novel class of diastereomeric detergents with a cyclopentane core unit, designated cyclopentane-based maltosides (CPMs), were prepared and evaluated for their ability to solubilize and stabilize several model membrane proteins. A couple of CPMs displayed enhanced behavior compared with the benchmark conventional detergent, n-dodecyl-β-D-maltoside (DDM) for all the tested membrane proteins including two G-protein-coupled receptors (GPCRs). Furthermore, CPM-C12 was notable for its ability to confer enhanced membrane protein stability compared with the previously developed conformationally rigid NBMs (JACS, 2017, 139, 3072) and LMNG. The effect of the individual CPMs on protein stability varied depending on both the detergent configuration (cis/trans) and alkyl chain length, allowing us draw conclusions on the detergent structure–property–efficacy relationship. Thus, this study not only provides novel detergent tools useful for membrane protein research but also reports on structural features of the detergents critical for detergent efficacy in stabilizing membrane proteins.

Graphical Abstract

Introduction

Integral membrane proteins are essential for cell functions such as inter- or intra-cellular material transfer, signal transduction, photosynthetic electron transport, protein trafficking, and cell adhesion and comprise more than 50% of human drug targets.¹ Structural and functional information of membrane proteins is essential for fundamental understanding of their mechanism of action as well as for rational design of new drug molecules. Unfortunately, these bio-macromolecules represent only ~2–3% of 3D-resolved protein structures,² even with the recent advances in cryo-electron microscopy and the substantial successes achieved with X-ray crystallography.³ Membrane protein extraction, purification, and structural investigation are often challenging mainly because of the low natural abundance of these molecules and their tendency to denature or aggregate once extracted from native membranes into aqueous buffer. A key prerequisite for isolation and structural studies of membrane proteins is that they must be maintained in a soluble and stable state in buffer solution by an amphiphilic additive that shields the large hydrophobic protein surfaces from polar aqueous environments. Conventional detergents with a polar head and a hydrophobic tail group such as n-dodecyl-β-D-maltoside (DDM), n-octyl-β-D-glucoside (OG), and lauryldimethylamine-N-oxide (LDAO) are widely used to extract membrane proteins from native lipid-bilayers and to maintain the native states of the proteins in solution.^4,5 However, in addition to being more dynamic than lipid assemblies, detergent micelles tend to expose hydrophobic regions of membrane proteins to buffer solution,^6,7 resulting in irreversible nonspecific aggregation. Thus, it is of great importance to develop novel agents or membrane-mimetic systems displaying favorable behaviors for membrane protein solubilization and stabilization.⁸ Notable examples of several large membrane-mimetic systems are bicelles,⁹ nanodiscs (NDs),¹⁰ amphiphilic polymers [styrene-maleic acid (SMA)^11a and diisobutylene-maleic acid (DIBMA)^11b copolymers, amphipols (APols)],¹² and peptide detergents [β-peptides (BPs),¹³ lipopeptide detergents (LPDs),^14a Salipro^14b, and peptidiscs^14c]. These agents have been shown to maintain several membrane proteins in native-like conformations but were often found to be inefficient at protein extraction and tend to form large protein-detergent complexes (PDCs). More importantly, with few exceptions (e.g., SMA),^11a they are yet to produce protein crystals with high quality. As an alternative strategy, several small amphiphilic agents have been developed as exemplified by neopentyl glycol (NG)-based amphiphiles (MNGs/GNGs),¹⁵ mannitol-based amphiphiles (MNAs),¹⁶ tripod amphiphiles (TPAs),¹⁷ calix[4]arene-based amphiphiles (C4Cs),¹⁸ tandem malonate-based glucosides (TMGs),¹⁹ penta-saccharide amphiphiles (PSEs),²⁰ butane-tetraol-based maltosides (BTMs),²¹ glycosyl-substituted dicarboxylate detergents (DCODs),²² dendronic group-containing trimaltosides (DTMs),²³ and 1,3,5-triazine- cored maltosides (TEMs).²⁴ GNG-3 and MNG-3 have contributed to the determination of more than 40 membrane protein crystal structures including a sodium-pumping pyrophosphatase, human aquaporin 2 (AQP2), and acetylcholine and opioid G-protein-coupled receptors (GPCRs) in the past ten years.²⁵ Departing from the canonical ‘polar head and nonpolar tail’ design of conventional detergents, facial amphiphiles (FAs) represent a highly innovative approach for studying membrane proteins and some of these amphiphiles (e.g., FA-5 and FA-7) were utilized for 3D crystal structure determinations of the ATP-binding cassette transporter (MsbA) and the GPCR-like bacteriorhodopsin.²⁶ Recently we developed norbornane (NB)-based maltosides (NBMs)²⁷ with two flexible alkyl arms and two maltoside head groups connected by a conformationally locked norbornane linker. Of these agents, X-NBM-C11 showed remarkable stabilization behavior with several model membrane proteins including human β₂ adrenergic receptor (β₂AR). Despite the favorable effects on protein stability, this NBM tends to form larger micelles (hydrodynamic radius (R_h)= 17.3 nm), which is potentially an unfavorable aspect for protein crystallization and NMR-based structural studies. In addition, the rigid NB linker used to build the NBMs could be associated with detergent efficacy suboptimal for protein stabilization. Herein, we made efforts to address these issues by converting the linker from the rigid NB (NBMs) to a more flexible cyclopentane (CP) unit (CPMs) (Fig. 1). This monocyclic linker provides conformational flexibility relative to the bicyclic NB linker, which can be an origin for enhanced detergent efficacy for protein stabilization. When the new detergents were evaluated with several model membrane proteins including two GPCRs, we found that CPM-C12 was significantly better than DDM and X-NBM-C11 at stabilizing the membrane proteins tested here.

Fig. 1.

Background for this study. The chemical structures of (a) previously reported X-NBM-C11 detergent with conformationally restricted maltoside head groups and (b) new cyclopentane-based maltoside-trans detergents (CPM-Ts) with the more conformationally flexible head groups. The CPM-Ts were created by disconnecting a C-C bond from the norbornane scaffold. The blue arrows show the relative conformational flexibility of the head groups. Unlike X-NBM-C11, these new detergents gave increased water-solubility and formed smaller micelles.

Results

Detergent structures and physical characterizations

The CPMs feature two alkyl chains and two dimaltosides as the hydrophobic and hydrophilic groups, respectively, connected via a monocyclic CP ring (Fig. 2). Depending on the relative orientation of the alkyl chains with respect to the head groups (cis/trans), these agents can be categorized into two sets. The two alkyl chains were connected to the C2 and C3 positions of the CP linker in a cis configuration (2R,3S) with respect to the head groups for CPM-Cs while a trans configuration was used for this connection of the alkyl chains in the case of CPM-Ts (Fig. 2). As a result, the CPM-Cs and CPM-Ts are CP variants of D-NBMs and X-NBMs, respectively; D (endo) and X (exo) notations were previously used to represent the relative orientation of the NBM tail groups with respect to their head groups.²⁷ Due to the torsional and angle strains of the central CP ring, the CPM-Cs and CPM-Ts are likely to preferentially adopt energy-minimized puckered conformations, half chair (C₂) and envelope (C_s), respectively (Fig. 2 & S1^†). This is in contrast with the conformationally locked NB linker in the NBMs. The configuration (cis/trans) and conformational variations (half chair (C₂)/envelope (C_s)) between the CPM-Cs and CPM-Ts could affect amphiphile efficacy for membrane protein stabilization in spite of their identical chemical compositions (i.e. identical polar and nonpolar segments). As hydrophile–lipophile balance (HLB) is important in determining detergent property,²⁸ we prepared detergent variants with two alkyl chain lengths (C11 and C12) for both sets of CPMs, used for detergent designation. Density functional theory (DFT) calculations at a B3LYP/6-31G* level was supportive for a half chair/twist conformation of the CP ring for CPM-C11 with a hydrophobic length of 15.2 Å, while its trans isomer (CPM-T11) was calculated to give an envelope conformation of the CP ring, with the hydrophobic length of 15.1 Å (Fig. S1^†). Thus, the two different conformations (half chair (C₂)/envelope (C_s)) along with alkyl chain length variations (C11/C12) serve as a way to change or fine-tune the detergent hydrophobic length. This is important as detergent hydrophobic length needs to be compatible with the hydrophobic dimensions of membrane protein for optimal protein stability in solution.

Fig. 2.

(a) Chemical structures of novel cyclopentane-based maltosides (CPMs) (middle-right) and their energy-optimized puckered conformers (far right). The CPM-Cs were derived from 5-norbornene-2-endo,3-endo-dimethanol, while CPM-Ts were derived from isomeric 2-exo,3-exo-dimethanol (far left). Syn-dihydroxylation using OsO₄-NMO was used for 1,2-diol generation (A/C, left). The inset within rectangle represents periodate oxidative cleavage of 1,2-diols (A and C), followed by NaBH₄ reduction to afford meso-1,5-diol derivatives (B and D). B and D are meso compounds due to the presence of a symmetry plane which perpendicularly bisects the central cyclopentane (CP) ring of the molecule (indicated by gray line in the Haworth projection). The CPM-C/Ts commonly contain a dimaltoside head group connected to the two alkyl chains using a cyclopentane linker (middle right). R or S designation was used to specify the stereochemistry of the two chiral carbons (C2 and C3). Half chair (C₂) and envelope (C_s) are two energy-minimized CP conformations of CPM-Cs and CPM-Ts, respectively, optimized by DFT calculations at the energy level of B3LYP/6-31G*) (far right).

The individual hydrophobic groups of the CPM-Cs/Ts are optically inactive meso compounds due to the presence of an internal symmetry plane dissecting the CP linker (compounds B and D in Fig. 2). Since these meso-1,5-diols (B and D) are non-superimposable stereoisomers, they are diastereomers to each other. The CPM-Cs/Ts are also diastereomers of each other, but are optically active because of the lack of an internal symmetry plane. The new agents were prepared according to a protocol comprising five high-yielding synthetic steps: (1) dialkylation, (2) alkene syn-dihydroxylation using osmium tetroxide–N-methyl morpholine-N-oxide (OsO₄−NMO) (i.e. 1,2-diol derivatives: A and C), (3) periodate-mediated oxidative cleavage of 1,2-diol, followed by in situ NaBH₄ reduction of di-aldehyde (B and D; inset in Fig. 2), (4) AgOTf-promoted glycosylation, and (5) global deprotection (see amphiphile synthesis in ESI for details). Glycosylation could generate two stereo-chemical outcomes depending on the approaching direction of a glycosyl acceptor toward a glycosyl donor, resulting in either an α or β−glycosidic bond. Consequently, the final products containing two newly formed glycosidic bonds could be a mixture of multiple diastereomers. We solved this stereochemistry issue by utilizing β-selective glycosylation attained via neighboring group participation.²⁹ The high diastereomeric purity of the CPMs was confirmed by ¹H NMR spectroscopic method (Fig. S2^† and S3^†). For example, the axial protons of CPM-C11 attached to the anomeric carbons, designated H_a, produce two narrowly separated peaks at 4.25 and 4.24 ppm as doublets (Fig. 3b & S2^†). By contrast, the same axial protons of the trans isomer (i.e., CPM-T11) gave two non-separable doublets, located at 4.26 ppm (Fig. 3c & S2^†). In addition, these anomeric protons (H_a) of both isomers interact with their neighboring protons (H) with a vicinal coupling constant (³J_aa) of 8.0 Hz, revealing that β-selective glycosylation had occurred. We also observed another doublet peak at 5.16 ppm with a relatively small coupling constant (³J_ae = 4.0 Hz), which corresponds to the α-anomeric protons (H_e) in the terminal glucose units of these detergents (Fig. 3& S2^†). Few additional peaks were detected in the α or β-anomeric region (4.3 ~ 5.2 ppm), indicative of high diastereomeric purity of the CPMs. The only detectable peaks in this region were the doublets at 4.48 and 5.09 ppm, which corresponds to the anomeric protons of maltose (Fig. S4). This disaccharide impurity in our detergent samples originates from the hydrolyzed product of perbenzoylated maltosylbromide used in glycosylation. Based on the NMR spectra of the CPMs (Fig. S2^† and S3^†), the amounts of this impurity varied from ~1 (CPM-T12) to ~5% (CPM-C11), but detergent efficacy for protein stabilization is unlikely to be affected by the presence of this highly hydrophilic compound.

Fig. 3.

(a) The chemical structure of the di-maltoside head group of the CPMs is shown to illustrate the anomeric protons of interest (H_e and H_a) and their vicinal couplings with the neighboring protons (H in blue). (b, c) Anomeric regions of the ¹H NMR spectra for CPM-C11 (b) and CPM-T11 (c) showing their high diastereomeric purity (see Fig. S2† for the full range of ¹H NMR spectra). Each isomer gave unique spectral features in the anomeric region, indicative of the clear differentiation of the individual isomers by their ¹H NMR spectra. Vicinal coupling constants (³J_aa & ³J_ae) are indicated above individual peaks to differentiate the α- and β-anomeric protons (H_e and H_a, respectively).

Because of the high efficiency of each synthetic step, the final amphiphiles could be prepared with overall yields of ~75%, making preparation of multi-gram quantities of material at a reasonable cost highly feasible.

High water-solubility (10 wt%) was found in all four new detergents, yet as for a long alkyl-chain detergent (i.e., CPM-C12/T12) a brief sonication was required for an initial dissolution (Table 1). Detergent solutions remained clear during a month of incubation at room temperature. Critical micelle concentrations (CMCs) were measured by monitoring dye solubilization using diphenylhexatriene³⁰ with increasing detergent concentration and the hydrodynamic radii (R_h) of the detergent micelles were estimated through dynamic light scattering (DLS) measurements. The summarized results for the CPMs along with D/X-NBM-C11 and DDM are presented in Table 1. The CMC values of all CPMs (from 3.8 to 6.7 μM) were more or less comparable to those of D/X-NBM-C11, but much smaller than that of DDM (170 μM), which indicates stronger tendencies to form self-assemblies than DDM (Fig. S5^†). Within the same set of detergents (e.g., the CPM-Cs), the CMC values decreased with increasing alkyl-chain length, due to the increased hydrophobicity. For instance, the CMCs of the CPM-Cs were lowered from 6.7 to 5.0 μM when the alkyl chain length increased from C11 to C12. Micelles formed by the individual sets of detergents were enlarged along with increasing alkyl chain length. Detergent micelle size increased from 4.8 (C11) to 5.5 nm (C12) for the trans isomers. Detergent micelle size is determined by the geometry of the detergent molecule, estimated by the volume ratio of detergent head and tail groups.³¹ It is interesting to note that the micelle size of the trans isomers significantly decreased with change from NB to CP linker. For instance, the R_h value of CPM-T11 micelles was 4.8 nm, smaller than X-NBM-C11 (17.3 nm). Even the C12 alkyl chain CPM (CPM-T12) formed smaller micelles than X-NBM-C11 with the shorter alkyl chain (5.5 vs 17.3 nm). This comparison reveals that the geometry of the detergent molecules is substantially changed from a cylindrical to a conical shape with the linker modification for the trans isomers. This change in detergent geometry likely originates from increased flexibility of the two maltoside head groups, resulting in an increased hydrophilic volume with little effect on the hydrophobic volume (Fig. 1). Interestingly, a different trend was observed for the cis isomers. CPM-C11/C12 were comparable to the D-NBM-C11 with respect to micelle size (3.8/4.0 vs 3.7 nm).

Table 1.

Molecular weights (MWs), critical micelle concentrations (CMCs; n = 3), water-solubility of the novel agents (CPM-Cs and CPM-Ts) and control detergents (DDM and X/D-NBM-C11), and hydrodynamic radii (R_h; n = 4) of their micelles in double-distilled water at room temperature.

Detergent	MWa (Da)	CMC (μM)	Rh (nm)b	Solubility (wt%)
CPM-C11	1147.4	6.7±0.3	3.8±0.1	~10
CPM-C12	1175.5	5.0±0.1	4.0±0.1	~10c
CPM-T11	1147.4	4.8±0.1	4.8±0.1	~10
CPM-T12	1175.5	3.8±0.5	5.5±0.1	~10c
X-NBM-C11	1145.4	6	17.3±0.2	~5c
D-NBM-C11	1145.4	7	3.7±0.1	~5
DDM	510.6	170	3.4±0.1	>10

^aMolecular weight of detergents.

^bHydrodynamic radius of detergent micelles measured at 1.0 wt% detergent concentration by dynamic light scattering.

^cSonication required to obtain a clear solution.

The variation in the conformation of the CP linker (half chair (C₂) or envelope (C_s)) is likely associated with the different behaviors of the CPM-Cs and CPM-Ts in self-assembly formation (Fig. S1^†). The DFT calculations show substantial variation in the linker conformation (NB vs CP) between D-NBM-C11 and CPM-C11, but show little variation in the linker conformation between CPM-T11 and X-NBM-C11. The CPM-Cs formed smaller micelles than the trans isomers (i.e., the CPM-Ts), as exemplified by CPM-C11 (3.8 nm) vs CPM-T11 (4.8 nm). Our results indicate that a small change in detergent architecture (i.e. just eliminating a single C-C bond) can result in a large variation in their self-assemblies, which could also affect detergent efficacy for membrane protein stabilization. When we investigated the size distribution of detergent micelles, all new agents showed only one set of micellar populations in number- or volume-weighted DLS profiles, indicative of high homogeneity (Fig. S6^† & S7^†). The appearance of a peak corresponding to large aggregates in the intensity-weighted DLS profiles results from the high sensitivity of light scattering to large particles.²⁷

Detergent micelle size was further investigated with increasing temperature from 15 to 65 °C. (Fig. S9^†). DDM gave little change in micelle size with this range of temperature variation. Consistent with the previous result,²⁷ the size of the D-NBM-C11 micelles was not affected by temperature, while micelles formed by X-NBM-C11 were substantially enlarged with increasing solution temperature. A similar trend was observed for the CPM analogs (CPM-C11 and CPM-T11), indicating that micelles formed by the endo/cis isomer are significantly more stable than the exo/trans isomer under the conditions. There was little difference in micelle sizes formed by CPM-C11 and D-NBM-C11, while micelles formed by CPM-T11 were substantially smaller than those formed by X-NBM-C11 over the temperature range tested. As the best protein stabilization efficiency was obtained from CPM-C12 (vide infra), we carried out in-depth physical characterizations for this agent along with X-NBM-C11 as a reference. X-NBM-C11 and CPM-C12 micelles investigated by multi-detection size exclusion chromatography (SEC) showed different distributions when detected using refractive index (RI) and right-angle light scattering (RALS) (Figs. 4 & S8). Micelles formed by X-NBM-C11 showed two separated peaks in the SEC profile, indicating a bimodal size distribution of these micelles (Fig. 4a). The peak at 8.4 mL corresponds to an aggregation number (N_agg) of between 90 and 180, while the peak at 9.1 mL gives N_agg of 180 ~ 750. By contrast, micelles formed by CPM-C12 gave a well-defined unimodal distribution, showing only one peak at 9.4 mL, corresponding to N_agg of 60–85 (Fig. 4b). This N_agg is much smaller than DDM micelles (~ 175).³¹ The higher N_agg of X-NBM-C11 than that of CPM-C12 reflects the bigger micelle size observed by DLS experiment (Table 1).

Fig. 4.

(a,b) SEC elution profiles and (c,d) particle size distributions from DLS for X-NBM-C11 and CPM-C12. (a) RI and RALS signals (left axis) of X-NBM-C11 showed a bimodal distribution corresponding to micellar molecular weights ranging from 860 kDa to 210 kDa across the first peak and from 210 kDa to 100 kDa across the second one, as indicated by the black line (right axis). (b) RI and RALS signals (left axis) of CPM-C12 showed a unimodal distribution giving molecular weights ranging from 70 to 100 kDa, as indicated by the black line (right axis). The black lines indicate peak widths correlated to homogeneity of detergent self-assemblies. (c) Intensity- or (d) volume-weighted particle size distribution profiles derived from DLS for mixtures of either X-NBM-C11 or CPM-C12 with POPC show a major population of small micelles. Profiles of POPC vesicles only were included for comparison. DLS: dynamic light scattering; SEC: size exclusion chromatography; RI: refractive index; RALS: right-angle light scattering.

In order to gain insights into the solubilizing efficiency of the new agents, unilamellar vesicles made of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) were separately mixed with 5 mM CPM-C12 and X-NBM-C11. Scattering intensity and mass-averaged (z-average) particle diameters decreased over incubation time for both detergents (Fig S10^†), thus demonstrating liposome solubilization. Interestingly, the solubilization kinetic of POPC vesicles was faster for CPM-C12 than for X-NBM-C11. DLS profiles following the liposome solubilization indicate the formation of the small aggregates with hydrodynamic diameters well below the initial size of the POPC vesicles (~120 nm) (Fig. 4c). Volume-weighted size distribution suggested the formation of small assemblies following detergent mixing, with hydrodynamic diameters close to 5.6 nm (CPM-C12) and 11.7 nm (X-NBM-C11) (Fig. 4d). Based on the liposome solubilization result, these two detergents along with DDM were further tested for extracting diverse membrane proteins from native Escherichia coli (E. coli) membranes. Of the tested detergents, CPM-C12 gave the highest extraction yields of various E. coli membrane proteins, followed by X-NBM-C11 and DDM (Fig. 5). The trans analog (i.e., CPM-T12) was inferior to CPM-C12 in this regard.

Fig. 5.

(a) SDS-PAGE of detergent-solubilized fractions and (b) relative protein extraction yields of CPM-C12 from the native E. coli membranes. X-NBM-C11 and DDM were used for comparison, with the amount of membrane protein extracted by 10 mM DDM serving as reference value (1.0 or 100%). Cell membrane fragments from E. coli BL21 (DE3) were incubated with three individual detergents (X-NBM-C11, CPM-C12, CPM-T12 and DDM) for 16 h at four different concentrations (1, 2, 5, and 10 mM). Band intensity in each lane was measured by densitometry using imageJ;³² error bars indicate standard errors of the mean from three or four separate solubilizations.

Detergent evaluation with diverse model membrane proteins

To assess the potential utility of new amphiphiles as tools for membrane protein study, we evaluated several model protein systems with the CPMs, using DDM and X/D-NBM-C11 as controls. The suitability of the isomeric CPMs (CPM-Cs and CPM-Ts) for membrane protein study was first investigated with the bacterial leucine transporter (LeuT), a prokaryotic homologue of the mammalian neurotransmitter: sodium symporter (NSS) family from Aquifex aeolicus.^33,34 This transporter was initially expressed and extracted from E. coli C41 (DE3) the membranes with 1.0 wt% DDM and purified in 0.05 wt% of the same detergent. DDM-purified LeuT was diluted into buffer solutions containing individual agents (CPM-C11/C12, CPM-T11/T12, D/X-NBM-C11, or DDM) to reach final detergent concentrations of CMCs + 0.04 wt% or CMCs + 0.2 wt%. We assessed protein stability by measuring its ability to bind radiolabeled leucine ([³H]-Leu) using scintillation proximity assay (SPA)³⁵ at regular intervals during a 13-day incubation at room temperature. At both detergent concentrations, LeuT in DDM-containing buffer underwent a gradual loss of protein activity over the incubation period, resulting in ~10% residual activity after the 13-day incubation (Fig. 6). Consistent with a previous result,²⁷ X/D-NBM-C11 was markedly superior to DDM in terms of preserving the functional state of the transporter. All CPMs were similar to X/D-NBM-C11 in maintaining transporter activity (Fig. 6a,b). No clear difference between the isomers (i.e., CPM-Cs vs CPM-Ts) was observed in this regard although the cis isomers look slightly better than the trans isomers. This result suggests that overall the CPM architecture is favorable for long-term LeuT stability. There was little observed difference in stability of the LeuT in the CPM agents with different stereo-chemistry.

Fig. 6.

Long-term stability of LeuT solubilized in CPMs (CPM-C11/T11/C12/T12) at the two detergent concentrations: (a) CMCs + 0.04 wt% and (b) CMCs + 0.2 wt%. X/D-NBM-C11 and DDM were used as positive controls. Ligand binding ability of the transporter was measured using the radio-labeled substrate ([³H]-Leucine (Leu)) via scintillation proximity assay (SPA). Protein stability was monitored at regular intervals during a 13-day incubation at room temperature. Data are shown as means ± SEM (error bars), n = 3.

The new agents were further investigated for the extraction and stabilization of melibiose permease from Salmonella typhimurium (MelB_St).³⁶ E. coli membrane fractions containing overexpressed MelB_St were treated with 1.5 wt% of individual detergents (DDM, D/X-NBM-C11, or CPM-C11/T11/C12/T12) for 90 min at 0 °C and the resulting detergent extracts were further incubated at an elevated temperature (45, 55 or 65 °C) for another 90 min. The amounts of soluble MelB_St under the tested conditions were quantified by Western blot analysis and expressed as percentages of the initial amount of MelB_St present in the untreated membranes. As a mild detergent is unlikely to destroy transporter integrity at low temperature, the amount of soluble MelB_St obtained at 0 °C would mainly reflect detergent extraction efficiency. If detergent-extracted MelB_St is further treated at a high temperature of 45, 55, or 65 °C, the amount of soluble MelB_St depends on detergent ability to prevent protein aggregation under the conditions tested. Consistent with a previous result,²⁷ X-NBM-C11 failed to solubilize MelB_St, while its endo isomer (D-NBM-C11) was efficient in this regard (Fig. 7a). DDM and D-NBM-C11 quantitatively extracted the transporter at 0 °C. Similar efficiencies for protein solubilization (90−100%) were observed for the CPM agents with the exception of CPM-T12. CPM-T12 was similar to X-NBM-C11 in terms of MelB_St extraction efficiency. At an elevated temperature of 45 °C, the amounts of soluble MelB_St were similar to those observed at 0 °C. When the incubation temperature was further increased to 55 °C, however, detergent efficacy for MelB_St solubilization was clearly differentiated. At this high temperature, DDM and X-NBM-C11 gave only ∼10% soluble MelB_St, while D-NBM-C11 yielded 75% solubilized MelB_St. CPM-T12 was inferior to D-NBM-C11, but the other three CPMs (CPM-C11, CPM-T11 and CPM-C12) were more effective than DDM at maintaining MelB solubility, with the best performance observed for CPM-C11and CPM-C12 (~100%). This result indicates that these CPM agents were not only efficient at extracting the transporter, but also effective at preserving the transporter in a soluble state upon heating. Additionally, the CPM-Cs appeared to be superior to the trans isomers (CPM-Ts) at maintaining MelB_St in a soluble form. To further evaluate relative detergent effectiveness to DDM, the three CPMs (CPM-C11, CPM-T11 and CPM-C12) were selected for MelB_St functional assay. MelB_St function was assessed by melibiose reversal of Förster resonance energy transfer (FRET) from tryptophan to 2′-(N-dansyl)aminoalkyl-1-thio-β-D-galactopyranoside (D²G).^36a,d,e An active transporter binds to both fluorescent galactoside ligand (D²G) and non-fluorescent substrate (melibiose). Consequently, D²G addition to active MelB_St gives a strong fluorescent signal that could be reversed by addition of a competitive melibiose as a ligand−substrate exchange occurs in the binding pocket. The DDM-solubilized MelB_St showed a response to the addition of both D²G and melibiose (Fig.7b). However, a complete loss in transporter function was observed when a less stable homologue, MelB_EC obtained from E. coli, was used under the same conditions.^36d By contrast, all the tested CPMs (CPM-C11, CPM-T11 and CPM-C12) preserved the functionality of both MelB homologues. Collectively, these three CPMs were superior to DDM at maintaining MelB in a soluble and functional form.

Fig. 7.

(a) Thermo-solubility of MelB_St solubilized in four CPM agents. DDM and D/X-NBM-C11 were used as controls. MelB_St was extracted from E. coli membranes using 1.5 wt% individual detergents for 90 min at 0°C. These detergent extracts were further incubated for another 90 min at an elevated temperature (45, 55 or 65 °C). Following ultracentrifugation to remove insoluble protein and cellular debris, the soluble MelB_St was separated by SDS-PAGE and visualized by Western blot (top panel). The amount of soluble MelB_St was expressed as a percentage of total MelB_St in the untreated membrane (Memb) and presented as a histogram (a, bottom panel). Error bars, SEM, n = 2. (b) Galactoside binding-mediated FRET reversal. Right-side-out (RSO) membrane vesicles containing MelB_St or MelB_Ec were solubilized with DDM, CPM-C11, CPM-T11, or CPM-C12. The detergent extracts were used to measure melibiose reversal of FRET from Trp to dansyl-2-galacotside (D²G). D²G at 10 μM and melibiose at a saturating concentration were added at 1 min and 2 min time points, respectively (black lines). Control data (red lines) were obtained by addition of water instead of melibiose.

We next assessed the new agents using a GPCR, the human β₂ adrenergic receptor (β₂AR).³⁷ The receptor was first extracted and purified using DDM. The DDM-purified receptor was diluted in buffer solutions supplemented with either the new individual agents without cholesteryl hemisuccinate (CHS) or DDM with CHS. The final detergent concentration was 0.2 wt% for all tested detergents. As a direct assessment of receptor stability, the ability of the receptor to bind the radioactive antagonist ([³H]-dihydroalprenolol (DHA)) was measured.^38–40 Preliminary results were obtained by measuring the initial ability of the detergent-solubilized receptor to bind the ligand. All CPM agents were as effective as DDM at maintaining receptor activity (Fig. 8a). In order to further investigate detergent efficacy, ligand binding activity of the receptor solubilized in the individual detergents was monitored at regular intervals over a 6-day incubation at room temperature (Fig. 8b). The DDM-solubilized receptor showed high initial activity, but rapidly lost its activity, giving only ~5% residual activity at the end of the incubation. A similar trend was observed for CPM-T12. X/D-NBM-C11-solubilized receptor retained approximately 50/30% of the initial activity at day 6. There is little difference in β₂AR stabilization between the CPMs and the NBMs, as exemplified by CPM-C11 vs D-NBM-C11 and CPM-T11 vs X-NBM-C11. CPM-C12-solubilized receptor showed the highest retention in receptor activity over the incubation period (Fig. 8b).

Fig. 8.

(a) Initial or (b) long-term ligand binding ability of β₂AR solubilized in individual detergents (DDM, CPM-Cs and CPM-Ts). DDM and X/D-NBM-C11 were used as positive controls. DDM-purified receptor was diluted into buffer solutions containing the individual new agents or DDM/CHS to reach the final detergent concentration of 0.2 wt%. Ligand binding activity of the receptor was measured using the radio-labelled ligand ([³H]-dihydroalprenolol (DHA)). Receptor activity was measured following 30-min dilution (a) or at regular intervals during a 6-day incubation (b) at room temperature. Error bars, SEM, n = 3.

The promising results with LeuT, MelB, and β₂AR prompted us to select three CPMs (CPM-C11, CPM-C12 and CPM-T11) for the further evaluation with another GPCR, namely, the mouse μ-opioid receptor (MOR).⁴¹ The individual detergents were used at 0.5 wt%. MOR stability was assessed by measuring receptor T_m via CPM assay. Along with DDM, LMNG, widely used for GPCR study, was included as a control in detergent evaluation. As expected, LMNG-solubilized MOR gave a higher T_m than found for the DDM-solubilized receptor (28.0 vs 33.7 °C) (Fig. 9). CPM-T11 was comparable to LMNG at stabilizing the receptor, while the cis isomers (CPM-C11 and CPM-C12) were notably more effective than LMNG. MOR solubilized in CPM-C11 and CPM-C12 gave T_ms of 36.3 and 39.3 °C, respectively. Receptor T_m was further increased by 5.6 °C when solubilized in CPM-C12 instead of LMNG, indicating the promise of this agent for GPCR structural study, particularly when combined with the β₂AR result.

Fig. 9.

Melting temperatures (T_m) of MOR solubilized in the designated detergents and derivative functions (normalized) of CPM profiles. T_m values were obtained from the derivative functions of the CPM profiles. For CPM assay, the receptor was solubilized in DDM (a), LMNG (b), CPM-C11 (c), CPM-T11 (d), and CPM-C12 (e) and temperatures of individual samples were increased from 15 to 70 °C. The value in parenthesis represents average receptor T_m ± SEM (n = 3). ‘1^st’, ‘2^nd’, and ‘3^rd’, represent experimental numbers of three independent samples.

As CPM-C12 was most effective in stabilizing multiple membrane proteins, we investigated whether this agent can be effectively used for protein extraction/solubilization and purification. To this end, we utilized the prokaryotic voltage-dependent potassium ion channel KvAP cloned from Aeropyrum pernix and expressed in E. coli.⁴² Cells were subjected to homogenization by sonication, followed by two-step centrifugation (7,000 and 100,000 g), and the resulting KvAP-containing membranes were incubated with CPM-C12 at a range of concentrations from 0 to 3.0 wt% for 3 h at room temperature. The amounts of KvAP solubilized under these conditions were estimated by SDS-PAGE (Fig S11^†) and Western blot (Fig. 10a). CPM-C12 was efficient at KvAP extraction and solubilization, and the amount of solubilized KvAP showed little dependency on detergent concentration over the range of 0.5–3.0 wt% tested. In order to gauge the compatibility of this detergent with protein purification, KvAP extracted and solubilized using 1.0 wt% CPM-C12 was loaded onto immobilized metal affinity chromatography (IMAC). CPM-C12-purified KvAP was eluted in good yields upon addition of 1.5 column volumes of elution buffer containing 0.25 wt% CPM-C12 and 400 mM imidazole at pH 8.0 (Fig. 10b and Fig. S12^†). Combined together, these results indicate that CPM-12 can be effectively used for membrane protein extraction and purification.

Fig. 10.

Western blots of prokaryotic voltage-dependent potassium channel (KvAP) over (a) E. coli BL21 membrane preparation and solubilization and (b) extraction and purification of the channel protein by immobilized metal affinity chromatography (IMAC) using Talon Co²⁺ beads. CPM-C12 was used at seven different concentrations (0, 0.5, 1, 1.5, 2, 2.5 and 3 wt %) (a) or 1.0 wt % (b) for KvAP extraction. CE: crude extract; S1 and P1: supernatant and pellet after the first centrifugation (7000 g); S2 and P2: supernatant and pellet after the second centrifugation (100,000 g); So: solubilized material; S3 and P3: supernatant and pellet after the third centrifugation (100,000 g); FT: flow through; WF: washed fraction; E1: eluted fraction.

Discussion

Membrane proteins display different tendencies to denature or aggregate in solution because of large variations in the structures and properties. This is the reason why we lack a magic bullet detergent suitable for working with all membrane proteins. Despite the protein-specific nature of detergent efficacy for protein stabilization, DDM is widely used for membrane protein research, and thus is an accepted gold standard for membrane protein structural study. In the current study, we developed a novel class of diastereomeric amphiphiles (CPMs) based on our previous NBM study and evaluated their efficacy for protein stabilization with multiple membrane proteins (LeuT, MelB, β₂AR, and MOR). The best detergent was CPM-C12, which turned out to be markedly superior to DDM for all the membrane proteins tested here. In addition, this C12 alkyl-chained CPM was even better than X/D-NBM-C11 at stabilizing the membrane proteins. When compared with LMNG, a particularly optimized detergent for GPCR stability, CPM-C12 was notably better than this NG class detergent for stabilization of two GPCRs (β₂AR and MOR). CPM-C12 was better at stabilizing β₂AR than X-NBM-C11, an agent shown previously to be better than LMNG at stabilizing the receptor in our previous NBM study.²⁷ CPM-C12 gave 5.6 °C higher T_m of the receptor than LMNG in the case of MOR. Furthermore, CPM-C12 was efficient at extracting membrane proteins (MelB_St) and successfully used for both solubilization and purification of the channel protein (KvAP). Thus, these results reveal that this CPM will find wide use in studying membrane proteins, particularly for GPCRs. Development of such detergents with enhanced protein stabilization efficacy and good protein extraction efficiency is challenging as multiple detergent properties need to be individually optimized within a single small architecture.

It is important to identify structural features responsible for the observed superiority of CPM-C12 over X/D-NBM-C11 in terms of protein stabilization. These detergents (CPM-C12 vs X/D-NBM-C11) mainly differ in their core structure (CP/NB) and alkyl chain length (C11/C12). In order to unravel the effects of the detergent core unit on protein stability, it is necessary to compare detergent efficacy between a pair of detergents with the same alkyl chain length and the same relative configuration of detergent head and tail groups (e.g., CPM-C11 vs D-NBM-C11). CPM-C11 was similar to D-NBM-C11 at stabilizing the membrane proteins (LeuT, MelB_St and β₂AR). A similar trend was observed for the trans/exo versions (CPM-T11 vs X-NBM-C11), with the exception of MelB_St stability. This comparison indicates that the presence of a CP rather than an NB core is unlikely to be responsible for the favorable protein stabilization behavior of the CPMs compared to the NBMs. In other words, the increase in the hydrophilic group flexibility attained by the introduction of the CP core is not a direct reason for the enhanced protein stabilization efficacy of CPM-C12 relative to X/D-NBM observed here. Rather, the increased hydrophilic group flexibility appears to give an indirect effect on protein stability as it conferred enhanced water-solubility to the CPM molecules, allowing for preparation of the most effective detergent with good water-solubility (i.e., CPM-C12). As a result, we conceive that CPM-C12 was observed to be better for protein stabilization than D/X-NBM-C11 not because of enhanced flexibility of this CPM in the hydrophilic region, but because of the increased alkyl chain length from C11 to C12. The increase in alkyl chain length could provide greater compatibility of CPM-C12 to the hydrophobic dimensions of protein surfaces as compared with D/X-NBM-C11. In addition, this alkyl chain extension endows detergent molecules with stronger interactions with the protein hydrophobic surfaces, leading to effective prevention of protein aggregation. At first glance, this conclusion seems inconsistent with the general concept that detergent flexibility is critical for membrane protein stabilization.^43,44 However, such detergent flexibility is associated with detergent hydrophobic group rather than hydrophilic group. Thus, the current result is still compatible with the previous results, yet it implies a distinctive role for detergent flexibility associated with the head group in protein stability.

It is interesting to compare the cis and trans versions of CPMs in terms of protein stabilization as this configuration difference generates variation in a relative direction of the detergent head and tail groups. The relative efficacy of the CPM-Cs and CPM-Ts for protein stabilization was dependent on detergent alkyl chain length (C11/C12) and a target membrane protein (MelB, MOR, LeuT, or β₂AR). In the cases of C11 versions, the cis-configured CPM (CPM-C11) was slightly better than the trans isomer (CPM-T11) for MelB and MOR stability, while an opposite trend was observed for β₂AR stability. When it comes to the C12 versions, CPM-C12 was clearly superior to CPM-T12 at stabilizing MelB and β₂AR, while little difference was observed for LeuT stability. Thus, there was no clear-cut trend of detergent efficacy between the CPMs with the cis and trans configurations. However, the CPM-Cs showed overall favorable behaviors for stabilizing the membrane proteins compared to the trans counterparts. The general outperformance of the cis isomers compared to the trans isomers is likely associated with a difference in conformation of the CP core unit (half-chair or envelope) between these stereoisomers, in addition to the relative configuration of the tail group to the head group (cis/trans). The conformation of the CP ring not only determines the relative directions of the alkyl chains and two maltose groups, but also affects detergent hydrophobic length and molecular symmetry. The half-chair conformation of the CP ring in the CPM-Cs places the two alkyl chains in a non-parallel arrangement: one alkyl chain in an axial position and the other alkyl chain in the equatorial position. This asymmetric conformation results in a difference in the effective chain length between the two alkyl chains. Membrane proteins have uneven protein hydrophobic surfaces and thus the difference in the effective chain length of the CPM-Cs could be beneficial for favorable interactions with membrane proteins than the trans counterparts (i.e., CPM-Ts).

Conclusions

Through variations of stereochemistry and alkyl chain length, we report herein two sets of diastereomeric cyclopentane-based maltosides. Of the new agents, we found CPM-C12 to be markedly more effective than optimized novel detergents (X-NBM-C11 and LMNG) as well as a gold standard detergent (DDM) at stabilizing GPCRs. Successful extraction and purification of a voltage-gated potassium ion channel, along with the efficient solubilization of MelB_St and E. coli membrane proteins, further underline the utility of this agent. Hence, this CPM represents an invaluable tool for membrane protein structural study. Additionally, detergent comparison enabled us to propose roles for both detergent flexibilities associated with the hydrophilic group and detergent core conformation in protein stability, which will assist rational design of novel detergents in future.