2022 Volume 70 Issue 8 Pages 514-518
Membrane cholesterol is an essential and abundant component of eukaryotic cell membranes. The unique chemical structure of cholesterol significantly influences the physicochemical properties of phospholipid bilayers, such as hydrophobic thickness and lateral pressure profile. However, the mechanisms by which these alterations regulate the balance of protein–lipid interactions in lipid bilayer environments remain unclear. To experimentally assess basic and common driving forces for helix associations in membranes, the self-associations of a de novo designed simple transmembrane helix (AALALAA)3 and its derivative helices were examined. Single-pair fluorescence resonance energy transfer (sp-FRET) experiments were performed to monitor the thermodynamic and kinetic stabilities of helix associations in single liposomes. The addition of cholesterol exerted both stabilizing and destabilizing effects on these associations, up to a change in ΔGa of approx. 10 kJ mol−1, and these effects were dependent on the association topology, amino acid sequence, and number of helices. These results demonstrate that cholesterol in the membrane regulates the stability of transmembrane proteins in a protein context-dependent manner through physicochemical mechanisms.
The cell membrane has a number of biological functions, including material transport, signal transduction, and energy conversion. The fundamental structure of the cell membrane is the lipid bilayer, which is spontaneously formed by amphiphilic phospholipids and is enriched with diverse membrane proteins that exert various biofunctions. Therefore, the lipid bilayer is a natural environment, or solvent, for transmembrane proteins, and is essential for protein stability and functions. Different types of lipids are heterogeneously distributed in cell membranes. Sphingolipids mainly exist in the outer leaflet of the plasma membrane of eukaryotic cells, whereas phosphatidylserine and phosphatidylinositol localize to the inner leaflet.1,2) Cholesterol and sphingolipid concentrations markedly differ among organelle membranes in eukaryotic cells.3) As shown in Fig. 1, cholesterol concentrations markedly increase from 1–10% in the endoplasmic reticulum (ER) to 25–50% in the plasma membrane.4) Cholesterol concentrations are controlled in animal cells.5,6) The concentration gradient of cholesterol contributes to the organelle-dependent regulation of protein activities in membranes. Previous studies reported that protein activities were strongly dependent on lipid compositions.7–9)
However, the mechanisms by which membrane physicochemical properties regulate the balance of protein–lipid interactions in lipid bilayers have not yet been elucidated. This review focuses on the basic and non-specific effects of membrane cholesterol on protein stability in lipid bilayers, and summarizes their potency based on the findings of our research using model transmembrane helices.10–12) Cholesterol and sphingolipids have been shown to cooperatively condense to form the liquid-ordered (Lo) phase, and laterally phase-separate from the liquid-disordered (Ld) phase in model membranes.3,13) Analogous to this phenomenon, the presence of Lo-like micro/nano-domains in plasma membranes (raft domains) and their relationship with biomembrane functions have been actively debated in recent years14,15); however, this topic is not the focus of this review.
The chemical structure of cholesterol markedly differs from that of typical phospholipids, such as palmitoyl-oleoyl-phosphatidylcholine (POPC) (Fig. 2A); however, POPC/cholesterol (7/3) membranes do not undergo macroscopic phase separation and have been suggested to submicroscopically demix into the Lo/Ld states.16) The rigid sterol ring of cholesterol restricts flexible fitting with the surface of proteins in the hydrophobic region, in contrast to the acyl chain of POPC. This unfavorable helix–cholesterol contact can indirectly drive the association of the proteins by exclusion around cholesterol in POPC/cholesterol membranes. Furthermore, the smaller polar headgroup (-OH) relative to the hydrophobic region changes the physicochemical properties of phospholipid bilayers. POPC exhibits a preference for geometrically flat monolayers (Fig. 2B, upper). However, the cholesterol-containing monolayer has a negative spontaneous curvature17) (Fig. 2B, middle). Higher lateral pressure (or curvature stress/strain) in the hydrocarbon core is needed to maintain the bilayer structure via the flattening of putative monolayers (Fig. 2B, lower). Acyl chains of phospholipids extend to thicken the hydrophobic region, which partially compensates for this pressure. A previous study reported that the incorporation of 30 mol% of cholesterol increased the hydrophobic thickness of the lipid bilayers by acyl chain ordering (approx. 4 Å) over that of pure POPC (approx. 27 Å).18) Similar effects due to spontaneous curvature are expected for small headgroup phospholipids, such as phosphatidylethanolamine.17)
(A) Chemical structures of POPC and cholesterol. Hydrophilic parts were positioned upwards and shaded in a blue-gray color. (B) Cholesterol-induced constraints and resulting alterations in the bilayer. See the text for details.
Figure 3 shows hypothetical cholesterol-induced excess pressure on transmembrane proteins due to changes in the lateral pressure profile.19) Lateral pressure is expressed as a function of depth in the lipid bilayer (z) (left panels).20,21) To maintain the bilayer structure, surface tension at the water–membrane interface (negative pressure) is compensated for by repulsive forces at the headgroup and hydrophobic regions (positive pressure). Therefore, the integral of the profile needs to be zero for the bilayer. In cholesterol-containing membranes, higher and lower lateral pressures are expected at the headgroup and hydrophobic regions, respectively, than in pure POPC membranes. These pressure profiles may induce an hourglass-shaped protein conformation.
Left panels show a schematic illustration of the lateral pressure profile as a function of membrane depth (z). The right lower panel shows excess lateral pressure induced by membrane cholesterol. See the text for details.
The self-association of transmembrane helices in the lipid bilayer environment has been detected by various techniques, including steric trap,22) thiol–disulfide exchange,23) and biochemical methods.24) However, difficulties are associated with detecting the dynamic/kinetic stability of this interaction using crosslinking or other indirect methods. The use of fluorescence resonance energy transfer (FRET) enables the real time and quantitative detection of helix associations in lipid bilayer environments; however, the effects of the attached fluorescent probes need to be carefully considered. The strength of the interaction may be quantitatively evaluated as association free energy (ΔGa) and other thermodynamic parameters upon association. In addition to ensemble FRET measurements (Fig. 4A), which are suitable for assessing weak helix associations (ΔGa of approximately −10 to −25 kJ mol−1),10,25–27) a single-pair FRET (sp-FRET) technique (Fig. 4B) was devised to directly detect the association–dissociation dynamics of transmembrane helices in single vesicles in time scales from milliseconds to seconds.10–12) Furthermore, the association topology (parallel or antiparallel) may be controlled using disulfide-bridged dimer helices followed by the reduction of the disulfide bond before sp-FRET measurements.11) FRET fluctuations were observed between two transmembrane helices incorporated into large unilamellar vesicles (liposomes) (diameter of approx. 100 nm) attached to a glass surface (Fig. 4C). Thermodynamic and kinetic parameters may be extracted from analyses of traces.10–12)
Sample set-up for ensemble fluorescence resonance energy transfer (FRET) (A) and single-pair FRET (sp-FRET) (B) measurements. In the sp-FRET set-up (B), liposomes containing only a few transmembrane helices were attached to a glass surface via an interaction between biotin and avidin. Fluorescence signals from the FRET donors Cy3B and acceptor Cy5 were simultaneously imaged by total internal reflection microscopy under the excitation of the donor. After measurements, the number of helices was assessed by stepwise photobleaching. We selected vesicles that had incorporated only one donor helix and one acceptor helix for analysis. (C) Representative data for sp-FRET measurements. In the top panel, the fluorescence intensity (F.I.) of the FRET donor (upper trace, green) negatively correlated with that of the FRET acceptor (lower trace, red), reflecting the monomer–dimer dynamic equilibrium of the two helices, whereas only the FRET donor was emitted in the bottom panel, indicating no detectable association between the helices.
Synthetic transmembrane peptides are useful for investigating protein folding in lipid bilayer environments because transmembrane helical structures are regarded as stable folding units for the formation of tertiary and oligomeric structures.28,29) Previous studies that used model and natural transmembrane helices reported the stabilizing effects of cholesterol on helix associations.23,30–35) Transmembrane helices are also useful in computer simulation studies to examine the effects of membrane physicochemical properties on the associations of transmembrane helices at the molecular level.36–39) My collaborators and I used three types of model transmembrane helices (1TM, GXXXG, and 2TM) to examine the effects of cholesterol (Fig. 5). The helices were prepared by Fmoc solid-phase peptide synthesis, and the N-termini were labeled with cyanine dyes Cy3B (FRET donor) and Cy5 (FRET acceptor). sp-FRET measurements were initiated using the de novo-designed transmembrane helix (AALALAA)3 (1TM), which was devoid of the sequence motifs that drive the dimer formation of helices10) (Fig. 5A). The peptide was confirmed to adopt stable transmembrane helical structures in lipid bilayers with various lipid compositions.10,26,27) 1TM helices preferentially formed antiparallel dimers due to electrostatic attraction between helix macrodipoles, which was enhanced in thicker membranes with their partial charges at helix termini embedded in a less polar environment26) (Fig. 5B). The membrane thickening effect of cholesterol was estimated to stabilize the helix macrodipole interaction by approx. −2 kJ mol−1.10)
(A) Amino acid sequences for the host (1TM), GXXXG, and 2TM. (B) Helix macrodipole interactions in lipid bilayer environments. Dielectric constants around helix termini (ε) decreased with increases in the thickness of the bilayer (from C14 to C22).
Introduction of the GXXXG motif, which has been shown to drive helix–helix associations,40–42) at the center of the sequence of 1TM (AALALAA-AGLALGA-AALALAA) (Fig. 5A, GXXXG) is useful to investigate the role of the motif in the cholesterol efect.11) Dimerization of the GXXXG helices in POPC membranes was stronger than that of the 1TM helices in both parallel and antiparallel association topologies.11)
Moreover, to examine the stability of transmembrane helix bundles, the (AALALAA)3-G5-(AALALAA)3 (2TM) peptide was designed, which comprised two 1TM helices connected to a flexible pentaglycine linker12) (Fig. 5A).
It is important to note that host helices did not have a specific interaction motif with cholesterol molecules, such as CRAC domains.43) The transmembrane GXXXG sequence was proposed to bind cholesterol.4) However, our FRET experiments with fluorescent sterol did not detect specific interactions between GXXXG helices and sterols.11) Therefore, these helices are considered to reflect the general/indirect effects of cholesterol on helix associations.
Figure 6 summarizes the thermodynamic stabilities (ΔGa) and kinetic stabilities (dimer lifetime) of the helix bundles of 1TM–1TM (antiparallel), 1TM–2TM, 2TM–2TM, and GXXXG–GXXXG (parallel and antiparallel) helices in the absence and presence of 30 mol% cholesterol in POPC bilayers at 25 °C. In all cases, except for GXXXG–GXXXG, membrane cholesterol enhanced the thermodynamic and kinetic stabilities of the helix associations. The 1TM–2TM dimer (3-helix bundle) was significantly stabilized by cholesterol. In the 1TM–1TM and 1TM–2TM associations, more tilted helices were observed in the presence of cholesterol (measured with Fourier transform (FT)IR polarized attenuated total reflection spectroscopy), which was consistent with the formation of hourglass-shaped helix bundles to compensate for high lateral pressure at the center of cholesterol-containing membranes.10,12)
X and Y axes indicate the association free energy and dimer lifetime at 25 °C, respectively. Open and closed symbols represent associations in POPC and POPC/cholesterol (7/3) bilayers, respectively. Circles: antiparallel 1TM–1TM; diamonds: 1TM–2TM (the topology was not specified); rectangles: 2TM–2TM (the topology was not specified); triangles: antiparallel GXXXG–GXXXG; inverted triangles: parallel GXXXG–GXXXG. Weak associations (1TM–1TM in POPC and GXXXG associations in POPC/cholesterol) were not detectable by the sp-FRET measurements (gray region).
The GXXXG motif frequently drives helix associations in membranes. Parallel and antiparallel dimers were consistently observed in POPC bilayers. However, these GXXXG dimers were destabilized in the presence of cholesterol (associations were not detectable by sp-FRET measurements).11) Two dimensional (2D)-IR and FTIR studies revealed that GXXXG dimers had small crossing angles (approx. 10°).11) Therefore, cholesterol-containing membranes destabilize these dimers by lateral pressure constraints.
The significant and complex effects of membrane cholesterol on the stability of transmembrane helix bundles, which were dependent on the association topology, number, and amino acid sequence of helices, were discussed in this review.
These studies revealed that small differences in the helix markedly affect the cholesterol dependence of interhelical interactions. Even in the ranges of the conditions examined, cholesterol-induced constraints (rigidity and lateral pressure profiles) to stabilize helix bundles were not unidirectional. In addition to the amino acid sequences of proteins, the effects of membrane properties controlled by lipids, cholesterol in this case, are important for understanding the folding, stability, and dynamic functions of transmembrane proteins in biomembranes.
This article was written based on studies performed in Prof. Katsumi Matsuzaki’s lab (Graduate School of Pharmaceutical Sciences, Kyoto University). I sincerely thank Prof. Matsuzaki and colleagues for their continuous support on research projects. I would also like to thank Prof. Martin Zanni (Univ. Wisconsin-Madison, U.S.A.), Prof. Shinya Oishi (Kyoto Pharmaceutical Univ.), Prof. Nobutaka Fujii (Kyoto Univ.), Prof. Hiroyuki Yasui (Kyoto Pharmaceutical Univ.), and Prof. Shiroh Futaki (Kyoto Univ.) for their support of research projects. This work was financially supported by JSPS KAKENHI (19K07013, 16K08194, 25460034, 21107514, 18790025, and 03J05377).
The author declares no conflict of interest.
