Cholesterol Oxidation Modulates the Formation of Liquid-Ordered Domains in Model Membranes

7-ketocholesterol (KChol) is one of the most cytotoxic oxysterols found in the plasma membrane, and increased levels of KChol are associated with numerous pathologies. It is thought to induce apoptosis via inactivation of the phosphatidylinositol 3-kinase/Akt signaling pathway — a pathway that depends on lipid-rafts as signaling platforms. By means of coarse-grained molecular dynamics simulations, we demonstrate that KChol disrupts the liquid-liquid phase separation seen in an equimolar mixture of (dipalmitoylphosphatidylcholine) DPPC, (dioleoylphosphatidylcholine) DOPC, and Cholesterol (Chol). This disruption occurs via two mechanisms: i) KChol adopts a wider range of orientations with the membrane, which disrupts the packing of neighboring lipids and ii) KChol has no preference for DPPC over DOPC, which is the main driving force for lateral demixing in DPPC/DOPC/Chol membranes. This provides a molecular description of the means by which KChol induces apoptosis, and illustrates that a single chemical substitution to cholesterol can have a profound impact on the lateral organization of lipid membranes. Graphical TOC Entry

Since Simons and Ikonen first described lipid rafts, 1 the existence, origin and nature of these structures in cellular membranes has been hotly debated. [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] However, there is now direct evidence of microdomains in live yeast cell organelles; 18,19 of nanodomains in live plant cell plasma membranes; 20 of functional membrane microdomains in live bacteria; 21,22 and of nanodomains in isolated mammalian cell plasma membranes. 23 The ubiquitous presence of lipid-raft-like structures across the domains of life means their biological significance is no longer in doubt. This is further emphasized by their suspected roles in many membrane processes: from membrane signaling 24 to membrane trafficking, 25 from membrane deformation 26 to membrane vesiculation, 27 and from sites for oligomerization of peptides 28 to sites for attachment of pathogens. 29 Given the biological importance of lipid rafts, the disruption of liquid-ordered domains has the potential to impact myriad biological pathways and processes. Elevated levels of ring-oxidised sterols -produced by the autoxidation of cholesterol 30 -are implicated in numerous pathologies, [31][32][33][34][35][36][37][38][39][40][41][42][43] and have been speculated to prevent liquid-ordered domain formation. 44-46 7-ketocholesterol (KChol; Figure 1A) is one of the most abundant and cytotoxic oxysterols, 44 and its presence in lipid rafts can induce cell death. 33 KChol causes apoptosis via inactivation of the phosphatidylinositol 3-kinase/Akt signaling pathway 47 -a pathway that depends on lipid rafts as signaling platforms. 24 Further, by excluding KChol from lipid rafts, cell death is avoided. 48 It is therefore possible that KChol induces apoptosis via disrupting the formation of liquid-ordered domains in the plasma membrane.
The concept of lipid rafts originated as an explanation for the dynamic clustering of cholesterol (Chol; Figure 1A) and sphingolipids in the plasma membrane, and the preferential sorting of certain proteins into these domains. 1 Since then, many different lipid mixtures have been found to be capable of nano-or micro-domain formation. 17,[49][50][51][52][53][54][55][56] Indeed, the plasma membrane is thought to consist of many different raft-like and non-raft-like regions of varying lipid composition. 6,10,53 These raft-like regions may arise through many different physical processes, 9,57,58 with different physical mechanisms dominating at different stages of domain formation. 59 Given the complexity of the plasma membrane, model membranes are typically employed for the study of domain formation. Membranes consisting of 1,2-dipalmitoyl-sn-3-phosphocholine (DPPC), 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) and Chol were the first phase-separating ternary mixture to have its phase boundaries fully mapped, 49 and has since become the canonical mixture for studying phase separation in lipid membranes.
While this mixture produces macroscopic phase separation, nanodomains behave surprisingly like genuine phases and so studying macroscopic phase separation may also inform us about nanodomains and lipid rafts. In this work, we report on the effect of cholesterol oxidation on domain formation studied should be equivalent to those at the thermodynamic limit. 65 We ran two replicas of each mixture for 20 µs at a temperature of 310 K. At this temperature, the equimolar mixture is at a critical point on the phase diagram generated by Carpenter et alii. 62 We have chosen a part of the phase diagram near the critical point as the plasma membrane is thought to be in such a region.  Here we see that KChol, an oxidation product of Chol, has significantly less affinity for DPPC over DOPC than Chol -and the result is a disruption of the macroscopic phase separation.
As a result of the phase separation, we see a large order gradient across the DPPC:DOPC:Chol membrane -the Chol-depleted region in Figure 1B is significantly more disordered than the Chol-enriched region. It has a larger area per lipid, smaller membrane thickness, and more disordered acyl tails than the Chol-enriched region ( Figure  Then, for each lipid species, we binned these thicknesses into nine states, which served as the emission states of the model. We used a Gaussian mixture model to initialize the parameters (µ, σ) of the hidden Gaussian distributions, before using the Baum-Welch algorithm to fit the model parameters based on the emission states and initial parameters. Finally, we used the Viterbi algorithm to decode the most likely time series of hidden states for each lipid.
The lateral distribution of ordered states can be seen in Figure 3.    Figure S4). Chol has a strong tendency to be oriented at around 10°(and 170°), whereas KChol has a broader distribution of orientations with a peak at around 15°(and 165°). KChol adopts a wider range of orientations in the membrane so that its hydrophilic ketone group can be exposed to the solvent. This increased orientational freedom of KChol will likely lead to an increased area per lipid. An implication of this is that KChol will disrupt the local packing of lipids in the L o phase. 44 Figure 4B). Domain registration in the Chol membrane, however, is also faster and more stable. It is not only the structure of the L o domains that changes upon oxidation, but also the dynamics of the molecules within the domains. In the DPPC:DOPC:Chol mixture, the lateral diffusion of L o -domain lipids is 1.5 times faster than those in the largest cluster of L d lipids (Table 2), which is in line with atomistic simulations and experimental measurements. 68 This difference is significantly reduced in the DPPC:DOPC:KChol membrane (Table 2)a result of the fact in this membrane we see nanodomain formation, with a smaller order gradient, rather than microdomain formation, with a larger order gradient. 77 We also find a substantial affect on interleaflet dynamics upon cholesterol oxidation. The  Figure 4C. For Chol, there is a barrier to flip-flop of around 5 kcal mol −1 . This is due to the unfavorable desolvation of the hydroxyl group during the flipflop process, which occurs as the sterol crosses through the hydrophobic core of the bilayer (−12Å< z < 12Å) and rotates to align with lipids in the apposing leaflet (65 • < θ TILT < 115 • ). The ring-oxidation of cholesterol into 7-ketocholesterol further increases this barrier by another 2 kcal mol −1 ( Figure 4C). This is because both the ketone and hydroxyl groups must be desolvated for flip-flop to occur. The result is that KChol is less likely move to the midplane and we thus see a reduced rate of flip-flop in the DPPC:DOPC:KChol mixture.
We have shown that the macroscopic phase separation seen in a DPPC:DOPC:Chol membrane is disrupted by the autoxidation of cholesterol into 7-ketocholesterol. In a DPPC:DOPC:KChol membrane, we instead see nanodomain formation that is more akin to that expected in the plasma membrane. 6,10,15,53 This disruption arises from the hydrophilicity of the ketone group of KChol, which has two effects on the domain formation. First, to allow for the hydration of the ketone group, KChol adopts a broader distribution of orientations in the membrane.
This disrupts the local packing of lipids, inducing disorder in L o regions. [44][45][46] Second, the reason Chol prefers to interact with DPPC over DOPC is because DPPC is better at shielding the hydrophobic rings of cholesterol from the surrounding solvent. The tetracyclic rings of KChol, however, are less hydrophobic due to the presence of the ketone group; KChol tends to expose this moiety to the solvent rather than seeking refuge in the hydrophobic core of the bilayer, and we thus find that KChol has less of a preference for DPPC over DOPC. This reduced preference for DPPC suppresses the lateral demixing of lipids, which in turn disrupts the liquid-liquid phase separation seen in the DPPC:DOPC:Chol mixture.
We also see that the hydrophilicity of the KChol ketone group suppresses translocation. The reduced rate of translocation has little effect on domain registration at in the equilibrated mixture studied here, but in more physiologically-relevant mixtures sterol flip-flop is required for interleaflet domain registration. 54 Chol preferentially mixes with sphingolipids over glycerophospholipids for the same rea-son it prefers DPPC over DOPC -sphingolipids are better at shielding cholesterol from the surrounding solvent. 78 We can therefore expect the increased hydrophilicity of KChol to diminish its affinity for sphingolipids compared to cholesterol. This would either disrupt the formation of Chol-sphingolipid nanodomains in biological membranes, or at least reduce the lateral order gradient as seen here. The reduced order gradient would have implications lipid-raft protein-sorting due to the hydrophobic mismatch between raft regions and their embedded proteins. Such implications include the disruption of cell-signaling pathways, via which KChol is known to induce apoptosis.