Effects of Transmembrane Î±-Helix Length and Concentration on

Article pubs.acs.org/JPCB

Effects of Transmembrane α‑Helix Length and Concentration on Phase Behavior in Four-Component Lipid Mixtures: A Molecular Dynamics Study David G. Ackerman and Gerald W. Feigenson* Department of Molecular Biology and Genetics, Cornell University, Ithaca, New York 14853, United States S Supporting Information *

ABSTRACT: We used coarse-grained molecular dynamics simulations to examine the effects of transmembrane α-helical WALP peptides on the behavior of four-component lipid mixtures. These mixtures contain a high-melting temperature (high-Tm) lipid, a nanodomain-inducing low-Tm lipid, a macrodomain-inducing low-Tm lipid and cholesterol to model the outer leaflet of cell plasma membranes. In a series of simulations, we incrementally replace the nanodomaininducing low-Tm lipid by the macrodomain-inducing low-Tm lipid and measure how lipid and phase properties are altered by the addition of WALPs of different length. Regardless of the ratio of the two low-Tm lipids, shorter WALPs increase domain size and all WALPs increase domain alignment between the two leaflets. These effects are smallest for the longest WALP tested, and increase with increasing WALP concentration. Thus, our simulations explain the experimental observation that WALPs induce macroscopic domains in otherwise nanodomain-forming lipid-only mixtures (unpublished). Since the cell plasma membrane contains a large fraction of transmembrane proteins, these findings link the behavior of lipid-only model membranes in vitro to phase behavior in vivo.

1. INTRODUCTION The cell plasma membrane (PM) is a bilayer comprised of hundreds of different lipid and protein species that give rise to distinct nanoscale environments sometimes termed “rafts”.1,2 Here we use “nanoscale” to mean not visible with an optical microscope, thus a size scale ∼5−200 nm. The properties and lipid compositions of the raft and nonraft environments suggest that they might be nanoscopic phase domains. Indeed, simplified lipid mixtures representing the PM can produce coexisting liquid-ordered (Lo) and liquid-disordered (Ld) phases, similar to the raft and nonraft PM environments, respectively.1,3 Modeling the complexity of the PM using these simplified lipid mixtures elucidates many of the fundamental features of phase separation that underlie raft formation. However, a key feature commonly omitted from many experimental model membranes is the presence of proteins. Since transmembrane protein domains make up 15−20% of the volume of the PM bilayer,4,5 a more complete understanding of phase behavior and raft formation requires inclusion of proteins in the well-established, but more simplistic, lipid-only model membranes. The simplest biologically relevant model membrane mixtures giving rise to coexisting Lo and Ld domains contain three lipid types found in the PM: a high-melting temperature (high-Tm) lipid, a low-Tm lipid and cholesterol (chol).6−9 Depending on the particular mixture, the Lo + Ld domains will either be nanoscopic or macroscopic.3 In both cases, the Ld phase is © 2016 American Chemical Society

characterized by low acyl chain order and is enriched in the low-Tm lipid, whereas the Lo phase is characterized by high acyl chain order, is enriched in the high-Tm lipid and requires chol to form.3,10 Similarly, rafts are high-order environments enriched in high-Tm lipids and sometimes enriched in chol compared to the rest of the membrane,1 indicating that the simplistic lipid-only membrane models capture the general nature of coexisting PM domains. In addition to lipids, proteins within the PM affect raft behavior. In cells, raft size seems to be on the order of tens of nanometers and thus below optical resolution,1 but cooled PM vesicle blebs that are separated from the proteins of the cytoskeleton can exhibit coexisting micron-scale fluid domains.11,12 These experimental findings are supported by theoretical models showing that coupling to cytoskeletal proteins can prevent large-scale domains while also stabilizing small-scale critical fluctuations above the critical point.13 Altering these fluctuations is one way that proteins can play an active role in changing domain size and morphology within the PM. But proteins also respond to the underlying phase behavior of the PM, with peripheral and integral membrane proteins preferentially partitioning into certain phases of blebs.11,12 The ability of proteins to alter and respond to Received: January 19, 2016 Revised: March 29, 2016 Published: April 15, 2016 4064

DOI: 10.1021/acs.jpcb.6b00611 J. Phys. Chem. B 2016, 120, 4064−4077

Article

The Journal of Physical Chemistry B

PC in CG), the nanodomain-forming low-Tm lipid PUPC (4:0,4:2-PC in CG), and cholesterol at a composition of DPPC/[DUPC+PUPC]/chol = 0.4/0.4/0.2 (Figure 1). We

phase behavior is also captured by model membrane experiments in which protein-induced cross-linking of gangliosides causes visible phase separation, which in turn alters the partitioning of transmembrane peptides.14,15 Here we use coarse-grained (CG) molecular dynamics (MD) simulations of quaternary lipid mixtures together with transmembrane peptides to elucidate the molecular underpinning of this twoway relationship between lipid phases and membrane proteins. By simplifying the representation of molecules, CG models allow access to the size-scales and time-scales necessary for studying membrane phase separation and protein-phase interactions.16−19 Simulations of phase-separated model membranes show that membrane-bound proteins such as H-Ras, and protein-bound lipids such as gangliosides, can exhibit preferential membrane phase partitioning20 and even interfacial partitioning that lowers the energy penalty of the phase interface.21 Interactions of the membrane anchors with the surrounding membrane in turn influence overall protein behavior, with the peripheral membrane proteins Hedgehog, H-Ras and N-Ras partitioning based on the nature of their cholesterol (Hedgehog) or prenyl or palmitoyl (H- or N-Ras) anchors.20 Simple α-helical peptides have also been useful at elucidating protein−membrane interactions as they model the ubiquitous transmembrane α-helical domains of integral membrane proteins.22 CG simulations of phase-separated bilayers with α-helices have revealed much about peptidephase interactions: tight lipid packing in the Lo phase causes αhelices to partition into the Ld phase;23 partitioning into the Ld phase decreases with increased helix length;24 fixed α-helices can recruit phase domains to their locations;24 interactions of helix extracellular domains can affect helix clustering and diffusion, in turn altering domain patterning;25 and high concentrations of helices can induce domain formation.26 Because domain size scale can be important, further control and evaluation of domain properties and peptide effects in phaseseparated bilayers requires expanding beyond the minimal three-component lipid mixtures used to achieve phase separation in CG simulations. This can be accomplished by adding just one more lipid component, as described below. Experiments with four-component lipid mixtures can make use of the influential role of the low-Tm lipid: by changing just the type of low-Tm lipid, the size scale of domains can go from nanoscopic (e.g., DSPC/POPC/chol27) to macroscopic (e.g., DSPC/DOPC/chol28).29−31 The domain size can then be controlled in four-component mixtures by fixing the overall fraction of high-Tm lipid/[macrodomain-forming low-Tm lipid + nanodomain-forming low-Tm lipid]/chol (e.g., DSPC/[DOPC +POPC]/chol), and incrementally replacing the nanoscopic low-Tm lipid with the macroscopic low-Tm lipid. As replacement increases, domains grow several orders of magnitude from nanoscopic to macroscopic.29 Specific properties of the phases can be measured throughout this replacement, with features such as Lo−Ld thickness mismatch found to be correlated with domain size.32 Since cells may similarly be able to use composition to control raft behavior and size, four-component systems are a next logical step for modeling protein-phase interactions that can occur in vivo. Here we report the effects of simple transmembrane α-helical WALP peptides on phase behavior in four-component lipid mixtures using CG MD. To model experimental fourcomponent mixtures, we use the same system described in our previous work:33 the high-Tm lipid DPPC (4:0,4:0-PC in CG), the macrodomain-forming low-Tm lipid DUPC (4:2,4:2-

Figure 1. Molecules used in this study. (A) Lipids, with unsaturated beads colored orange, and (B) WALPs used in the simulations. (C) Snapshot of bilayer at ρ = 0.8 with 2 mol % WALP-23, with DPPC (blue), DUPC (red), PUPC (green), chol (yellow), and WALP-23 (purple cylinders). Molecules and bilayer visualized with VMD version 1.9.1.

define a replacement ratio of PUPC by DUPC as ρ = [DUPC]/ [DUPC + PUPC] and simulate along a compositional trajectory (ρ-trajectory) of increasing ρ. Low ρ simulations have large fractions of PUPC and small domains, and high ρ simulations have large fractions of DUPC and large domains. As a function of ρ, we investigate how domain properties are affected by WALP length and concentration. Two main results are that WALPs thinner than the Lo phase, regardless of concentration, increase domain size and that all WALPs, including WALPs as thick as the Lo phase, increase domain alignment.

2. COMPUTATIONAL METHODS All simulations were performed at a lipid composition of DPPC/[DUPC + PUPC]/chol = 0.4/0.4/0.2. Whereas the DPPC/PUPC/chol system is properly characterized as having nonideal mixing rather than nanoscopic phase separation,26,34 it enables us to study changes in domain size and behavior from small clusters at low ρ to large phase-separated patches at high ρ. Particular WALPs were selected to span the range of phase thicknesses in the simulations. The thicknesses (measured as the distance between phosphate beads, see section 3.4) of a pure DUPC Ld phase, a pure PUPC Ld phase and a DPPC/ chol = 0.68/0.32 Lo phase are ∼3.53, ∼3.76, and ∼4.47 nm, respectively. We therefore used WALP-17, WALP-23, and WALP-29 which have thicknesses (measured as the distance between the first and last bead of the WALP) of ∼2.6, ∼ 3.6, and ∼4.5 nm, respectively. In previous work,33 we found that the steepest change in phase behavior for the peptide-free mixture DPPC/[DUPC 4065


Article

The Journal of Physical Chemistry B Table 1. Simulation Informationa no. of lipids no. of WALPs ρ run time (μs) a

no WALP

2 mol % WALP-17

2 mol % WALP-23

2 mol % WALP-29

0.4 mol % WALP-23

1 mol % WALP-23

4 mol % WALP-23

4608 0 0.4−0.8 25

4608 90 0.4−0.8 10

4608 90 0.4−0.8 10

4608 90 0.4−0.8 10

4608 18 0.6 10

4608 54 0.6 10

4608 180 0.6 10

Number of lipids, number of WALPs, ρ values, and run times for the seven different systems discussed in the text.

+PUPC]/chol = 0.4/0.4/0.2 occurred between ρ = 0.6 and ρ = 0.8. To capture this transition in our current work which makes use of the additional variables of peptide length and concentration, here we simulated only from ρ = 0.4 to ρ = 0.8 in increments of ρ ∼ 0.1. Four ρ-trajectories were simulated in this way: one ρ-trajectory with either 2 mol % WALP-17, 2 mol % WALP-23, or 2 mol % WALP-29, and one ρ-trajectory without WALP (Table 1). Three additional simulations were performed at ρ = 0.6 with WALP-23 concentrations of 0.4, 1, or 4 mol % (Table 1). All simulations contained 4,608 total lipids (2,304 per leaflet) and were fully solvated with ∼11 water beads per molecule. We note here that the 1 mol % WALP-23 simulation does not always follow the trends of the other simulations. We suspect that this may be a sampling issue and that future work with longer simulations will reveal that it should indeed follow the expected trends. 2.1. Molecular Parameters. All simulations were performed with Gromacs35 version 4.6 and the Martini 2.1 CG force field.36,37 The standard water and lipid parameters38 were used, in addition to the newest cholesterol model.39 All WALPs had the amino acid sequence AGAW(LA)nLWAGA, where n = 4, 7, and 10 for WALP-17, WALP23, and WALP-29 respectively. These WALPs have a single tryptophan near each end to simplify the dynamics.40,41 Atomistic structures of WALP were first built using PyMol,42 followed by conversion to the Martini 2.1 force field using martinize.py.43 We found these “single Trp WALPs” to strongly cluster (data not shown), yet we are interested in how isolated WALPs and small WALP clusters affect domain behavior. To prevent excessive WALP clustering we changed the amino acid AC1AC1 bead interaction from “intermediate” to “super-repulsive.” This modification is mild in only affecting WALP−WALP and internal WALP interactions without modifying lipid interactions, but it significantly inhibits large-scale clusters observed with the default MD parameters (data not shown). Our main observations, that WALPs can increase domain size and alignment, were seen both with and without the super-repulsive AC1−AC1 bead interactions. In agreement with published studies, we found that WALPs with two tryptophans at each end do not exhibit large aggregates in simulations.23 2.2. Bilayer Assembly. Initial template bilayers containing 512 lipids and proteins at the desired concentrations were built along the xy plane using in-house code. To ensure complete solvation, ∼11 water beads were added per molecule. During an initial 3 ns equilibration step at 295 K and 1 bar, the GL1 bead of the lipids, the R2 bead of cholesterols, and the two end beads of the WALPs were weakly position-restrained in the zdimension (force constant = 20 kJ mol−1 nm−2) to prevent excessive bilayer deformation. Position restraints were then removed and the system was further equilibrated for 3 ns at 295 K and 1 bar. The resultant bilayer was tiled 3 × 3 times, yielding the final bilayer sizes (Table 1).

Concentrations mentioned throughout the paper are the desired concentrations, but are only approximate due to the limited number of molecules. In the most extreme case, the desired 1 mol % WALP-23 concentration is actually 54/4,608 = 1.17 mol %. After bilayer assembly, the WALP-free systems were run for 25 μs (Table 1). The WALP-containing simulations were also initially run for 25 μs, but after finding that clustering occurred (see section 2.1), they were remade and rerun with the new super-repulsive interaction for 10 μs. The shorter run times were necessitated by limited computational resources, but were sufficient for equilibration and provided enough data for proper analysis, as discussed in section 2.5. 2.3. Simulation Parameters. Simulations were run in the NPT ensemble with 20 fs time steps. A temperature of 295 K was maintained using the V-rescale thermostat44 with a time constant of 1 ps. All molecule types were individually coupled to the temperature bath. The Berendsen semi-isotropic barostat45 was used to maintain a pressure of 1 bar in the xy and z dimensions, with a time constant of 4 ps. van der Waals interactions were shifted to zero from 0.9 to 1.2 nm. Electrostatic interactions were shifted to zero from 0 to 1.2 nm. The center of mass motion of the system was removed every ten time steps. Periodic boundary conditions were used in all three dimensions, and the LINCS35,46 algorithm was used to constrain bonds. 2.4. Phase Determination. To identify phases, we start by partitioning the molecules into leaflets. If a lipid center of mass is above (below) the local bilayer center of mass, then it is considered in the top (bottom) leaflet. If a WALP’s first and last beads are above (below) the local bilayer center of mass, then it is considered in the top (bottom) leaflet; if the first and last beads are in opposite leaflets, then the WALP is considered in both leaflets. Phases are then determined as in ref 33: First, the centers of mass of all molecules in a given leaflet are projected onto the xy plane. Next, a Voronoi tessellation is performed on the centers of mass. A molecule’s local environment is then defined to be all molecules that share a Voronoi edge with it. If the local environment is enriched in DPPC and cholesterol compared to the rest of the leaflet, the molecule of interest is considered Lo. Otherwise it is considered Ld. Large-scale phases are then determined based on connectivity of like-phase molecules, where two molecules are connected if they share a Voronoi edge. Clusters smaller than 10 molecules are considered part of the surrounding phase. Boundaries between phases are then the phase interface. This methodology was implemented in Matlab version 2014b. 2.5. Equilibration and Data Analysis. On the basis of equilibration of box area, phase interface length, and phase alignment (Figure S1), we determined that the simulations were sufficiently equilibrated by 5 μs. Therefore, we used only the last 5 μs for data analysis of the peptide-containing mixtures. For the peptide-free mixtures, which were run for 25 4066


Article


Figure 2. Domain size and alignment between leaflets change in the presence of WALP, shown for ρ = 0.6. (A−D) Voronoi tesselation for one leaflet of the bilayer with DPPC (blue), DUPC (red), PUPC (green), chol (yellow) and WALP-23 (purple) for (A) the peptide-free system, (B) 0.4 mol % WALP-23, (C) 1 mol % WALP-23, and (D) 4 mol % WALP-23. Phase boundaries demarcated by thick black lines. (E−H) Corresponding plots overlaying the bilayer leaflets, with Lo−Lo overlap in white, Ld−Ld overlap in black, and Lo−Ld overlap in gray, shown for (E) the peptidefree system, (F) 0.4 mol % WALP-23, (G) 1 mol % WALP-23, and (H) 4 mol % WALP-23.

μs, we allowed for an extended equilibration of 15 μs before data were acquired over the final 10 μs. To calculate standard errors, each simulation data set was split into 250 ns subsets, with each subset considered independent. To confirm that each subset is sufficiently independent, we examined the autocorrelations of two main parameters of interest: normalized phase interface length, and normalized alignment fraction (discussed below). The autocorrelations are plotted in Figure S2, which shows that the data become sufficiently uncorrelated by 250 ns, and so each 250 ns subset can be considered independent. Where applicable, each subset was further split into two for the two leaflets. Means and standard errors were calculated from these subsets. With the limited simulation size together with the somewhat arbitrary choice of some minimum number of molecules that can be called a phase, we do not attempt to distinguish between highly nonideal mixing and the start of phase separation. In the Supporting Information (section S2), we follow the methodology of Domański et al.26 to measure demixing. While we find that demixing does increase with ρ and in the presence of WALP (Figure S3), we do not distinguish between nonideal mixing and phase separation. Therefore, in this work, all simulations are termed phase-separating, and all phase properties (e.g., line tension) assume that the simulations are exhibiting stable coexisting phases.

aligned Ld phases, and gray to an Lo phase across from an Ld phase. Even in these one-frame snapshots, there is a clear increase in domain size and alignment as WALP concentration increases. We quantify these effects in Figure 3. In Figure 3A, we plot the average domain area, normalized by the box area A. Compared to the lipid-only mixtures, 2 mol % WALP-17 and 2 mol % WALP-23 increase domain size along the ρ-trajectory by, respectively, 5−44% and 5−68%. In contrast, 2 mol % WALP-29 decreases domain size at some ρ values and only increases the domain size by a maximum of 14% at ρ = 0.6. The largest effect is observed for 4 mol % WALP-23, with an increase in average domain area of 111% compared to the WALP-free ρ = 0.6 simulation. These findings are supported by other simulations which have shown WALPinduced domain size growth in the ternary mixture DPPC/ PUPC/chol.26 We also measured interface length as a function of ρ since it is less distorted by the sporadic fracturing and fusing of domains as compared to domain size. Figure 3B shows how the ρ value changes the interface length between coexisting phases (normalized by the box length L) for all simulations. Since the area fractions of each phase are ∼50 ± 5% for all simulations (data not shown), changes in interface length can be predominantly attributed to coalesced domains rather than changes in phase fractions. Consistent with the domain coalescence as DUPC fraction increases, all bilayers exhibit a decreasing interface length along the ρ-trajectory. Compared to the WALP-free ρ-trajectory, 2 mol % WALP-17 decreases the interface length by between 16% and 25.9%, and 2 mol % WALP-23 decreases the interface length by between 9.9% and 33%. A 2 mol % WALP-29 sample has the smallest effect on morphology compared to the other WALPs, decreasing the interface length by a maximum of 9% at low ρ and increasing it by up to 15% at high ρ. Of all systems studied, the largest effect occurs for 4 mol % WALP-23, which decreases the interface length at ρ = 0.6 by 43% compared to the corresponding WALP-free system. Lowering the concentration of WALP-23 generally reduces this effect, but a detectable decrease in interface length of 6% is still observed at 0.4 mol % WALP-23.

3. RESULTS In the following sections we describe WALP-induced changes on overall phase behavior (section 3.1), the behavior of the WALPs themselves (section 3.2), and the effect of WALPs on specific lipid and phase properties (sections 3.3 and 3.4). 3.1. Shorter WALPs Increase Domain Size. All WALPs Increase Domain Alignment. To visualize changes in domain morphology, snapshots are shown of the WALP-free and WALP-23 simulations for the representative case of ρ = 0.6 (Figure 2). Parts A−D of Figure 2 show snapshots of individual leaflets with phase boundaries in black. Parts E−H of Figure 2 show alignment of phase domains between leaflets, with white corresponding to aligned Lo phases, black corresponding to 4067


Article


fold increase at ρ = 0.7−0.8. Again we find that higher WALP concentrations have more pronounced effects, and that 4 mol % WALP-23 induces the largest alignment of any system at ρ = 0.6, increasing the alignment by a factor of 1.53 compared to the lipid only mixture. We compare the size-scales of domains and alignment using pair correlation functions. In Figure 4, we plot the interleaflet

Figure 4. Pair correlation functions, shown for the example case of ρ = 0.6, reveal the increase in size-scale and alignment of domains in the presence of WALP. Centers of mass of Lo phase molecules were used to calculate (A) intraleaflet pair correlation functions that describe the characteristic size of Lo domains and (B) interleaflet pair correlation functions that describe the characteristic length-scale of Lo domain alignment. Representative error bars shown at r ∼ 2.5 nm. In general, WALPs increase the size-scale and alignment of phases. Results colored as follows: peptide-free (black), 2 mol % WALP-17 (red), 2 mol % WALP-23 (green), 2 mol % WALP-29 (orange), 0.4 mol % WALP-23 (purple), 1 mol % WALP-23 (pink), and 4 mol % WALP-23 (blue).

Figure 3. Domain size and alignment between leaflets increase in the presence of WALP. (A) Area per domain, normalized by the box area A, increases along the ρ-trajectories and is larger in the presence of WALPs. (B) Interface length, normalized by the box length L, decreases along the ρ-trajectories and is generally smaller in the presence of WALPs. (C) Aligned area of like-phases, normalized by the box area A, increases along the ρ-trajectories and is larger in the presence of WALPs. The color scheme used here is used for other Figures, unless otherwise specified.

and intraleaflet pair correlation functions for the centers of mass of Lo phase molecules for the example case of ρ = 0.6. Even at concentrations as low as 0.4 mol % WALP-23, there is an increase in both size and alignment of domains compared to the WALP-free bilayer, as indicated by increased correlations for r < 10 nm. Increasing concentration beyond 0.4 mol % generally drives the formation of larger, more aligned domains with correlations increasing for r = 0−10 nm. A uniquely largescale effect is observed for the 1 mol % WALP simulation. While the three different 2 mol % systems all increase the sizescale of domains and alignment, as before we see that the effects of WALP-17 and WALP-23 are more significant over the range r = 0−10 nm compared to those of WALP-29. 3.2. WALP Behavior: Partitioning, Local Environment, Clustering, and Tilt. In Figure 5, the WALP concentration is shown as a function of distance from the phase interface for the representative case of ρ = 0.6, normalized by the maximum concentration in each simulation. WALPs are almost entirely found in the Ld phase, consistent with previous MD and experimental work by other researchers.23 Figure 5A also indicates that shorter WALPs prefer being farther from the interface, and Figure 5B suggests that WALP concentration increases near the interface as overall WALP concentration

In addition to their intraleaflet effects on domain coalescence, WALPs induce a pronounced increase of interleaflet domain alignment. In Figure 3C, the area fraction of aligned domains is plotted (i.e., the area fraction of white or black in Figure 2E− H) for all simulations. Given that phase area fractions are ∼50%, alignment fractions of ∼0, ∼ 0.5, and ∼1 correspond respectively to complete antialignment of domains, random alignment of domains, and complete alignment of domains. The WALP-free systems exhibit essentially random alignment for ρ ≤ 0.6, with alignment increasing more steeply beyond ρ = 0.6 to ∼80% aligned at ρ = 0.8. On the basis of previous lipidonly simulations with an older cholesterol model,33 we expect these trends to continue beyond the ρ values simulated here. In the current simulations, we find that the addition of WALPs tends to increase domain registration, most significantly at lower ρ values where peptide-free mixtures exhibit nonaligned domains. The 2 mol % WALP-17 and 2 mol % WALP-23 samples drive domains to become 1.1−1.35 times more aligned than the corresponding lipid-only mixtures. A 2 mol % WALP29 sample also causes increased alignment, but the effect is smaller compared to the other WALPs, with only a 1.01−1.02 4068


Article


interface. We also see that shorter WALPs are farther from the interface than longer WALPs, expanding on the simulation findings of Liang et al. where fixed WALPs were increasingly found in Lo phases if their hydrophobic length was increased.24 For each simulation, we measured the average distance of Ld WALPs to the interface divided by the average distance of Ld lipids to the phase interface (Figure 6B). Using this method, we see that Ld WALPs are found farther from the interface than an average Ld lipid. At each ρ, WALP-17 is farthest, being 29− 121% farther from the interface than an average Ld lipid. As WALP length increases, the disparity between WALP and lipid distances decreases. Figure 6B also reveals differences in distance ratios based on WALP concentration: WALP-23 at 0.4 mol % is 55% farther from the interface than an average Ld lipid, whereas at 4 mol %, WALP-23 is only 25.5% farther from the interface. This could be due to high concentrations of WALP making it difficult for all the WALPs to pack deep in the Ld phase. Another explanation of this behavior is that larger domains exist at higher WALP concentrations, and since large domains have a higher fraction of lipids far from the interface, this results in a decrease in the distance ratio. We can determine the area occupied by the WALPs from the Voronoi tessellations. For each leaflet, a WALP occupies ∼1− 1.3 times more area than its fraction in that leaflet. WALPs are predominantly found in the Ld phase, making up ∼2−2.6 times more Ld area than their mole fraction in that leaflet. For example, 2 mol % WALP-23 at ρ = 0.6 makes up 4 mol % of each leaflet, occupies ∼4.6% of the leaflet area, and comprises ∼8.8% of the Ld phase area. To study WALP behavior more closely, we look at the local WALP environment, defined to be all molecules that share a Voronoi edge with WALP. In Figure 7, the fraction of these

Figure 5. WALP length and concentration determine WALP preference for phase interface. WALP concentration as a function of distance from the phase interface, normalized by the peak concentration for each simulation, shown for the example case of ρ = 0.6 for (A) different WALP lengths and (B) different WALP concentrations. All molecules farther than 5.5 nm from the interface were included in the same bin, centered at 5.75 nm. Results colored as follows: 2 mol % WALP-17 (red), 2 mol % WALP-23 (green), 2 mol % WALP-29 (orange), 0.4 mol % WALP-23 (purple), 1 mol % WALP23 (pink), and 4 mol % WALP-23 (blue).

increases. The unique behavior of WALP-17, peaking in concentration at intermediate distances in the Ld phase, might be due to WALP-17 clustering as discussed later. Similar trends are observed for all ρ values (data not shown), and nonnormalized concentrations for ρ = 0.6 are shown in Figure S4. The average distance of Ld phase WALPs from the interface is shown in Figure 6A. As domains increase in size with increasing ρ, WALPs are located farther from the Lo/Ld

Figure 7. WALPs in the Ld phase are predominantly surrounded by DUPC, with longer WALPS having a higher fraction of contacts with Lo lipids. Fraction of Voronoi contacts between Ld WALPs and other molecules shown for 2 mol % WALP-17, 2 mol % WALP-23, and 2 mol % WALP-29 at ρ = 0.4, ρ = 0.6, and ρ = 0.8. Figure 6. Longer WALPs and higher concentrations drive WALPs closer to the interface. (A) Average distance between Ld WALPs and the phase interface shows that shorter WALPs are farther from the interface. (B) Average Ld WALP interface distance divided by average Ld lipid interface distance confirm that WALPs in the Ld phase are farther from the interface than an average lipid, but that this proportion decreases with increased WALP length and concentration. Results colored as follows: 2 mol % WALP-17 (red), 2 mol % WALP23 (green), 2 mol % WALP-29 (orange), 0.4 mol % WALP-23 (purple circles), 1 mol % WALP-23 (pink circles), and 4 mol % WALP-23 (blue circles).

Voronoi contacts between Ld phase WALPs and each molecule type is tabulated. Regardless of ρ value or WALP type, WALPs tend to predominantly neighbor DUPCs. This is true even at ρ = 0.4, where there is 60% more PUPC than DUPC but WALPs have ∼40% of their contacts with DUPC and only 15−27% of their contacts with PUPC. As WALP length increases and WALPs partition somewhat closer to the interface, the fraction of contacts between WALP and the Lo lipids DPPC and chol also increases, in agreement with previous simulations.24 For 4069


Article

The Journal of Physical Chemistry B WALP-29, 18−27% of all their contacts are with DPPC and chol for the ρ values shown. For all WALPs, there is also a significant fraction of WALP−WALP contacts, ranging from ∼6 to 36% of all contacts for the ρ values shown. This implies that the WALPs are clustering to some extent. The clustering of WALPs seen in Figure 7 is further quantified in Figure 8, where we plot the average WALP cluster

Figure 9. Longer WALPs tilt more than shorter WALPs. (A) Tilt angle distributions for different WALPs at ρ = 0.6. The high tilt angles for WALP-17 are due to WALPs aligning with the bilayer plane. (B) Average tilt angles for WALPs, omitting WALPs with tilts greater than 60°, as a function of ρ show that WALP-29 tilts more than WALP-17 or WALP-23. Results colored as follows: 2 mol % WALP-17 (red), 2 mol % WALP-23 (green), 2 mol % WALP-29 (orange), 0.4 mol % WALP-23 (purple, circle in B), 1 mol % WALP-23 (pink, circle in B), and 4 mol % WALP-23 (blue, circle in B). Circles overlap in part B.

Figure 8. Large-scale WALP clustering only occurs for high WALP concentrations or short WALPs. Average number of WALPs per cluster, where two WALPs are considered in the same cluster if two of their beads are within 0.7 nm of eachother. Clustering increases with WALP concentration, but 2 mol % WALP-17 exhibits the most clustering. Results colored as follows: 2 mol % WALP-17 (red), 2 mol % WALP-23 (green), 2 mol % WALP-29 (orange), 0.4 mol % WALP23 (purple circle), 1 mol % WALP-23 (pink circle), and 4 mol % WALP-23 (blue circle).

and WALP-23 tilts an intermediate amount regardless of WALP concentration. WALP-17 tilt decreases with ρ, which could be due to the WALP-17 ends being more able to anchor themselves near the bilayer-water interfaces in the thinner Ld phases that exist at high ρ (see section 3.4 for phase thicknesses). Conversely, the larger tilts that occur at high ρ for WALP-29, and to a smaller extent for WALP-23, are due to the thicker WALPs adjusting to the thinner phases. These larger tilts observed for the thicker WALPs are in agreement with previous studies showing that hydrophobic thickness mismatch between peptide and bilayer is correlated with peptide tilt.57 Other simulations48,58 have also shown a nonzero tilt for WALPs thinner than the surrounding bilayer, as we find for WALP-17. Tilt could also be affected by whether or not a WALP is in a cluster, though this is not a focus of the present study. 3.3. WALPs Tend to Increase Lipid Demixing. Although a small mole fraction of the bilayer, the WALPs can promote demixing and change the compositions of coexisting Lo and Ld phases. Parts A−D of Figure 10 show the mol % of DPPC, DUPC, PUPC, and chol in Lo and Ld phases. In order to compare to the WALP-free case, WALPs are not included in the concentration calculations. Along the WALP-free ρtrajectory, demixing increases between the Lo phase lipids (DPPC and chol) and the Ld phase lipids (DUPC and PUPC). In the presence of WALP, there is a further depletion of DPPC in the Ld phase, but a smaller change in the Lo phase. Similarly, WALPs tend to decrease cholesterol concentration in the Ld phase and increase cholesterol concentration in the Lo phase. An exception is the 2 mol % WALP-29 ρ-trajectory, which has the opposite behavior. For the low-Tm lipids, in most cases WALPs deplete DUPC from the Lo phase and enrich it in the Ld phase, whereas WALP-induced compositional changes for PUPC are less significant. For all lipids, demixing increases as WALP-23 concentration increases.

size. As in Schäfer et al.,23 we consider two WALPs to be in the same cluster if any of their beads are within 0.7 nm of each other. At 2 mol %, WALP-23 and WALP-29 clusters remain small, with an average size of ∼2−4 WALPs per cluster over the entire range of ρ. Circles show the size of clusters increasing with increasing WALP-23 concentration at fixed ρ = 0.6. Interestingly, 2 mol % WALP-17 simulations have the largest clusters at each ρ, and the cluster sizes decrease as ρ increases. We note that peptide clustering in response to bilayer phase and hydrophobic mismatch has been explored extensively.23,47−56 However, due to our implementation of a super-repulsive interaction to decrease clustering, we cannot conclude whether the large WALP-17 clusters are due to WALP-17 being the thinnest of all WALPs tested, or whether clustering is due to WALP-17 having fewer super-repulsive beads than the other WALPs. What Figure 8 conclusively shows is that we were generally successful at preventing the extremely large WALP clusters that were observed with the standard parametrization. In addition to clustering, another characteristic of the WALP molecules is their tilt. The tilts of WALP molecules with respect to the local bilayer normal are plotted in Figure 9A. WALP-17 and WALP-23 exhibit similar tilt distributions, but WALP-29 tilts more. WALP-17 also has a small probability increase for high tilt angles, corresponding to a small fraction of WALP-17s that orient perpendicular to the bilayer normal. Figure 9B shows the average tilt angle for all the WALP simulations, omitting WALPs perpendicular to the bilayer normal (here taken to be WALPs with tilts greater than 60°). At all ρ values, WALP-29 tilts the most, WALP-17 tilts the least, 4070


Article


3.4. WALPs Alter Lipid Order and Phase Thickness. The average order of a CG lipid is calculated using the equation Sz =

1 8

8

∑ n=1

⟨3 cos2 αn − 1⟩ 2

where α is the angle between a bond in the lipid acyl chain and the bilayer normal, and the sum is over the four bonds in each acyl chain.59 In Figure 11 the average order of each lipid in the

Figure 10. WALPs increase demixing. Mole percent of (A) DPPC, (B) chol, (C) PUPC, and (D) DUPC for the Lo phases (solid line or filled circles) and Ld phases (dashed line or empty circles) (E) Length of tieline connecting Lo and Ld compositions, in arbitrary units. Tieline length calculated based on pseudo 3-component phase diagram of DPPC/[DUPC + PUPC]/chol where the two low-Tm lipids are grouped together. Results colored as follows: peptide-free (black), 2 mol % WALP-17 (red), 2 mol % WALP-23 (green), 2 mol % WALP29 (orange), 0.4 mol % WALP-23 (purple circles), 1 mol % WALP-23 (pink circles), and 4 mol % WALP-23 (blue circles).

Figure 11. Lipid order is only perturbed near the interface. Average lipid order, Sz, for (A) DPPC, (B) PUPC,and (C) DUPC as a function of distance from the phase interface. Darkness increases with ρ, from ρ = 0.4 (light gray) to ρ = 0.8 (black). All molecules farther than 5.5 nm from the interface were included in the same bin, centered at 5.75 nm.

WALP-free bilayers is plotted as a function of distance to the Ld/Lo phase interface from ρ = 0.4 (gray) to ρ = 0.8 (black). As we have shown previously,33 the order increases across the interface from the Ld phase to the Lo over a span of only ∼3 nm. To measure the effects of WALP on lipid order, we normalized for the distance-dependent effects near the phase interface shown in Figure 11. In Figure 12, we plot the average order of Ld phase DUPCs for the WALP-containing mixtures at ρ = 0.8, divided by the corresponding WALP-free results. Using this ratio we find that WALP-17 is at one extreme, dramatically decreasing the order of neighboring lipids by ∼30% compared to the WALP-free mixtures. WALP-23 has a negligible effect, only slightly raising the bilayer order without a significant dependence on WALP-23 distance. At another extreme, WALP-29 increases the overall average order of neighboring lipids by ∼15%, and slightly raises the average order at farther distances. We find similar behavior for DPPC and PUPC (data not shown). Such WALP-induced local

We summarize these findings by measuring the lengths of tielines between coexisting phases, i.e., measuring the distance between compositions of the coexisting phases as they would appear on the pseudo 3-component phase diagram of DPPC/ [DUPC + PUPC]/chol where the two low-Tm lipids are grouped together. Larger compositional differences between coexisting phases give longer tielines. In Figure 10E, we plot the length of the tielines for all systems studied. In nearly every case, WALPs increase the compositional differences between coexisting Lo and Ld phases. The extent of demixing is largest for WALP-17 and WALP-23, increasing with WALP concentration. Our findings are in agreement with CG simulations of similar mixtures showing WALP-enhanced demixing that increases with WALP concentration.26 However, Fö rster resonance energy transfer experiments indicate that in similar mixtures, WALPs do not change coexisting phase compositions (unpublished), so the CG results may not be fully capturing all the subtleties of demixing. 4071


Article


Figure 12. WALPs at 2 mol % perturb nearby lipids, shown for the representative case of ρ = 0.8. Average Ld phase DUPC order ratio as a function of distance to the interface and distance to the nearest WALP for (A) WALP-17, (B) WALP-23, and (C) WALP-29. Ratio is taken by dividing DUPC order at a given interface distance by the corresponding DUPC order from the peptide-free simulation. All DUPCs farther than 4 nm from the interface were included in the same bin, centered at 4.5 nm.

Figure 13. WALPs change the thickness mismatch between phases. Thicknesses of (A) the Lo phase and (B) the Ld phase for all simulations. Thickness was measured as the average phosphatephosphate distance between pairs of “valid” PC lipids in opposed leaflets, as defined in the main text. (C) Thickness mismatch between coexisting phases differs for the different WALPs and different ρ values. Results colored as follows: peptide-free (black), 2 mol % WALP-17 (red), 2 mol % WALP-23 (green), 2 mol % WALP-29 (orange), 0.4 mol % WALP-23 (purple circles), 1 mol % WALP-23 (pink circles), and 4 mol % WALP-23 (blue circles). Circles overlap in parts A−C.

perturbations that die off within a couple of nanometers have also been observed in previous simulations56 and are predicted by theoretical models.54 As expected with changes in order, bilayer thickness is also altered in the presence of WALP. To measure thickness, we first pair each top leaflet PC lipid with the nearest bottom leaflet PC lipid. We consider the pair of lipids to be “valid” for this measurement if both lipids are in the same phase, and if they are both more than 2 nm from the phase interface, to avoid interfacial perturbations. Phase thickness is then defined to be the average phosphate-phosphate distance of all such valid lipid pairs in that phase. We plot the coexisting phase thickness, and difference in thickness between the phases, in Figure 13. For all systems, the Lo phase thickness is ∼4.26 nm, and since Lo phase lipid chains are always nearly fully extended, the thickness only slightly increases by ∼0.1 nm from ρ = 0.4 to ρ = 0.8. In contrast, differences in Ld thickness are much more pronounced. WALP-17-containing Ld phases are always the thinnest (thinner than ∼3.5 nm), WALP-29-containing Ld phases are always the thickest (thicker than ∼3.75 nm), and both WALP-23 and the WALP-free Ld phases are of moderate thickness (∼3.75 nm). Figure 13C shows the thickness mismatch between coexisting Lo + Ld phases. Thickness mismatch increases as WALP length decreases, with WALP-17-containing bilayers having the largest thickness mismatches, WALP-23-containing and WALP-free bilayers exhibiting similar thickness mismatches, and WALP-29-containing bilayers having the smallest thickness mismatches. At low ρ values, the difference in thickness between coexisting phases can vary dramatically. For

example, the thickness mismatch for the simulations with WALP-17 at ρ = 0.5 (0.89 nm) is more than 2.5 times the corresponding WALP-29 simulation thickness mismatch (0.33 nm). As ρ increases, the difference in thickness mismatch decreases. Similar peptide-induced order and thickness changes have been predicted theoretically,54,60,61 and measured in simulations23,48,53,56 and experiments,62,63 but here we show for the first time how thickness varies with peptide length and concentration along compositional trajectories.

4. DISCUSSION Our main findings from section 3 are that shorter WALPs induce an increase in phase domain size and all WALPS increase domain alignment. In sections 4.1 and 4.2 we discuss how the other results from section 3 help to explain these findings. In section 4.3, we discuss the implications of our results for experiments on similar systems, and for the life of a cell. 4.1. Shorter WALPs Increase Intraleaflet Line Tension. Line tension is the energy per unit length of the phase interface. It has been measured both experimentally,64−66 and in simulations18,26,67 using a variety of techniques. Typically, large stable domains are required for such measurements. However, many of our simulations exhibit small transient 4072


Article


higher line tension observed at higher ρ. In the presence of WALP, thickness mismatch changes. Compared to the peptidefree simulations, WALP-17 increases the mismatch by thinning the Ld phase, WALP-23 barely changes the thickness mismatch since it does not significantly perturb either phase, and WALP29 decreases the thickness mismatch by thickening the Ld phase. If line tension in the WALP-containing systems were solely determined by thickness mismatch of the coexisting phases, then compared to the peptide-free simulations we would expect a very large increase in line tension for WALP-17, a negligible change in line tension for WALP-23 and a decrease in line tension for WALP-29. We would also expect that different WALP-23 concentrations would not significantly affect line tension since they do not significantly affect thickness mismatch. This is not the case in our simulations. Instead, WALP-17 and WALP-23 dramatically increase the line tension compared to the WALP-free systems, with higher WALP-23 concentrations leading to increased line tensions. Furthermore, WALP-29 raises line tension at some ρ values and lowers it at others, even though the presence of WALP-29 always decreases the thickness mismatch between coexisting phases. On the basis of our results, thickness mismatch is not the dominant factor in determining line tension and domain size in systems with transmembrane peptides. 4.2. WALPs Increase Surface Tension and Anchor Domains in Opposed Leaflets. As reported in section 3.1, all WALPs at all concentrations increase domain alignment between opposed leaflets. The increased alignment in the presence of WALP is in part driven by the WALP-induced increase in both demixing and domain size, which increase surface tension between the contact areas of different phases in opposed leaflets. As we have shown previously,33 the increase of both DUPC fraction and demixing along the WALP-free ρtrajectory results in a higher fraction of unsaturated beads at the bilayer midplane in the Ld phase. These beads interact unfavorably with the saturated beads in the bilayer midplane in the Lo phase, driving like-phase domains to align. That domains are larger at higher ρ further drives alignment due to the increased energy penalty that accompanies larger misaligned domains.72−74 Shorter WALPs are thus able to increase interleaflet alignment by increasing both demixing and domain size. Another factor contributing to increased alignment is the interleaflet nature of the transmembrane WALPs. Since the WALPs span the bilayer and partition into the Ld phase, it is reasonable that they would increase domain alignment by “anchoring” the Ld domains in one leaflet to Ld domains in the other leaflet. This is consistent with our finding that all WALPs increase alignment and that alignment increases with WALP concentration, since more WALPs mean more interleaflet anchors. Not surprisingly, simulations have shown that this effect can be lessened by palmitoylating one end of α-helical transmembrane domains, which allows clustering of Ld lipids to the nonpalmitoylated end and promotes clustering of Lo lipids to the palmitoylated end.20,25 Therefore, the ability of the WALP to partition into the Ld phase in both leaflets helps to drive alignment of domains. Alignment might also be affected by domain fluctuations. Lo + Ld mixtures with higher line tension have fewer, more stable domains than mixtures with lower line tension. Lower line tension, as with WALP-29, might then contribute to decreased alignment due to large, uncorrelated thermal domain fluctuations in the two leaflets.

domains that are not easily characterized by existing methods. Here we use interface length as a surrogate for line tension in our simulation analysis. An increase in line tension drives domains to coalesce, reducing the total length of interface. Since the area fractions of the phases remain at ∼50%, changes in the normalized area per domain shown in Figure 3A and normalized interface length in Figure 3B are directly correlated to changes in line tension. We find a pronounced increase in domain size and decrease in interface length in the presence of WALP-17 and WALP-23, whereas trends due to WALP-29 are less clear. We therefore conclude that the two shorter WALPs, regardless of concentration, raise Lo/Ld line tension compared to corresponding WALP-free simulations and so decrease the perimeter-to-area ratio of the domains. Previous CG simulations had shown higher line tensions in systems with WALP23,26 and our results extend this research to a wider range of WALPs, over a wider range of lipid compositions and line tensions. The observed changes in line tension are connected to WALP partitioning. Even at low concentrations, molecules can alter phase domain size depending on how they partition, and in turn, affect line tension. Molecules that partition to the interface inherently lower the line tension,21,67,68 whereas those that partition away from the interface inherently increase line tension. In Figures 5 and 6, we show that not only do WALPs partition preferentially into the Ld phase, in agreement with other work,23 but also they avoid the interface, tending to be deep within the Ld phase. On the basis of partitioning, WALPs must then raise line tension and thereby drive the formation of larger domains. We found that among the three peptide lengths we studied, WALP-29 tended to be closest to the interface, meaning that its partitioning would be expected to raise line tension the least. While this simple partition-driven line tension reasoning is in agreement with our interface length and domain size results, which showed that the effect of WALP-29 on line tension is small, other effects must be accounted for to fully explain the behavior of the different WALP types and concentrations. Another contribution to line tension in the simulations comes from the WALP-induced demixing of the coexisting phases. In Figure 10 we showed that the addition of WALP-17 and WALP-23 to the simulations increases demixing between the coexisting phases, predominantly by depleting DPPC and cholesterol from the Ld phase and enriching it in DUPC. Such an increase in compositional differences between coexisting phases raises the line tension, as has been measured experimentally for lipid-only systems.65 This is consistent with our simulation findings of increased demixing and increased line tension for WALP-17 and WALP-23 mixtures compared to WALP-29 and the lipid-only mixtures. Other simulations have even shown that high WALP concentrations increase demixing to the point where domains are stabilized in otherwise mixed systems.26 While experiments do not show a change in mixing in the presence of WALPs, demixing seems to play a role in the simulations. In addition to partitioning and demixing, one of the key factors often identified as raising line tension is the thickness mismatch between coexisting phases.26,69−71 Thickness mismatch forces the phases to adjust at the interface, which is energetically more costly for larger mismatches. In the WALPfree simulations studied here, there is always a thickness mismatch between the phases (Figure 13). This thickness mismatch increases along the ρ-trajectory, contributing to the 4073


Article


stabilization and registration in asymmetric bilayers, and describe our preliminary work in the Supporting Information (Section S4). The ability of transmembrane peptide domains to affect raft size and registration in vivo could be an essential mechanism by which cells alter localization and interaction of molecules in the two leaflets. The alignment of phase domains in the two PM leaflets might be part of a mechanism of transmembrane signaling.

Interestingly, some phase properties are actually proposed to favor antialignment of domains. For instance, thickness mismatch between coexisting phases is thought to inhibit alignment as a way to minimize the exposure of hydrocarbon to water at the phase interface.74−76 However, we find that compared to the WALP-free mixtures, there is a significant increase in alignment for all the 2 mol % WALP-17 simulations which have the largest thickness mismatches. Evidently, thickness mismatch is not a decisive factor in alignment for these simulations, possibly because the mismatches are not large enough to overcome other factors that drive alignment. 4.3. Implication of WALP Effects for Experiments and Cells. Experiments making use of ρ-trajectories enable finding the compositional crossover region between nanodomains and macrodomains.29,30 The transition can be modeled as a competition between line tension, which favors large domains, and some competing interaction(s) that favors broken-up domains.77,78 When line tension exceeds this competing interaction, nanodomains give way to macrodomains. Line tension has been experimentally measured along ρ-trajectories,79 but data are limited to the macroscopic regime. Experimental measurements show that macroscopic domains first form along the ρ-trajectory when line tension reaches a value of ∼0.3−0.4 pN, regardless of the lipid mixture (unpublished). On the basis of our findings, we predict that WALPs in vitro raise line tension and increase domain size, and so would shift the nano to macro transition to lower ρ values compared to the corresponding lipid-only mixtures. This effect should be least pronounced for WALPs thicker than the bilayer, and should increase with increasing WALP concentration. That the nano to macro transition shifts to lower ρ as WALP concentration increases is indeed supported by experimental findings (unpublished). Our results also suggest that WALPs in vitro should increase the register of domains. This property is harder to measure experimentally because macroscopic lipid bilayer domains in symmetric bilayers are always observed to be in register within the resolution of optical microscopy, likely due to a high surface tension between the leaflets.80 However, antiregistration has been observed in simulations exhibiting a large thickness mismatch between coexisting phases.75 If observed experimentally, such a system would be useful to test for WALP-induced alignment. Additionally, if the recent experimental work of Blosser et al. to measure the coupling parameter between phase separated leaflets81 could be extended to bilayers with transmembrane peptides, then the increased coupling due to WALP could be directly measured. Our findings in model membranes suggest that transmembrane α-helical peptides in the cell PM can stabilize rafts and interleaflet alignment. However, the cell PM differs from most model membranes since it has different lipid compositions in its two leaflets,82 each with distinct properties. For instance, symmetric membranes modeling the outer leaflet composition (enriched in sphingomyelins and PCs) can phase separate, but those modeling inner leaflet compositions (enriched in phosphatidylethanolamines and phosphatidylserines) do not.83 The effects of asymmetry have been studied with MD75,84,85 and experiments,86,87 in some cases showing that a phase-separated leaflet can induce phase separation in the otherwise uniform apposing leaflet.85−87 Though these studies were in lipid-only systems, it is possible that transmembrane proteins can further enhance interleaflet interactions.88 We have begun to measure the effect of WALP on domain

5. CONCLUSIONS

■

1 Transmembrane proteins make up a significant fraction of cell plasma membrane volume. Model transmembrane α-helical WALP peptides increase both bilayer phase domain size and alignment. 2 Regarding domain size: (a) phase thickness mismatch plays a smaller role in determining domain size compared to WALP partitioning and WALP-induced demixing; (b) WALPs that are shorter than the Ld phase increase domain size the most; (c) increased WALP concentration increases domain size; and (d) domain size increases for all ρ values for WALP-17 and WALP-23. 3 Regarding domain alignment: (a) alignment increases in WALP-containing systems by a combination of increased demixing, domain growth, and transmembrane anchoring of Ld phases; (b) WALPs longer than the Ld phase increase alignment the least; (c) increased concentration of WALP increases alignment; and (d) increase in alignment occurs for all WALPs at all ρ. 4 Different length WALPs significantly differ in their effects on lipid order and Ld phase thickness. Thus, raft order, size, and interleaflet alignment within a cell plasma membrane are likely affected by the presence and properties of transmembrane proteins, which may in turn affect the partitioning and functionality of other membrane proteins.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcb.6b00611. Figures showing equilibration of box size, autocorrelations of interface length and domain alignment, demixing parameter measurements, and WALP concentrations as a function of distance to the interface in addition to demixing calculations and description of preliminary asymmetry studies (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*(G.W.F.) E-mail: [email protected]. Telephone: (607) 2554744. Address: 201 Biotechnology Building, Cornell University, Ithaca, NY 14853. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by grants from the U.S. National Science Foundation (MCB-1410926) and the U.S. National Institutes of Health (GM105684) to G.W.F. D.G.A. was funded by the U.S. National Science Foundation Graduate Research Fellowship (DGE-114153) and National Institutes of Health 4074


Article


(19) Bennett, W. F. D.; Tieleman, D. P. Computer Simulations of Lipid Membrane Domains. Biochim. Biophys. Acta, Biomembr. 2013, 1828, 1765−1776. (20) De Jong, D. H.; Lopez, C. A.; Marrink, S. J. Molecular View on Protein Sorting into Liquid-Ordered Membrane Domains Mediated by Gangliosides and Lipid Anchors. Faraday Discuss. 2013, 161, 347−363. (21) Janosi, L.; Li, Z.; Hancock, J. F.; Gorfe, A. A. Organization, Dynamics, and Segregation of Ras Nanoclusters in Membrane Domains. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 8097−8102. (22) Killian, J. A. Synthetic Peptides as Models for Intrinsic Membrane Proteins. FEBS Lett. 2003, 555, 134−138. (23) Schäfer, L. V.; de Jong, D. H.; Holt, A.; Rzepiela, A. J.; de Vries, A. H.; Poolman, B.; Killian, J. A.; Marrink, S. J. Lipid Packing Drives the Segregation of Transmembrane Helices into Disordered Lipid Domains in Model Membranes. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 1343−1348. (24) Liang, Q.; Wu, Q.-Y.; Wang, Z.-Y. Effect of Hydrophobic Mismatch on Domain Formation and Peptide Sorting in the Multicomponent Lipid Bilayers in the Presence of Immobilized Peptides. J. Chem. Phys. 2014, 141, 074702. (25) Parton, D. L.; Tek, A.; Baaden, M.; Sansom, M. S. P. Formation of Raft-Like Assemblies within Clusters of Influenza Hemagglutinin Observed by MD Simulations. PLoS Comput. Biol. 2013, 9, e1003034. (26) Domański, J.; Marrink, S. J.; Schäfer, L. V. Transmembrane Helices Can Induce Domain Formation in Crowded Model Membranes. Biochim. Biophys. Acta, Biomembr. 2012, 1818, 984−994. (27) Heberle, F. A.; Wu, J.; Goh, S. L.; Petruzielo, R. S.; Feigenson, G. W. Comparison of Three Ternary Lipid Bilayer Mixtures: FRET and ESR Reveal Nanodomains. Biophys. J. 2010, 99, 3309−3318. (28) Zhao, J.; Wu, J.; Heberle, F. A.; Mills, T. T.; Klawitter, P.; Huang, G.; Costanza, G.; Feigenson, G. W. Phase Studies of Model Biomembranes: Complex Behavior of DSPC/DOPC/cholesterol. Biochim. Biophys. Acta, Biomembr. 2007, 1768, 2764−2776. (29) Konyakhina, T. M.; Goh, S. L.; Amazon, J.; Heberle, F. A.; Wu, J.; Feigenson, G. W. Control of a Nanoscopic-to-Macroscopic Transition: Modulated Phases in Four-Component DSPC/DOPC/ POPC/Chol Giant Unilamellar Vesicles. Biophys. J. 2011, 101, L8− L10. (30) Goh, S. L.; Amazon, J. J.; Feigenson, G. W. Toward a Better Raft Model: Modulated Phases in the Four-Component Bilayer, DSPC/ DOPC/POPC/CHOL. Biophys. J. 2013, 104, 853−862. (31) Konyakhina, T. M.; Wu, J.; Mastroianni, J. D.; Heberle, F. A.; Feigenson, G. W. Phase Diagram of a 4-Component Lipid Mixture: DSPC/DOPC/POPC/chol. Biochim. Biophys. Acta, Biomembr. 2013, 1828, 2204−2214. (32) Heberle, F. A.; Petruzielo, R. S.; Pan, J.; Drazba, P.; Kučerka, N.; Standaert, R. F.; Feigenson, G. W.; Katsaras, J. Bilayer Thickness Mismatch Controls Domain Size in Model Membranes. J. Am. Chem. Soc. 2013, 135, 6853−6859. (33) Ackerman, D. G.; Feigenson, G. W. Multiscale Modeling of Four-Component Lipid Mixtures: Domain Composition, Size, Alignment, and Properties of the Phase Interface. J. Phys. Chem. B 2015, 119, 4240−4250. (34) Rosetti, C.; Pastorino, C. Comparison of Ternary Bilayer Mixtures with Asymmetric or Symmetric Unsaturated Phosphatidylcholine Lipids by Coarse Grained Molecular Dynamics Simulations. J. Phys. Chem. B 2012, 116, 3525−3537. (35) Hess, B.; Kutzner, C.; van Der Spoel, D.; Lindahl, E. GROMACS 4: Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simulation. J. Chem. Theory Comput. 2008, 4, 435− 447. (36) Marrink, S. J.; Risselada, H. J.; Yefimov, S.; Tieleman, D. P.; de Vries, A. H. The MARTINI Force Field: Coarse Grained Model for Biomolecular Simulations. J. Phys. Chem. B 2007, 111, 7812−7824. (37) Monticelli, L.; Kandasamy, S. K.; Periole, X.; Larson, R. G.; Tieleman, D. P.; Marrink, S.-J. The MARTINI Coarse-Grained Force Field: Extension to Proteins. J. Chem. Theory Comput. 2008, 4, 819− 834.

Training Grant 1-T32-GM08267. This work used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation Grant No. ACI-1053575. We thank Lars V. Schäfer and Siewert J. Marrink for providing the reference simulation parameterization and data that allowed us to determine why WALPs cluster. We also thank Roger E. Koeppe II and Denise Greathouse for useful discussions about the WALP peptides.

■

REFERENCES

(1) Lingwood, D.; Simons, K. Science 2010, 327, 46−50. (2) Simons, K.; Sampaio, J. L. Membrane Organization and Lipid Rafts. Cold Spring Harbor Perspect. Biol. 2011, 3, a004697. (3) Feigenson, G. W. Phase Diagrams and Lipid Domains in Multicomponent Lipid Bilayer Mixtures. Biochim. Biophys. Acta, Biomembr. 2009, 1788, 47−52. (4) Sheetz, M. P. Glycoprotein Motility and Dynamic Domains in Fluid Plasma Membranes. Annu. Rev. Biophys. Biomol. Struct. 1993, 22, 417−431. (5) Golan, D. E.; Alecio, M. R.; Veatch, W. R.; Rando, R. R. Lateral Mobility of Phospholipid and Cholesterol in the Human Erythrocyte Membrane: Effects of Protein-Lipid Interactions. Biochemistry 1984, 23, 332−339. (6) Dietrich, C.; Bagatolli, L. A.; Volovyk, Z. N.; Thompson, N. L.; Levi, M.; Jacobson, K.; Gratton, E. Lipid Rafts Reconstituted in Model Membranes. Biophys. J. 2001, 80, 1417−1428. (7) Samsonov, A. V.; Mihalyov, I.; Cohen, F. S. C. Characterization of Cholesterol-Sphingomyelin Domains and Their Dynamics in Bilayer Membranes. Biophys. J. 2001, 81, 1486−1500. (8) Veatch, S. L.; Keller, S. L. Separation of Liquid Phases in Giant Vesicles of Ternary Mixtures of Phospholipids and Cholesterol. Biophys. J. 2003, 85, 3074−3083. (9) Veatch, S. L.; Keller, S. L. Organization in Lipid Membranes Containing Cholesterol. Phys. Rev. Lett. 2002, 89, 268101. (10) Marsh, D. Cholesterol-Induced Fluid Membrane Domains: A Compendium of Lipid-Raft Ternary Phase Diagrams. Biochim. Biophys. Acta, Biomembr. 2009, 1788, 2114−2123. (11) Baumgart, T.; Hammond, A. T.; Sengupta, P.; Hess, S. T.; Holowka, D. A.; Baird, B. A.; Webb, W. W. Large-Scale Fluid/fluid Phase Separation of Proteins and Lipids in Giant Plasma Membrane Vesicles. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 3165−3170. (12) Sengupta, P.; Hammond, A.; Holowka, D.; Baird, B. Structural Determinants for Partitioning of Lipids and Proteins between Coexisting Fluid Phases in Giant Plasma Membrane Vesicles. Biochim. Biophys. Acta, Biomembr. 2008, 1778, 20−32. (13) Machta, B. B.; Papanikolaou, S.; Sethna, J. P.; Veatch, S. L. Minimal Model of Plasma Membrane Heterogeneity Requires Coupling Cortical Actin to Criticality. Biophys. J. 2011, 100, 1668− 1677. (14) Hammond, A. T.; Heberle, F. A.; Baumgart, T.; Holowka, D.; Baird, B.; Feigenson, G. W. Crosslinking a Lipid Raft Component Triggers Liquid Ordered-Liquid Disordered Phase Separation in Model Plasma Membranes. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 6320−6325. (15) Lingwood, D.; Ries, J.; Schwille, P.; Simons, K. Plasma Membranes Are Poised for Activation of Raft Phase Coalescence at Physiological Temperature. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 10005−10010. (16) Meinhardt, S.; Vink, R. L. C.; Schmid, F. Monolayer Curvature Stabilizes Nanoscale Raft Domains in Mixed Lipid Bilayers. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 4476−4481. (17) Stevens, M. J. Complementary Matching in Domain Formation within Lipid Bilayers. J. Am. Chem. Soc. 2005, 127, 15330−15331. (18) Risselada, H. J.; Marrink, S. J. The Molecular Face of Lipid Rafts in Model Membranes. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 17367− 17372. 4075


Article


(59) Baoukina, S.; Mendez-Villuendas, E.; Tieleman, D. P. Molecular View of Phase Coexistence in Lipid Monolayers. J. Am. Chem. Soc. 2012, 134, 17543−17553. (60) Mouritsen, O. G.; Bloom, M. Mattress Model of Lipid-Protein Interactions in Membranes. Biophys. J. 1984, 46, 141−153. (61) Dan, N.; Pincus, P.; Safran, S. A. Membrane-Induced Interactions between Inclusions. Langmuir 1993, 9, 2768−2771. (62) De Planque, M. R. R.; Greathouse, D. V.; Koeppe, R. E.; Schäfer, H.; Marsh, D.; Killian, J. A. Influence of Lipid/peptide Hydrophobic Mismatch on the Thickness of Diacylphosphatidylcholine Bilayers. A 2H NMR and ESR Study Using Designed Transmembrane A-Helical Peptides and Gramicidin A. Biochemistry 1998, 37, 9333−9345. (63) Nezil, F. A.; Bloom, M. Combined Influence of Cholesterol and Synthetic Amphiphillic Peptides upon Bilayer Thickness in Model Membranes. Biophys. J. 1992, 61, 1176−1183. (64) Esposito, C.; Tian, A.; Melamed, S.; Johnson, C.; Tee, S.-Y.; Baumgart, T. Flicker Spectroscopy of Thermal Lipid Bilayer Domain Boundary Fluctuations. Biophys. J. 2007, 93, 3169−3181. (65) Tian, A.; Johnson, C.; Wang, W.; Baumgart, T. Line Tension at Fluid Membrane Domain Boundaries Measured by Micropipette Aspiration. Phys. Rev. Lett. 2007, 98, 208102. (66) Baumgart, T.; Hess, S. T.; Webb, W. W. Imaging Coexisting Fluid Domains in Biomembrane Models Coupling Curvature and Line Tension. Nature 2003, 425, 821−824. (67) Schäfer, L. V.; Marrink, S. J. Partitioning of Lipids at Domain Boundaries in Model Membranes. Biophys. J. 2010, 99, L91−L93. (68) Palmieri, B.; Yamamoto, T.; Brewster, R. C.; Safran, S. A. Line Active Molecules Promote Inhomogeneous Structures in Membranes: Theory, Simulations and Experiments. Adv. Colloid Interface Sci. 2014, 208, 58−65. (69) García-Sáez, A. J.; Chiantia, S.; Schwille, P. Effect of Line Tension on the Lateral Organization of Lipid Membranes. J. Biol. Chem. 2007, 282, 33537−33544. (70) Kuzmin, P. I.; Akimov, S. A.; Chizmadzhev, Y. A.; Zimmerberg, J.; Cohen, F. S. Line Tension and Interaction Energies of Membrane Rafts Calculated from Lipid Splay and Tilt. Biophys. J. 2005, 88, 1120− 1133. (71) Muddana, H. S.; Chiang, H. H.; Butler, P. J. Tuning Membrane Phase Separation Using Nonlipid Amphiphiles. Biophys. J. 2012, 102, 489−497. (72) Hakobyan, D.; Heuer, A. Key Molecular Requirements for Raft Formation in Lipid/cholesterol Membranes. PLoS One 2014, 9, e87369. (73) Pantano, D. A.; Moore, P. B.; Klein, M. L.; Discher, D. E. Raft Registration across Bilayers in a Molecularly Detailed Model. Soft Matter 2011, 7, 8182−8191. (74) May, S. Trans-Monolayer Coupling of Fluid Domains in Lipid Bilayers. Soft Matter 2009, 5, 3148−3156. (75) Perlmutter, J. D.; Sachs, J. N. Interleaflet Interaction and Asymmetry in Phase Separated Lipid Bilayers: Molecular Dynamics Simulations. J. Am. Chem. Soc. 2011, 133, 6563−6577. (76) Williamson, J. J.; Olmsted, P. D. Registered and Antiregistered Phase Separation of Mixed Amphiphilic Bilayers. Biophys. J. 2015, 108, 1963−1976. (77) Amazon, J. J.; Goh, S. L.; Feigenson, G. W. Competition between Line Tension and Curvature Stabilizes Modulated Phase Patterns on the Surface of Giant Unilamellar Vesicles: A Simulation Study. Phys. Rev. E 2013, 87, 022708. (78) Amazon, J. J.; Feigenson, G. W. Lattice Simulations of Phase Morphology on Lipid Bilayers: Renormalization, Membrane Shape, and Electrostatic Dipole Interactions. Phys. Rev. E 2014, 89, 022702. (79) Hassan-Zadeh, E.; Baykal-Caglar, E.; Alwarawrah, M.; Huang, J. Complex Roles of Hybrid Lipids in the Composition, Order, and Size of Lipid Membrane Domains. Langmuir 2014, 30, 1361−1369. (80) Collins, M. D. Interleaflet Coupling Mechanisms in Bilayers of Lipids and Cholesterol. Biophys. J. 2008, 94, L32−L34. (81) Blosser, M. C.; Honerkamp-Smith, A. R.; Han, T.; Haataja, M.; Keller, S. L. Transbilayer Colocalization of Lipid Domains Explained

(38) Marrink, S. J.; de Vries, A. H.; Mark, A. E. Coarse Grained Model for Semiquantitative Lipid Simulations. J. Phys. Chem. B 2004, 108, 750−760. (39) Melo, M. N.; Ingólfsson, H. I.; Marrink, S. J. Parameters for Martini Sterols and Hopanoids Based on a Virtual-Site Description. J. Chem. Phys. 2015, 143, 243152. (40) Sparks, K. A.; Gleason, N. J.; Gist, R.; Langston, R.; Greathouse, D. V.; Koeppe, R. E., II Comparisons of Interfacial Phe, Tyr, and Trp Residues as Determinants of Orientation and Dynamics for GWALP Transmembrane Peptides. Biochemistry 2014, 53, 3637−3645. (41) Gleason, N. J.; Greathouse, D. V.; Grant, C. V.; Opella, S. J.; Koeppe, R. E., II Single Tryptophan and Tyrosine Comparisons in the N-Terminal and C-Terminal Interface Regions of Transmembrane GWALP Peptides. J. Phys. Chem. B 2013, 117, 13786−13794. (42) The PyMOL Molecular Graphics System, Version 1.7.4; Schrödinger, LLC. (43) De Jong, D. H.; Singh, G.; Bennett, W. F. D.; Arnarez, C.; Wassenaar, T. A.; Schäfer, L. V.; Periole, X.; Tieleman, D. P.; Marrink, S. J. Improved Parameters for the Martini Coarse-Grained Protein Force Field. J. Chem. Theory Comput. 2013, 9, 687−697. (44) Bussi, G.; Donadio, D.; Parrinello, M. Canonical Sampling through Velocity Rescaling. J. Chem. Phys. 2007, 126, 014101. (45) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R. Molecular Dynamics with Coupling to an External Bath. J. Chem. Phys. 1984, 81, 3684−3690. (46) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. LINCS: A Linear Constraint Solver for Molecular Simulations. J. Comput. Chem. 1997, 18, 1463−1472. (47) Parton, D. L.; Klingelhoefer, J. W.; Sansom, M. S. P. Aggregation of Model Membrane Proteins, Modulated by Hydrophobic Mismatch, Membrane Curvature, and Protein Class. Biophys. J. 2011, 101, 691−699. (48) Kaiser, H.-J.; Orlowski, A.; Róg, T.; Nyholm, T. K. M.; Chai, W.; Feizi, T.; Lingwood, D.; Vattulainen, I.; Simons, K. Lateral Sorting in Model Membranes by Cholesterol-Mediated Hydrophobic Matching. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 16628−16633. (49) Schmidt, U.; Weiss, M. Hydrophobic Mismatch-Induced Clustering as a Primer for Protein Sorting in the Secretory Pathway. Biophys. Chem. 2010, 151, 34−38. (50) De Meyer, F. J.-M.; Venturoli, M.; Smit, B. Molecular Simulations of Lipid-Mediated Protein-Protein Interactions. Biophys. J. 2008, 95, 1851−1865. (51) Sparr, E.; Ash, W. L.; Nazarov, P. V.; Rijkers, D. T. S.; Hemminga, M. A.; Tieleman, D. P.; Killian, J. A. Self-Association of Transmembrane A-Helices in Model Membranes: Importance of Helix Orientation and Role of Hydrophobic Mismatch. J. Biol. Chem. 2005, 280, 39324−39331. (52) Scarpelli, F.; Drescher, M.; Rutters-Meijneke, T.; Holt, A.; Rijkers, D. T. S.; Killian, J. A.; Huber, M. Aggregation of Transmembrane Peptides Studied by Spin-Label EPR. J. Phys. Chem. B 2009, 113, 12257−12264. (53) Schmidt, U.; Guigas, G.; Weiss, M. Cluster Formation of Transmembrane Proteins due to Hydrophobic Mismatching. Phys. Rev. Lett. 2008, 101, 128104. (54) Marčelja, S. Lipid-Mediated Protein Interaction in Membranes. Biochim. Biophys. Acta, Biomembr. 1976, 455, 1−7. (55) Dan, N.; Berman, A.; Pincus, P.; Safran, S. Membrane-Induced Interactions between Inclusions. J. Phys. II 1994, 4, 1713−1725. (56) Castillo, N.; Monticelli, L.; Barnoud, J.; Tieleman, D. P. Free Energy of WALP23 Dimer Association in DMPC, DPPC, and DOPC Bilayers. Chem. Phys. Lipids 2013, 169, 95−105. (57) Strandberg, E.; Esteban-Martín, S.; Ulrich, A. S.; Salgado, J. Hydrophobic Mismatch of Mobile Transmembrane Helices: Merging Theory and Experiments. Biochim. Biophys. Acta, Biomembr. 2012, 1818, 1242−1249. (58) Kim, T.; Im, W. Revisiting Hydrophobic Mismatch with Free Energy Simulation Studies of Transmembrane Helix Tilt and Rotation. Biophys. J. 2010, 99, 175−183. 4076


Article

The Journal of Physical Chemistry B via Measurement of Strong Coupling Parameters. Biophys. J. 2015, 109, 2317−2327. (82) van Meer, G. Cellular Lipidomics. EMBO J. 2005, 24, 3159− 3165. (83) van Meer, G.; Voelker, D. R.; Feigenson, G. W. Membrane Lipids: Where They Are and How They Behave. Nat. Rev. Mol. Cell Biol. 2008, 9, 112−124. (84) Polley, A.; Vemparala, S.; Rao, M. Atomistic Simulations of a Multicomponent Asymmetric Lipid Bilayer. J. Phys. Chem. B 2012, 116, 13403−13410. (85) Polley, A.; Mayor, S.; Rao, M. Bilayer Registry in a Multicomponent Asymmetric Membrane: Dependence on Lipid Composition and Chain Length. J. Chem. Phys. 2014, 141, 064903. (86) Kiessling, V.; Crane, J. M.; Tamm, L. K. Transbilayer Effects of Raft-like Lipid Domains in Asymmetric Planar Bilayers Measured by Single Molecule Tracking. Biophys. J. 2006, 91, 3313−3326. (87) Lin, Q.; London, E. Ordered Raft Domains Induced by Outer Leaflet Sphingomyelin in Cholesterol-Rich Asymmetric Vesicles. Biophys. J. 2015, 108, 2212−2222. (88) Devaux, P. F.; Morris, R. Transmembrane Asymmetry and Lateral Domains in Biological Membranes. Traffic 2004, 5, 241−246.

4077


Effects of Transmembrane Î±-Helix Length and Concentration on

Recommend Documents