Molecular Dynamics Simulations Reveal Leaflet Coupling in

Apr 22, 2019 - †Department of Physics and ‡Department of Molecular Biology and Genetics, Cornell University , Ithaca , New York 14853 , United Sta...
0 downloads 0 Views 13MB Size
Subscriber access provided by UNIV OF LOUISIANA

B: Biomaterials and Membranes

Molecular Dynamics Simulations Reveal Leaflet Coupling in Compositionally Asymmetric Phase-Separated Lipid Membranes Michael D. Weiner, and Gerald W. Feigenson J. Phys. Chem. B, Just Accepted Manuscript • DOI: 10.1021/acs.jpcb.9b03488 • Publication Date (Web): 22 Apr 2019 Downloaded from http://pubs.acs.org on April 23, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Molecular Dynamics Simulations Reveal Leaflet Coupling in Compositionally Asymmetric Phase-Separated Lipid Membranes Michael D. Weiner† and Gerald W. Feigenson∗,‡ †Department of Physics, Cornell University, Ithaca, New York 14853, United States ‡Department of Molecular Biology and Genetics, Cornell University, Ithaca, New York 14853, United States E-mail: [email protected] Phone: +1 (607) 255-4744

1

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract The eukaryotic plasma membrane has an asymmetric distribution of its component lipids. Rafts that result from liquid-liquid phase separation are a feature of its exoplasmic leaflet, but how these exoplasmic leaflet domains are coupled to the cytoplasmic leaflet is not understood. These rafts can be studied in model membranes of three-component mixtures that produce coexisting liquid ordered and liquid disordered domains. We conducted all-atom Molecular Dynamics simulations of compositionally asymmetric lipid bilayers that reflect a more realistic model of the plasma membrane. One leaflet contained phase-separated domains with phosphatidylcholine and cholesterol, representing the exoplasmic leaflet, while the other contained phosphatidylethanolamine, phosphatidylserine, and cholesterol, which are the predominant components of the cytoplasmic leaflet. Inspired by findings of domain alignment across the two leaflets in compositionally symmetric model membranes, we examined coupling between the two leaflets to see how the single-phase cytoplasmic leaflet would respond to phase separation in the other leaflet and if information could be communicated across the membrane. We found the region of the single-phase leaflet apposing the Lo domain to be slightly more ordered and thicker than the region apposing the Ld domain. The region across from the Lo domain is somewhat enriched in cholesterol and significantly depleted of polyunsaturated lipids.

Introduction Lipid bilayers play a central role in living organisms, cordoning off regions of space to allow the formation of different environments. Among these lipid membranes is the plasma membrane (PM) of the eukaryotic cell, which defines the cell itself and serves as the interface between the cell’s interior and its surroundings. The PM consists of numerous types of lipids, which are distributed asymmetrically between the exoplasmic and cytoplasmic leaflets of the bilayer. 1 The exoplasmic leaflet is enriched in sphingomyelin and phosphatidylcholine (PC),

2

ACS Paragon Plus Environment

Page 2 of 27

Page 3 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

while the cytoplasmic leaflet features phosphatidylethanolamine (PE) and phosphatidylserine (PS). 2 Cholesterol makes up 30-45 mol% of the PM, but its distribution is not well characterized. 3,4 This asymmetric distribution of phospholipids, while recognized for several decades, has proved challenging to replicate in experimental model membranes. Typical synthetic membranes formed in the laboratory contain matching compositions of lipids in their two leaflets. Recent years have brought some advances in this field, especially with the introduction of methods based on cyclodextrin-mediated exchange, 5 where lipids on the outer leaflet of an initially-symmetric vesicle are partially exchanged for donor lipids of a different type. By enveloping the hydrophobic tail of the lipids, cyclodextrin solubilizes them and enables this process. The resultant vesicles are still limited in composition, achieving only 50%-70% asymmetry. 6,7 Others have studied vesicles budded directly from a cell PM, 8,9 preserving the bounty of lipid varieties in the natural system, though they may not maintain their full asymmetry. Fortunately, computational studies are unbound by laboratory limitations and can be parameterized from symmetric experiments and then used to explore complex asymmetric mixtures in silico. Researchers have taken advantage of this flexibility to simulate highly complex lipid compositions that are more representative of the animal cell PM. 10 In this study, we chose a system that retains many of the essential features of the PM lipid composition but reduces the number of components, both to allow study of specific properties and to offer the hope of a laboratory comparison with model membranes in the coming years. The exoplasmic leaflet is thought to contain rafts, lipid domains with different physical properties from the surrounding membrane. 11 These rafts offer an environment for recognition sites on the cell’s exterior, which can favor either raft or non-raft regions, and can be involved in signaling processes. Rafts, which require the presence of cholesterol, may form as a separate lipid phase with a large fraction of high-melting-temperature lipid, such as sphingomyelin. Since the essential physical property of coexisting domains can be reproduced

3

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

by model membranes, both in vitro and in silico, they form an ideal tool for exploring the behavior of rafts. These model membranes contain coexisting liquid phases, both of which contain rapidly diffusing lipids. The liquid ordered (Lo) phase has higher order, more similar to that typically found in a solid, yet its components have mobility of a liquid, while the liquid disordered (Ld) phase has low order, like most liquids. A mixture of as few as three lipids may form coexisting Lo and Ld domains containing a phospholipid that is below its melting temperature, and phospholipid above its melting temperature, and cholesterol. 12 Phase diagrams built from experimental results on symmetric bilayers (Figure 1) reveal a variety of phase behaviors, including coexistence of two immiscible liquids at certain lipid component ratios. It is notable that the high-melting-temperature lipids, such as sphingomyelin, necessary to form the Lo phase are found mainly in the exoplasmic leaflet, so the raft exists only in that location. How information in the form of separated Ld+Lo domains in the exoplasmic leaflet could be transmitted across the membrane is the subject of this study. Studies of lipid bilayers, 13–15 in simulation, 16–20 in theory, 21–24 and in the laboratory, 6,25–27 have revealed coupling between the two leaflets of the bilayer. The study of coupling has incorporated consideration of interacting domains of different lipid composition (see following paragraph), where interleaflet coupling determines how each leaflet’s composition and phase behavior affects domain formation. Simulations generally employ standard force fields without coupling as an explicit input, so findings of coupling, such as ours, depend on the use of a well-parameterized force field. Some simulations have measured coupling through small changes in the number of lipids present in each leaflet, employing this numerical asymmetry to illustrate coupling. 20 Many of the simulations have employed coarse-grained force fields, which simplify the model of lipid molecules. This simplification increases the speed of simulation sufficiently to observe spontaneous liquid-liquid demixing and the formation of coexisting domains. 17,28,29 Without the speed of coarse-grained simulations, atomistic studies can use pre-separated phases, as we did, but lose the ability to study the separation

4

ACS Paragon Plus Environment

Page 4 of 27

Page 5 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

itself. Experimentalists have constructed both symmetric 25 and asymmetric 6,26,27 vesicles and supported bilayers to examine the effect of an ordered or cholesterol-enriched domain in one leaflet on the apposed leaflet. They have observed the induction of a distinct region opposite the ordered domain identified by a change in the partition coefficient of a fluorescent dye, perhaps due to a change in order, even in compositions that have no regions of distinct physical properties in symmetric bilayers. Most often, coupling in symmetric systems appears as registration of phase-separated domains in each leaflet, meaning the alignment of like domains. In model membrane vesicles 12,25 and simulations, 17–19 phase-separating macrodomain-forming mixtures frequently form registered bilayers, with Lo domains apposing Lo domains and Ld domains apposing Ld domains. In certain cases, however, antiregistration is observed, 18,19,30 and it has been theorized that such behavior is due to the balance of competing interactions, with interactions favoring like-domain contact at the bilayer midplane, which promotes registration, balanced against minimizing hydrophobic mismatch at domain interfaces, which promotes antiregistration. 24 Chain interdigitation at the bilayer midplane 16 is but one interaction that could favor registered domains. We extend this exploration to asymmetric bilayers, specifically systems in which only one leaflet is phase-separated, and ask if coupling can still occur in some form. We have found that coexisting phase domains in one leaflet can affect the order and local composition of the other leaflet, even if it lacks separated domains. This coupling suggests how rafts in the exoplasmic leaflet could affect the cytoplasmic leaflet of the PM and interact with the cell’s interior environment.

Methods We conducted all-atom Molecular Dynamics simulations on the Anton2 supercomputer, 31 using version 1.27.0 of its customized Desmond software. We employed a multi-step process

5

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

to generate asymmetric bilayers of appropriate composition. The composition for the exoplasmic leaflet was selected to produce approximately equal fractions of coexisting Lo and Ld phases; the two cytoplasmic leaflet compositions explored contrasting systems with and without polyunsaturated acyl chains while representing the most common lipid headgroup components of the physiological leaflet. We began by building four symmetric bilayers of homogeneous lipid distribution in CHARMM-GUI. 33,34 Two symmetric bilayers contained DSPC (distearoylphosphatidylcholine, di-18:0 PC), DOPC (dioleoylphosphatidylcholine, di-18:1 PC), and cholesterol, with one system in the ratio of an Lo mixture and one in the ratio of an Ld mixture. These phase compositions were the endpoints of a tieline measured experimentally 12 that passed through 30 mol% cholesterol near its midpoint (Figure 1). These three lipids spontaneously form coexisting liquid phases at the averaged composition and have been characterized experimentally in great detail. They represent a minimal model of plasma membrane rafts that would occur in the exoplasmic leaflet. Portions of these two single-phase bilayers were merged to create an already-demixed initial condition for the symmetric exoplasmic leaflet model containing 30 mol% DSPC, 40 mol% DOPC, and 30 mol% cholesterol. These two coexisting domains were observed to mix only slightly during simulation. The other two independent symmetric bilayers generated in CHARMM-GUI represented two different models for the cytoplasmic leaflet, which is rich in PE and PS. 2 The "simple" model contained 40 mol% POPE (palmitoyl,oleoyl-phosphatidylethanolamine, 16:0,18:1PE), 30 mol% POPS (palmitoyl,oleoyl-phosphatidylserine, 16:0,18:1-PS), and 30 mol% cholesterol, while the "complex" model replaced half of the POPE with SAPE (stearoyl,arachidonoylphosphatidylethanolamine, 18:0,20:4-PE). The simple model, containing only a single type of acyl chain, allowed for exploration of the differences between the two most prominent headgroups of the cytoplasmic leaflet, PE and PS. Adding a polyunsaturated lipid, SAPE, permitted examination of the affect of different acyl chains, comparing monounsaturated and polyunsaturated lipids. Both models were simulated from homogeneous initial conditions.

6

ACS Paragon Plus Environment

Page 6 of 27

Page 7 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Ld

Lo

Pre-separated domains Lo

Ld

DSPC DOPC chol POPE POPS SAPE

Figure 1: Experimental Phase Diagram and Methods. This experimentally-derived phase diagram (upper panel), modified from ref 12 for DSPC/DOPC/chol, displays the behavior of the three-component lipid system used for the exoplasmic leaflet in this study. The gray star marks the simulated system, containing 30 mol% DSPC, 40 mol% DOPC, and 30 mol% cholesterol. It is found in the Ld+Lo coexistence region, along a tieline (gray line) whose endpoints formed the mixtures for the initial single-phase bilayers. Mixtures based on these endpoints, one in the Lo phase (red star) and one in the Ld phase (blue star), were placed into single-phase leaflets. A cylinder representing half of the bilayer’s area was taken from the Lo bilayer (red) and placed into the Ld bilayer (blue), from which an equivalent disk had been removed, 32 forming a pre-separated bilayer (lower panel). 7

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Each system consisted of approximately 1000 total lipids and was solvated with approximately 50 water molecules per lipid, plus neutralizing potassium ions. These systems were energy minimized and then equilibrated for 4.75 ps using the CHARMM-GUI equilibration scripts in Gromacs 5.1.3. 35 Following 10 ns of additional equilibration in Gromacs, the system was converted for work on Anton2 using InterMol. 36 There, the system underwent a 5-µs production run on Anton with a Nose-Hoover thermostat (310 K) and semiisotropic barostat in an NPT ensemble. This production run employed the CHARMM36 all-atom force field 37–40 with TIP3P water. 41 The simulations were conducted with a 2.5 fs timestep using u-series electrostatics with a 0.88 nm cutoff. Next, results from the symmetric simulations were used to form two asymmetric bilayers, one with each cytoplasmic leaflet model. Cross-sectional area was calculated for each symmetric system, and asymmetric bilayers were formed in CHARMM-GUI with the number of lipids in each leaflet chosen to produce an expected tension-free bilayer after equilibration based on matching leaflet equilibrated areas (not initial areas). 10,19 While this method may be imperfect, it is in common use, and a mismatch of less than 5% is unlikely to severely affect the results, especially for measurements of saturated acyl chains. 42 These asymmetric bilayers were simulated using the same protocols described above. Data analysis was performed using custom Python scripts incorporating the MDTraj library. 43 Autocorrelation measurements of each key measured observable were performed, and findings are described in the Supporting Information. In order to ensure accurate statistics through the use of independent frames, 44 results were calculated from frames in the final microsecond of simulation, separated by twice the autocorrelation decay time, which differed for each property. The final microsecond, while more stable, does not represent a system at equilibrium, as demonstrated in the Supporting Information graphs of transbilayer partition coefficients (Figure S1). Since partition coefficient, described in detail in Results and Discussion below, represents the slow process of diffusion of lipids throughout the bilayer, it will adjust more slowly than molecular properties, such as order parameter. Even with

8

ACS Paragon Plus Environment

Page 8 of 27

Page 9 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

the Anton2 supercomputer and pre-separated phases, all-atom simulation equilibrium occurs too slowly when the Lo phase is present to be simulated. Even 9-µs simulations may not reach equilibrium. 32 However, while the bilayers may not be at equilibrium, the key finding of a difference between the cytoplasmic leaflet region apposing an Lo domain and the region apposing an Ld domain is robust. The Supporting Information also shows that the thickness of each region converges quickly and remains distinct over the full simulation (Figure S2). To identify the Lo and Ld domains in the exoplasmic leaflet, we employed a Voronoi tesselation of the phosphorus (for phospholipids) or oxygen (for cholesterol) atoms, which divided the xy-plane of the bilayer into cells, with one for each lipid in the leaflet. 32 For each lipid, the local composition, taken from the lipid itself and the other lipids sharing a Voronoi edge, was compared to that of the full leaflet. An enrichment of DSPC and cholesterol labelled the lipid Lo. For all other portions of the analysis, a lipid’s neighbors consisted of those nearby lipids in both leaflets. Specifically, all phosphorus or oxygen atoms were projected into the xy-plane (where the z-axis is the bilayer normal), and lipids within 2 nm of the lipid of interest were termed neighbors; using a smaller radius produced similar results in most cases but occasionally failed to find any neighbors in the apposed leaflet. These neighbors were used throughout the analysis. The apposed phase of lipids in the cytoplasmic leaflet was defined as the predominant phase among its exoplasmic leaflet neighbors.

Results and discussion We measured the partitioning of inner leaflet lipids, comparing their prevalence across from an Lo domain to their prevalence across from an Ld domain. This partition coefficient, the ratio of each lipid’s prevalence across from (apposing) Lo to its prevalence apposing Ld, normalized by the leaflet-wide ratio for all lipids, is termed the "transbilayer Kp." Table 1 displays the Kp values for each asymmetric system. Transbilayer Kp values for the symmetric phase-separated bilayer are presented for comparison in Table 2. The registration

9

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Table 1: Asymmetric Partition Coefficients. Transbilayer Kp is the ratio of each lipid’s presence apposing Lo phase to apposing Ld phase, normalized by the leaflet-wide ratio for all lipids. Higher numbers in the right column indicate preference of that lipid to locate apposing an Lo domain. A value of 1 represents no preference. Simple and Complex refer to the two cytoplasmic leaflets tested in this study, without and with polyunsaturated acyl chains presents, respectively. Values include standard error. Mixture Simple Simple Simple Complex Complex Complex Complex

Lipid POPE POPS cholesterol POPE POPS SAPE cholesterol

Transbilayer Kp 0.93 ± 0.03 0.95 ± 0.03 1.16 ± 0.04 1.00 ± 0.02 1.02 ± 0.02 0.81 ± 0.03 1.13 ± 0.04

Table 2: Symmetric Partition Coefficients. Transbilayer Kp is the ratio of each lipid’s presence apposing Lo phase to apposing Ld phase, normalized by the leaflet-wide ratio for all lipids. Higher numbers in the right column indicate preference of that lipid to locate apposing an Lo domain. A value of 1 represents no preference. Since both leaflets contain coexisting Lo and Ld domains, the partitioning of DSPC towards Lo and DOPC towards Ld is indicative of registered phase domains. Values include standard error. Lipid Transbilayer Kp DSPC 2.40 ± 0.07 DOPC 0.38 ± 0.01 cholesterol 1.32 ± 0.03

10

ACS Paragon Plus Environment

Page 10 of 27

Page 11 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(alignment) of phases is apparent, as DSPC and cholesterol, the primary components of Lo, are prevalent across from Lo, and DOPC, the primary component of Ld, is prevalent across from Ld. For the asymmetric cases, no phase separation exists in the cytoplasmic leaflet, but some lipids do partition unequally across from a particular phase in a less extreme manner. Cholesterol prefers to partition across from Lo, which is a high-cholesterol phase, while the polyunsaturated SAPE favors apposing Ld. In the complex system, POPE and POPS do not appear strongly affected by the other leaflet, with values indistinguishable from 1.00. However, the location of the monounsaturated lipids may be driven primarily by the strong partitioning (Kp = 0.81) of the polyunsaturated lipid, which forces many of the POPE and POPS into the region apposing Lo. In the simple system, they exhibit some preference for apposing Ld, though this may be explained simply by the paucity of cholesterol in those regions and the absence of a polyunsaturated chain that more strongly prefers the Ld-influenced regions. The statistically significant decrease in their partition coefficients in the system lacking SAPE reveals the importance of including polyunsaturated chains, a major component of the PM’s cytoplasmic leaflet, in the model. The transbilayer Kp values form clear evidence of a weak form of alignment across the bilayer with chain polyunsaturation dominating over headgroups. Another measure of coupling is acyl chain order parameter. The carbon-hydrogen order parameter (-SCH ) measures the alignment of the phospholipid acyl chains relative to the bilayer normal. Higher values (maximum 0.5, minimum -1.0) represent a more aligned state, where acyl chains follow the bilayer normal. Lo phase lipids have a higher value than Ld phase lipids. The order parameter is defined as

−S CH = −

3hcos2 (α)i − 1 2

(1)

where α is the angle between the carbon-hydrogen bond in the acyl chain and the bilayer normal. Figures 2 and 3 show the average order parameter of the different lipids present in

11

ACS Paragon Plus Environment

The Journal of Physical Chemistry

A. DSPC in Complex Mixture

Mixture Symmetric Symmetric Asymmetric Asymmetric

Phase Lo Ld Lo Ld

B. DOPC in Complex Mixture

Order parameter (-SCH)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

C. DSPC in Simple Mixture

D. DOPC in Simple Mixture

sn1 acyl chain carbon number

Figure 2: Exoplasmic Chain Order Parameter. The carbon-hydrogen order parameter for the sn1 chain of each PC lipid is presented by carbon number in the acyl chain, where higher values represent a more ordered chain. Both the simple and complex asymmetric bilayers (dashed lines) are compared to the same symmetric bilayer (solid lines). The higher order of the Lo phase (black) over the Ld phase (gray) is clear, as is the double bond and corresponding dip in order for DOPC at carbon 9. Error bars represent standard error.

12

ACS Paragon Plus Environment

Page 12 of 27

Page 13 of 27

A. POPE in Complex Mixture

Mixture Symmetric Asymmetric Asymmetric

Apposed Lo Ld

B. POPS in Complex Mixture

Order parameter (-SCH)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

C. SAPE in Complex Mixture

D. POPE in Simple Mixture

E. POPS in Simple Mixture

Saturated acyl chain carbon number

Figure 3: Cytosolic Chain Order Parameter. The carbon-hydrogen order parameter for the saturated chain of each PE or PS lipid is presented by carbon number in the acyl chain, where higher values represent a more ordered chain. For each case, the order of the lipid in the region apposing an Lo (dashed black 13 lines) or Ld (dashed gray lines) domain in the ACS Paragon Plus Environment asymmetric case is compared to a symmetric bilayer (solid black lines). The disordering of the lipids across from an Ld domain appears. Error bars represent standard error.

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the study. The average order for each carbon in the acyl chain is graphed, beginning with the third carbon from the glycerol. The order parameter generally declines towards the bilayer center, and a sharp drop occurs near a double bond, as at carbons 9-11 of DOPC (Figure 2B,D). Since the double bond can mask comparisons of order, we present order of sn1 chains only, which are saturated for all lipids present except DOPC. For the exoplasmic phase-separated leafet PCs (Figure 2), the effect of the apposed leaflet’s composition is small. In every case, DSPC and DOPC in the Lo domain are more ordered than those lipids in Ld, and this is the case at every carbon along the acyl chains. The finding reveals that, in a system of coexisting liquid macrodomains, the order is robustly determined by the phases in that leaflet, with the apposed leaflet having only a small influence. This is particularly clear when one considers that the symmetric bilayer is registered, with the Lo domains aligned across the bilayer. Replacing a phase-separated leaflet with a single-phase Ld leaflet does not significantly disorder the remaining Lo domain. This behavior applies to both DSPC and DOPC, which always exhibit order consistent with their phase and neighboring lipids, not with their order in a homogeneous environment (e.g., DOPC, a "disordered" unsaturated lipid, has high order when in an Lo phase). The PE and PS of the cytoplasmic leaflet (Figure 3), in contrast, exhibit a larger change in order depending on the apposed leaflet. When the other leaflet contains Lo and Ld domains, the order of both PE and PS decreases across from the Ld domain. The order of PE and PS across from Lo match their order in a PE/PS/cholesterol symmetric bilayer. Although the cytoplasmic leaflet contains only a single (Ld) phase, its regions exhibit coupling to their partner lipids across the bilayer, mimicking the registration behavior commonly found in symmetric phase-separated systems. Here, we see the region opposite Lo having higher order and the region opposite Ld lower. While less dramatic than phase separation, it reveals how a mixture of lipids incapable of forming the Lo phase can still respond to phase separation in the other leaflet by having regions of order separation. Such behavior has been observed experimentally, 6 as mentioned in the introduction. A small difference in order

14

ACS Paragon Plus Environment

Page 14 of 27

Page 15 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

between the regions apposing Lo and Ld domains, as we find, could appear as a significant difference in the partition coefficient of a fluorescent dye, even if both regions are part of an Ld phase. Because of this possibility of creating distinct bilayer environments that could sort other components of the membrane, even order differences within a phase are significant to interpreting experimental findings. Comparing sections of Figure 3 shows that POPE and POPS have nearly identical order parameter curves, demonstrating that these headgroups do not determine order. This finding, like all the others regarding order, appears in both asymmetric systems, with or without the polyunsaturated acyl chain present. The polyunsaturated SAPE has order only slightly lower than that of monounsaturated lipids in the same leaflet but does respond to the disorder induced by an apposed Ld domain. While the cytoplasmic leaflet mixture of PE, PS, and cholesterol forms an Ld phase, it is less disordered than the DOPC-rich Ld domain of the exoplasmic leaflet (compare Figures 2 and 3). Both cytoplasmic models employed represent a relatively ordered Ld phase. Compared to the DOPC-rich Ld domain of the exoplasmic leaflet (Figure 1), the cytoplasmic Ld has a high fraction of cholesterol, which promotes order among neighboring phospholipids. Both leaflets contain 30 mol% cholesterol, but much of the exoplasmic cholesterol partitions into the Lo domain, while the single-phase cytoplasmic leaflet has nowhere to send cholesterol away from the Ld phase. Care should be taken when comparing our findings to those of asymmetric bilayer experiments, which often employ a single-phase Ld leaflet even more highly disordered than our phase-separated Ld domain, such as a cholesterol-free POPC and POPE mixture. 7 In addition to examining chain order parameter of each lipid, we also measured the thickness by phase of each leaflet, and of the bilayer as a whole, for each system. The bilayer’s thickness was defined as the average distance from the phosphorus atoms in one leaflet to those in the other along the bilayer normal, while the thickness of an individual leaflet was defined as the average distance from its phosphorus atoms to the carbon atoms of its acyl

15

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Symmetric Exoplasmic

Ld

2.27 4.66 2.06

Symmetric Complex Cytoplasmic

Lo

1.99 4.30 2.29 4.85 Simple Asymmetric

Lo

Symmetric Simple Cytoplasmic

Page 16 of 27

Ld 2.00 4.39 2.03

2.06 4.48 Complex Asymmetric

Lo 2.28 4.63 2.05

Ld 2.03 4.36 2.02

2.05 4.43 Lipid components: DSPC DOPC chol POPE POPS SAPE All measurements ± 0.01 nm

Figure 4: Leaflet and Bilayer Thickness. The five bilayers simulated are presented, with lipids colored by type. The simple cytoplasmic system lacks the polyunsaturated SAPE. Phase-separated exoplasmic leaflets display an Lo phase, enriched in DSPC and cholesterol, and an Ld phase, enriched in DOPC. Numbers indicate the thickness of a leaflet or full bilayer. Leaflet thickness is defined as the average distance from the phosphorus atom to the terminal methyl’s carbon atom, while the full bilayer thickness is measured from phosphorus to phosphorus. The thinning of the cytoplasmic leaflet region apposing an Ld phase is visible. Standard error is ± 0.01 nm per measurement. chain terminal methyl groups. The local thickness can be measured for each lipid based on its neighbors in both leaflets, and the value for a domain is determined by the average of the lipids in that domain. Thickness is closely correlated with order parameter, as a more ordered lipid is taller and occupies less cross-sectional area. The measured thicknesses (Figure 4) confirm the most important finding of the order parameter measurements: that the cytoplasmic leaflet apposing an Lo domain matches the behavior of that same lipid mixture in a symmetric system, while those lipids apposing an Ld domain are comparatively disordered, resulting in a thinner leaflet. The regions of the single-phase cytoplasmic leaflet across from a phase-separated Ld domain were found to be approximately 0.03 nm thinner than those across from the neighboring Lo domain. From the symmetric exoplasmic-leaflet mixture, it was seen that an Lo bilayer is approximately 0.6 nm thicker than an Ld bilayer, which is consistent with extrapolating experimental findings at lower cholesterol concentrations. 45,46 The finding of the symmetric cytoplasmic Ld bilayers having phosphate-phosphate thick-

16

ACS Paragon Plus Environment

Page 17 of 27

ness of 4.48 nm (simple) or 4.43 nm (complex) makes them about 0.6 nm thicker than a single-component DOPC bilayer, 16 which confirms that our cytoplasmic leaflets represent a significantly more ordered mixture than the pure DOPC leaflets sometimes used in experimental asymmetric bilayers.

1

Probability

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

0.8 0.6 0.4 POPE (18:1) POPS (18:1) SAPE (20:4)

0.2 0

0.5

0.7

0.9

1.1

POPE (16:0) POPS (16:0) SAPE (18:0) 1.3

1.5

1.7

Distance from phosphorus (nm)

1.9

2.1

Figure 5: Acyl chain carbon density. Each acyl chain’s carbon probability density reveals the distribution of its position in the leaflet. Saturated acyl chains in the sn1 position (dashed lines) have distributions shifted towards the interior of the bilayer, while unsaturated acyl chains in the sn2 position (solid lines) are shifted towards the bilayer-water interface. Above the graph, an image of a POPE molecule aligned to the graph elucidates the x-axis distance of the carbons (black) from the phosphorus atom (orange). Tails of the curves are hidden beyond the axes limits to focus on the primary region of interest. Comparing the thickness measurements for the individual leaflets and the bilayer as a whole reveals that the leaflet sum is about 0.3 nm less than the total thickness. Since the leaflet thickness is calculated from the average position of acyl chain terminal methyl groups along the bilayer normal, this reveals that they are, on average, above the bilayer midplane; i.e., there is little interdigitation. This finding confirms the idea of the middle of the bilayer as a region of low density. 47 In order to better understand the structure of the cytoplasmic leaflet lipids in their environment and the lack of interdigitation, we examined the spatial distribution of their acyl chains. Figure 5 presents the probability density of acyl chain carbons along the bilayer normal. The six acyl chains, two from each of the hybrid

17

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

lipids of the cytoplasmic leaflet of the complex asymmetric bilayer, have differing carbon densities as a function of the distance from the phosphorus atom of the lipid along the bilayer normal. POPE and POPS have very similar distributions, once again highlighting the lack of headgroup dependence. The sn1 chains, which are saturated, have higher probability to be towards the bilayer center, while the unsaturated sn2 chains, which are bent by double bonds, have little presence close to the other leaflet and tend towards the bilayer-water interface. The polyunsaturated 20:4 chain of SAPE is the most shifted outward. The large number of kinks from double bonds leave many of its chains near the surface, far from the center of the bilayer. All chains reach a very low probability density beyond the midplane (about 2.05 nm), again indicating a low probability of interdigitation, which would see acyl chains extending beyond the midplane. Unlike phospholipids, cholesterol molecules are nearly completely hydrophobic. This property allows them to pass easily through the center of the bilayer, moving from one leaflet to the other. While this cholesterol flip-flop is widely observed in coarse-grained simulations, it is less frequently observed in all-atom ones, due to the shorter time scales and higher energy barriers. 48 Over the five-µs simulations, net movement of cholesterol, i.e. the change in cholesterol composition of each leaflet, was recorded. In no case was there a significant net movement of cholesterol from one leaflet to the other, though there was a slow movement of about 1% of cholesterols towards the exoplasmic leaflet (see Supporting Information, Figure S3). This small value indicates that the initially chosen asymmetric compositions were well-suited as the starting configuration for the simulations. Had the number of lipids in each leaflet been chosen poorly, we would have expected to see a net movement of cholesterol from the leaflet where it had higher chemical potential to the leaflet where it had lower chemical potential. Since some cholesterols flipped, we know this would be possible, so the result reassures suitability of the preparation of the initial states. It is also of interest to measure the rate of cholesterol flip-flop, not just its net result. As with the other results, we analyzed the final microsecond of simulation to count cholesterol

18

ACS Paragon Plus Environment

Page 18 of 27

Page 19 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

flipping events. In all cases, the total number of flips was quite small. The symmetric outer leaflet mixture saw only one flip, from one Lo domain to the one apposing it. The simple asymmetric bilayer saw one cholesterol move in each direction, both involving the Ld domain. The complex asymmetric bilayer contained five flipping cholesterol molecules, with three passing towards the inner leaflet and two moving outwards. With such a small number of observations, we cannot make generalized claims about the nature of these transitions, but flipping of hydrophobic cholesterol is expected in simulations, and the small values indicate that they are not changing the composition of any leaflet or domain.

Conclusion Asymmetric model membranes offer increased insight into physiological behavior. We examined a system containing a phase-separated exoplasmic leaflet of DSPC, DOPC, and cholesterol with a cytoplasmic leaflet of PE, PS, and cholesterol. The order and composition of the phase-separated exoplasmic leaflet did not change much relative to those found in a symmetric bilayer. The cytoplasmic leaflet, a single phase system, reacts to the presence of coexisting domains in the apposed leaflet through small compositional inhomogeneities and changes in order and thickness of the regions of the leaflet across from the phase-separated domains. The slight reorganization of lipids produces a slightly thicker, more ordered region apposing Lo and a slightly thinner, less ordered region apposing Ld. This differentiation is achieved by disordering the region across from Ld from its original state in a symmetric bilayer. Notably, if a polyunsaturated lipid is added, it favors neighboring the Ld domain by about 20%; cholesterol always favors a location near the Lo domain by about 15%. Our results reveal the application of studies of domain registration in symmetric bilayers to asymmetric models and suggest how rafts, a phenomenon of the PM’s exoplasmic leaflet, affect cytoplasmic leaflet lipids. The relationship between physical properties in the cytoplasmic and exoplasmic leaflets connects the cell’s interior and exterior and can communicate

19

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

information between the cell and its surroundings.

Supporting Information Available The Supporting Information is available free of charge. • Uncertainty and autocorrelation methods; time series of transbilayer partition coefficient, leaflet thickness, and cholesterol distribution.

Acknowledgement This work was supported by funding from the U.S. National Science Foundation (MCB1410926) and the U.S. National Institutes of Health (GM105684) to G.W.F. Anton 2 computer time was provided by the Pittsburgh Supercomputing Center (PSC) through Grant R01GM116961 from the NIH. The Anton 2 machine at PSC was generously made available by D.E. Shaw Research. This work also used the Extreme Science and Engineering Discovery Environment (XSEDE), 49 which is supported by National Science Foundation grant number ACI-1548562, on Bridges at the Pittsburgh Supercomputing Center (TG-MCB130010 to G.W.F.). M.D.W. was supported in part by NIH Training Grant 1-T32-GM08267. The authors thank Edward Lyman, Milka Doktorova, and George Khelashvili for useful conversations.

References (1) Fujimoto, T.; Parmryd, I. Interleaflet coupling, pinning, and leaflet asymmetry-major players in plasma membrane nanodomain formation. Frontiers in Cell and Developmental Biology 2017, 4, 155. (2) Verkleij, A. J.; Zwaal, R. F. A.; Roelofsen, B.; Comfurius, P.; Kastelijn, D.; van Dee-

20

ACS Paragon Plus Environment

Page 20 of 27

Page 21 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

nen, L. L. M. The asymmetric distribution of phospholipids in the human red cell membrane. A combined study using phospholipases and freeze-etch electron microscopy. BBA - Biomembranes 1973, 323, 178–193. (3) Devaux, P. F.; Morris, R. Transmembrane asymmetry and lateral domains in biological membranes. Traffic 2004, 5, 241–246. (4) Mondal, M.; Mesmin, B.; Mukherjee, S.; Maxfield, F. R. Sterols are mainly in the cytoplasmic leaflet of the plasma membrane and the endocytic recycling compartment in CHO cells. Molecular Biology of the Cell 2008, 20, 581–588. (5) Lin, Q.; London, E. Preparation of artificial plasma membrane mimicking vesicles with lipid asymmetry. PLoS ONE 2014, 9, e87903. (6) Lin, Q.; London, E. Ordered raft domains induced by outer leaflet sphingomyelin in cholesterol-rich asymmetric vesicles. Biophysical Journal 2015, 108, 2212–2222. (7) Eicher, B.; Marquardt, D.; Heberle, F. A.; Letofsky-Papst, I.; Rechberger, G. N.; Appavou, M. S.; Katsaras, J.; Pabst, G. Intrinsic curvature-mediated transbilayer coupling in asymmetric lipid vesicles. Biophysical Journal 2018, 114, 146–157. (8) Sezgin, E.; Kaiser, H.-J.; Baumgart, T.; Schwille, P.; Simons, K.; Levental, I. Elucidating membrane structure and protein behavior using giant plasma membrane vesicles. Nature Protocols 2012, 7, 1042–1051. (9) Levental, K. R.; Levental, I. Giant plasma membrane vesicles: Models for understanding membrane organization. Current Topics in Membranes 2015, 75, 25–57. (10) Ingólfsson, H. I.; Melo, M. N.; Van Eerden, F. J.; Arnarez, C.; Lopez, C. A.; Wassenaar, T. A.; Periole, X.; De Vries, A. H.; Tieleman, D. P.; Marrink, S. J. Lipid organization of the plasma membrane. Journal of the American Chemical Society 2014, 136, 14554–14559. 21

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(11) Lingwood, D.; Simons, K. Lipid rafts as a membrane-organizing principle. Science 2010, 327, 46–50. (12) 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. Biochimica et Biophysica Acta (BBA) - Biomembranes 2013, 1828, 2204–2214. (13) Nickels, J. D.; Smith, J. C.; Cheng, X. Lateral organization, bilayer asymmetry, and inter-leaflet coupling of biological membranes. Chemistry and Physics of Lipids 2015, 192, 87–99. (14) Kiessling, V.; Wan, C.; Tamm, L. K. Domain coupling in asymmetric lipid bilayers. Biochimica et Biophysica Acta - Biomembranes 2009, 1788, 64–71. (15) May, S. Trans-monolayer coupling of fluid domains in lipid bilayers. Soft Matter 2009, 5, 3148–3156. (16) Capponi, S.; Freites, J. A.; Tobias, D. J.; White, S. H. Interleaflet mixing and coupling in liquid-disordered phospholipid bilayers. Biochimica et Biophysica Acta - Biomembranes 2016, 1858, 354–362. (17) Ackerman, D. G.; Feigenson, G. W. Multiscale modeling of four-component lipid mixtures: Domain composition, size, alignment, and properties of the phase interface. Journal of Physical Chemistry B 2015, 119, 4240–4250. (18) Fowler, P. W.; Williamson, J. J.; Sansom, M. S. P.; Olmsted, P. D. Roles of interleaflet coupling and hydrophobic mismatch in lipid membrane phase-separation kinetics. Journal of the American Chemical Society 2016, 138, 11633–11642. (19) Perlmutter, J. D.; Sachs, J. N. Interleaflet interaction and asymmetry in phase separated lipid bilayers: Molecular dynamics simulations. Journal of the American Chemical Society 2011, 133, 6563–6577. 22

ACS Paragon Plus Environment

Page 22 of 27

Page 23 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(20) Tian, J.; Nickels, J.; Katsaras, J.; Cheng, X. Behavior of bilayer leaflets in asymmetric model membranes: Atomistic simulation studies. The Journal of Physical Chemistry B 2016, 120, 8438–8448. (21) Wagner, A. J.; Loew, S.; May, S. Influence of monolayer-monolayer coupling on the phase behavior of a fluid lipid bilayer. Biophysical Journal 2007, 93, 4268–4277. (22) Bossa, G. V.; Roth, J.; May, S. Modeling lipid-lipid correlations across a bilayer membrane using the quasi-chemical approximation. Langmuir 2015, 31, 9924–9932. (23) Putzel, G. G.; Schick, M. Phase behavior of a model bilayer membrane with coupled leaves. Biophysical Journal 2008, 94, 869–877. (24) Williamson, J. J.; Olmsted, P. D. Registered and antiregistered phase separation of mixed amphiphilic bilayers. Biophysical Journal 2015, 108, 1963–1976. (25) Korlach, J.; Schwille, P.; Webb, W. W.; Feigenson, G. W. Characterization of lipid bilayer phases by confocal microscopy and fluorescence correlation spectroscopy. Proceedings of the National Academy of Sciences 1999, 96, 8461–8466. (26) Kiessling, V.; Crane, J. M.; Tamm, L. K. Transbilayer effects of raft-like lipid domains in asymmetric planar bilayers measured by single molecule tracking. Biophysical Journal 2006, 91, 3313–3326. (27) Wan, C.; Kiessling, V.; Tamm, L. K. Coupling of cholesterol-rich lipid phases in asymmetric bilayers. Biochemistry 2008, 47, 2190–2198. (28) Bennett, W. F. D.; Tieleman, D. P. Computer simulations of lipid membrane domains. Biochimica et biophysica acta 2013, 1828, 1765–76. (29) Risselada, H. J.; Marrink, S. J. The molecular face of lipid rafts in model membranes. Proceedings of the National Academy of Sciences 2008, 105, 17367–17372.

23

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(30) Weiner, M. D.; Feigenson, G. W. The presence and role of midplane cholesterol in lipid bilayers containing registered or antiregistered phase domains. The Journal of Physical Chemistry B 2018, 122, 8193–8200. (31) Shaw, D. E.; Grossman, J. P.; Bank, J. A.; Batson, B.; Butts, J. A.; Chao, J. C.; Deneroff, M. M.; Dror, R. O.; Even, A.; Fenton, C. H. et al. Anton 2: Raising the bar for performance and programmability in a special-purpose molecular dynamics supercomputer. International Conference for High Performance Computing, Networking, Storage and Analysis, SC. 2014; pp 41–53. (32) Sodt, A. J.; Sandar, M. L.; Gawrisch, K.; Pastor, R. W.; Lyman, E. The molecular structure of the liquid-ordered phase of lipid bilayers. Journal of the American Chemical Society 2014, 136, 725–732. (33) Jo, S.; Kim, T.; Iyer, V. G.; Im, W. CHARMM-GUI: A web-based graphical user interface for CHARMM. Journal of Computational Chemistry 2008, 29, 1859–1865. (34) Qi, Y.; Ingólfsson, H. I.; Cheng, X.; Lee, J.; Marrink, S. J.; Im, W. CHARMM-GUI Martini Maker for coarse-grained simulations with the Martini force field. Journal of Chemical Theory and Computation 2015, 11, 4486–4494. (35) Abraham, M. J.; Murtola, T.; Schulz, R.; Páll, S.; Smith, J. C.; Hess, B.; Lindah, E. Gromacs: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 2015, 1-2, 19–25. (36) Shirts, M. R.; Klein, C.; Swails, J. M.; Yin, J.; Gilson, M. K.; Mobley, D. L.; Case, D. A.; Zhong, E. D. Lessons learned from comparing molecular dynamics engines on the SAMPL5 dataset. Journal of Computer-Aided Molecular Design 2017, 31, 147–161. (37) Klauda, J. B.; Venable, R. M.; Freites, J. A.; O’Connor, J. W.; Tobias, D. J.; Mondragon-Ramirez, C.; Vorobyov, I.; MacKerell, A. D.; Pastor, R. W. Update of

24

ACS Paragon Plus Environment

Page 24 of 27

Page 25 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

the CHARMM all-atom additive force field for lipids: Validation on six lipid types. Journal of Physical Chemistry B 2010, 114, 7830–7843. (38) Klauda, J. B.; Monje, V.; Kim, T.; Im, W. Improving the CHARMM force field for polyunsaturated fatty acid chains. The Journal of Physical Chemistry B 2012, 116, 9424–9431. (39) Lim, J. B.; Rogaski, B.; Klauda, J. B. Update of the cholesterol force field parameters in CHARMM. Journal of Physical Chemistry B 2012, 116, 203–210. (40) Beglov, D.; Roux, B. Finite representation of an infinite bulk system: Solvent boundary potential for computer simulations. The Journal of Chemical Physics 1994, 100, 9050– 9063. (41) Neria, E.; Fischer, S.; Karplus, M. Simulation of activation free energies in molecular systems. Journal of Chemical Physics 1996, 105, 1902–1921. (42) Park, S.; Beaven, A. H.; Klauda, J. B.; Im, W. How tolerant are membrane simulations with mismatch in area per lipid between leaflets? Journal of Chemical Theory and Computation 2015, 11, 3466–3477. (43) McGibbon, R. T.; Beauchamp, K. A.; Harrigan, M. P.; Klein, C.; Swails, J. M.; Hernández, C. X.; Schwantes, C. R.; Wang, L. P.; Lane, T. J.; Pande, V. S. MDTraj: A modern open library for the analysis of molecular dynamics trajectories. Biophysical Journal 2015, 109, 1528–1532. (44) Morales, J. J.; Nuevo, M. J.; Rull, L. F. Statistical error methods in computer simulations. Journal of Computational Physics 1990, 89, 432–438. (45) 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. Journal of the American Chemical Society 2013, 135, 6853–6859. 25

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(46) Heftberger, P.; Kollmitzer, B.; Rieder, A. A.; Amenitsch, H.; Pabst, G. In situ determination of structure and fluctuations of coexisting fluid membrane domains. Biophysical Journal 2015, 108, 854–862. (47) Janiak, M. J.; Small, D. M.; Shipley, G. G. Temperature and compositional dependence of the structure of hydrated dimyristoyl lecithin. Journal of Biological Chemistry 1979, 254, 6068–6078. (48) Choubey, A.; Kalia, R. K.; Malmstadt, N.; Nakano, A.; Vashishta, P. Cholesterol translocation in a phospholipid membrane. Biophysical Journal 2013, 104, 2429–2436. (49) Towns, J.; Cockerill, T.; Dahan, M.; Foster, I.; Gaither, K.; Grimshaw, A.; Hazlewood, V.; Lathrop, S.; Lifka, D.; Peterson, G. D. et al. XSEDE: Accelerating scientific discovery. Computing in Science and Engineering 2014, 16, 62–74.

26

ACS Paragon Plus Environment

Page 26 of 27

Page 27 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Graphical TOC Entry

Lo

Ld

27

ACS Paragon Plus Environment

DSPC DOPC chol POPE POPS SAPE