Influence of Ester-Modified Lipids on Bilayer Structure - Langmuir

Oct 24, 2013 - ... Justin A. Lemkul , Shuai Wei , Joshua Buckner , Jong Cheol Jeong ... III , Alexander D. MacKerell , Jr. , Jeffery B. Klauda , and W...
13 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/Langmuir

Influence of Ester-Modified Lipids on Bilayer Structure Diana Y. Villanueva,† Joseph B. Lim,† and Jeffery B. Klauda* Department of Chemical and Biomolecular Engineering, University of Maryland, College Park, Maryland 20742, United States S Supporting Information *

ABSTRACT: Lipid membranes function as barriers for cells to prevent unwanted chemicals from entering the cell and wanted chemicals from leaving. Because of their hydrophobic interior, membranes do not allow water to penetrate beyond the headgroup region. We performed molecular simulations to examine the effects of ester-modified lipids, which contain ester groups along their hydrocarbon chains, on bilayer structure. We chose two lipids from those presented in Menger et al. [J. Am. Chem. Soc. 2006, 128, 14034] with ester groups in (1) the upper half of the lipid chain (MEPC) and (2) the middle and end of the lipid chain (MGPC). MGPC (30%)/ POPC bilayers formed stable water pores of diameter 5−7 Å, but MGPC (22%)/POPC and MEPC (30%)/POPC bilayers did not form these defects. These pores were similar to those formed during electroporation; i.e., the head groups lined the pore and allowed water and ions to transport across the bilayer. However, we found that lateral organization of the MGPC lipids into clusters, instead of an electric field or charge disparity as in electroporation, was essential for pore formation. On the basis of this, we propose an overall mechanism for pore formation. The similarities between the ester-modified lipids and byproducts of lipid peroxidation with multiple hydrophilic groups in the middle of the chain suggest that free radical reactions with unsaturated lipids and sterols result in fundamental changes that may be similar to what is seen in bilayers with ester-modified lipids. process known as electroporation.4,14 These pores consist of water and ions and are lined with lipid head groups.4,14−16 The mechanism of pore formation consists of three steps:14 (1) the polar head groups tilt in response to the electric field, (2) a prepore forms with a water wire (WW) cutting through the membrane with the aid of some head groups protruding into the general hydrophobic region, and (3) more head groups line the pore, enabling larger pores to form fully solvated head groups at the center of the bilayer. Lipid peroxidation of cholesterol or other lipids can result in hydrophilic moieties attached to the cholesterol rings or the centers of the lipid chains.17 These can facilitate pore formation in membranes, as will be discussed later, but first a brief introduction to this phenomenon is presented here. Lipid peroxidation occurs through a free-radical chain reaction, which consists of initiation, propagation, and termination steps. Initiation of chain oxidation in lipids occurs when a free radical abstracts a proton from a methylene group. As the weakest parts of the lipid, allylic carbon positions are most prone to this free-radical attack.17 Propagation occurs in the presence of oxygen, leading to a peroxide radical that can react with other lipids. Free radicals preferentially react with polyunsaturated lipids; reaction rate constants decrease by an order of magnitude for cholesterol and then again for monosaturated lipids.17 Reaction of peroxide radicals with polyunsaturated lipids, such as those with arachidonic acid, can result in many stable modified lipids, e.g., cyclic peroxides, bicyclic endoper-

1. INTRODUCTION As one of the four major building blocks in biology, lipids play an integral role in cell membrane stability. Lipids self-assemble into bilayer structures due to their amphiphilic nature with a hydrophobic core and hydrophilic surface. The bilayer prevents transport of unwanted chemicals into and wanted chemicals out of the cell.1−3 This prevents water and water-soluble substances, such as ions, from residing in and transporting across the bilayer without the aid of proteins or an external electric field.4 Hence, the centers of typical phospholipid bilayers contain almost no water or water-dissolving substances. Electron density profiles from X-ray scattering clearly demonstrate this. For example, in bilayers comprised of 1,2dimyristoyl-sn-phosphatidylcholine, a common phospholipid, X-ray scattering showed that the water density decreases to zero at ∼10−12 Å from the center of the bilayer.5,6 Because the system has no water present at the center of the bilayer, no ions can pass through the membrane. Under normal conditions, the only way water or ions can pass through the membrane is by proteins. Cells typically use channel proteins to control the water and ion concentration gradients across the cell membrane.7−9 Some secondary active transport proteins also transport ions as part of their function.10−12 However, certain peptides, such as antimicrobial peptides, can create pores through which water and ions can pass.13 This destabilizes the membrane and ultimately kills microbes. Additionally, applied electric fields (greater than 0.1 V/nm) or vesicles with a charge imbalance between the intracellular and extracellular environments result in an increase in membrane permeability and thus induce pore formation, a © 2013 American Chemical Society

Received: July 1, 2013 Published: October 24, 2013 14196

dx.doi.org/10.1021/la403919h | Langmuir 2013, 29, 14196−14203

Langmuir

Article

oxides, and lipids with hydroxyl groups at positions as deep as C15 in the chain.17 For cholesterol, hydroperoxides occur at the C4 or C7 positions, commonly resulting in 7-ketocholesterol. When the peroxide radical reacts with an oleate, allylic hydroperoxides form at C8 through C11. The propagation step continues until a chain-breaking antioxidant such as vitamin E stops the reaction or two free radicals annihilate each other to end the chain in a termination step. Many negative consequences result from lipid peroxidation in a biological system. In biological membranes, lipid peroxidation causes loss of fluidity, increased permeability in endothelial cells, and a general change in the biophysical properties of the membrane.18 For instance, the inclusion of oxidatively modified lipids in 1,2-dipalmitoyl-sn-glycero-3phosphocholine monolayers resulted in expanded monolayers (increased surface area per lipid) until concentrations were high enough (>40%) that the oxidatively modified chains were believed to protrude into the aqueous phase.19 Preliminary molecular dynamics (MD) simulations on oxidized lipids have been performed with terminal aldehydes or hydroperoxides.20,21 Increasing the concentration of oxidized lipids up to 50% resulted in an increase in surface area per lipid, and hydroperoxides tended to more readily form hydrogen bonds with water.20 Water permeability increased with the amount of oxidized lipids; the membranes with less than 50% oxidized lipids did not form a water pore. However, at concentration levels of 75−100% aldehyde-modified lipids, a water pore formed with a 30% increase in the surface area (SA) per lipid.21 Our focus in this study was to use MD simulations to understand the changes in the biophysical properties of estermodified phospholipid bilayers.22 Ester-modified phospholipids have ester groups added to their hydrocarbon chains. Menger et al.22 prepared vesicles in aqueous media with 0.15 M NaCl and a NMR shift reagent, which downshifted the 23Na NMR signal by 30 ppm from that within the vesicle, allowing analysis of possible ion exchange. Pure 1-palmitoyl-2-oleoyl-sn-glycero-3phosphocholine (POPC) vesicles showed two peaks: one external and another internal to the vesicle. However, certain ester-modified lipids resulted in a single peak, indicating ion exchange faster than the NMR milliseconds-time scale. The authors suggested some defects occurred that allow Na+ exchange.22 At a critical concentration lower than those at which ions freely exchange, a new Na+ peak appeared. The authors believed that this signal corresponded to Na+ ions trapped within the lipid bilayer on the NMR time scale. Menger et al. hypothesized that initially randomly mixed ester-modified lipids, at certain critical concentrations, assemble into a domain that somehow traps Na+.22 Ester-modified lipids are similar to peroxidized lipids in that both have hydrophilic groups attached to the lipid chains; hence, ester-modified lipids may act as models for certain hydroperoxide lipids. Moreover, as we will demonstrate, these lipids appear to self-aggregate to form pores similar to those in electroporation, but without the need for an induced electric field or charge imbalance. In our MD simulation study, two different types of estermodified lipids (Figure 1) were used to examine the effects of these modifications on bilayer properties. Section 2 presents the methods used to build, equilibrate, and run the MD simulations. Details of the techniques used to analyze these bilayers are also given. Section 3 describes aspects of the overall bilayer structure, including the surface area per lipid, chain order, lateral bilayer density, and component distributions within the bilayer. Self-assembly of the ester-modified lipids and

Figure 1. Chemical structures of MEPC, MGPC, and POPC.

formation of the water pore are also described. The section concludes with a description of the dynamics of ions within the bilayer. An overall description of our findings is given in the discussion and conclusions (section 4) with an emphasis on the mechanism of pore formation and possible implications for peroxidized lipids.

2. METHODS Membrane coordinates for a bilayer of 100 POPC (Figure 1) lipids were generated using CHARMM-GUI Membrane Builder23,24 (see references for membrane building details). To build a mixed bilayer of ester-modified lipids used in previous experiments,22 randomly selected POPC lipids were adapted to these ester-modified lipids. The nomenclature follows that of ref 22 for ester-modified lipids E and G, which are henceforth named modified-E and modified-G PC, i.e., MEPC and MGPC (Figure 1). The CHARMM program25 was used to implement this modification by adding ester groups at various places along the carbon chain of POPC and adjusting the saturation and carbon length. Ester groups were added to the C5 and C9 positions of both chains in MEPC and to the C9 and C16 positions of both chains in MGPC. Three systems, MEPC (30%)/POPC, MGPC (30%)/ POPC, and MGPC (22%)/ POPC, were created with details given in Table 1. A total of 19 Na+ and 19 Cl− ions were added to the system (consistent with experiments at 0.15 M)22 with a hydration of 70 waters/lipid.

Table 1. Details of the Three Membrane Systems Simulated and Simulation Times: MEPC-30 = MEPC (30%)/POPC, MGPC-30 = MGPC (30%)/POPC, and MGPC-22 = MGPC (22%)/POPC no. of atoms no. of lipids no. of waters no. of Cl− and Na+ simulation time [ns]

MEPC-30

MGPC-30

MGPC-22

35 002 100 7048 19 150

34 822 100 7048 19 150

34 758 100 7048 19 200

After the membrane systems were generated in CHARMM-GUI Membrane Builder,23,24 each was minimized for 1000 steepest descent steps and 2000 adopted basis Newton−Raphson steps. The CHARMM36 (C36)26 lipid force field was used with the modified TIP3P water model.27,28 For MEPC and MGPC, ester parameters were transferred from the C36 lipid force field maintaining the same charge and van der Waals interactions. Associated parameters, topologies, and coordinates for these systems are available for 14197

dx.doi.org/10.1021/la403919h | Langmuir 2013, 29, 14196−14203

Langmuir

Article

Figure 2. Snapshots of MGPC(30%)/POPC run 1 over 150 ns time span: water molecules (blue), Na+ ions (red spheres), and Cl− ions (yellow spheres). Snapshots are at (A) 0, (B) 75, and (C) 150 ns. download from http://terpconnect.umd.edu/∼jbklauda/research/ download.html. A smoothing function was used to switch the van der Waals interactions from 8 to 10 Å. The particle-mesh Ewald method29 was used for long-range electrostatic interactions with an interpolation order of 4 and a direct space tolerance of 10−6. All MD simulations were performed using NAMD30 with 2 fs time steps and the SHAKE algorithm. A tetragonal unit cell was used to maintain equal dimensions in X and Y (in the plane of the membrane) while Z varied independently in the NPT ensemble. Langevin dynamics was used to maintain a constant temperature for each system at 294.65 K (to match experiments), while the Nosé−Hoover Langevin-piston algorithm31,32 was used to maintain constant pressure at 1 bar. The production run times are shown in Table 1, and two replicate simulations (different initial velocities) were performed for each system. The visual molecular dynamics (VMD)33 program was used to create images. At the end of each simulation, the output data were analyzed for the SA per lipid based on the box dimensions and the number of lipids per leaflet. The deuterium order parameters (SCD) were obtained from |SCD| =

(Figure S1). Similarly, the MEPC (30%)/POPC bilayers lacked the ability to form a water pore, as the ester groups of each MEPC lipid lie closer to the top of the chain (Figure S2). The average SA per lipid of each system was calculated over time to determine the stability of the bilayers. All bilayers equilibrated quickly from the initial SA per lipid (Figure S3). The two runs for the MEPC (30%)/POPC and MGPC (22%)/ POPC systems maintained approximately equal average SAs per lipid (Table 2). However, the two runs for MGPC (30%)/ Table 2. Surface Area per Lipid [(A2)/Lipid] of Equilibrated Systems with ± Standard Error (from Last 50 ns of the Simulation with 1 ns Blocks To Obtain the Errors)

3 1 cos2 θ − 2 2

system

run 1

run 2

MEPC (30%)/POPC MGPC (30%)/POPC MGPC (22%)/POPC

80.4 ± 0.2 86.7 ± 0.2 81.3 ± 0.2

80.8 ± 0.3 85.5 ± 0.3 80.8 ± 0.2

POPC differed slightly, likely due to the less developed water pore in run 1. There was a ∼5.0 Å2 difference between the MEPC and MGPC bilayers at the same concentration level. The MGPC (30%)/POPC system, having a fully developed water pore, resulted in the highest average SA per lipid (X*Y/ (no. of lipids)). The other two systems, MGPC (22%)/POPC and MEPC (30%)/POPC, did not exhibit water pore formation and consequently had similar average SAs per lipid (Table 2). All values were significantly greater than the SA per lipid of pure POPC bilayers from experiment (68.7 ± 1.5 Å2)37 or simulated using C36 (64.7 ± 0.2 Å2).26 The SCDs of the three systems are compared in Figures 3 and S4. There was an overall decrease in the SCDs of POPC in the mixed bilayers compared to those of pure POPC for the sn-1 chain from experiment and simulated using C3638 (SCDs of pure POPC simulated using C36 match experiment well26). This was consistent with the increase in overall SA per lipid. For the sn-2 chain (Figure S4), the differences between pure POPC and POPC in the mixtures in this study were diminished for the upper half of the chain of MGPC (22%)/POPC. However, the SCDs were significantly lower in the mixed bilayers after the double bond for all lipid mixtures. The SCDs for MGPC and MEPC were significantly lower than those of POPC due to interaction of the ester groups with water and the ability to form chain orientations parallel to the membrane surface. Of the two ester-modified lipid types, MEPC had the higher SCDs, which approached the values for POPC after carbon position 10 (Figure S4). Both concentrations of MGPC

where θ is the angle of a C−H vector with respect to the bilayer normal. SCD are typically reported as absolute values, which is assumed in our abbreviation. The electron density profiles (EDPs) were calculated using a binning method along the bilayer normal,34 assuming minimal lipid membrane undulation for these relatively small bilayer patches. The SIMtoEXP software35 was used to combine the atomic densities for particular chemical groups. The head-to-head distance (DHH) was calculated from the maximum of the density. The bilayer thickness (DB) was calculated from the Gibbs dividing surface of the water volume probability and similarly the hydrophobic thickness (DC) from the dividing surface of the hydrocarbon tail. The bilayer center of mass was placed at z = 0 for the EDP analysis. The water pore radius analysis calculations were done using the program HOLE36 based on defining the center of the hole in the cluster of MGPC lipids. Unless otherwise specified, reported analysis is based on data from 100 ns to the end of the MD simulation.

3. RESULTS 3.1. Overall Bilayer Structure. Both simulations of the MGPC (30%)/POPC bilayer resulted in the formation of a water pore (Figure 2) that fully developed around 100 ns. This pore allowed water to freely flow across the lipid membrane. As shown in Figure 2A, initially water pools (WPs) penetrated the average hydrophobic core of the membrane for each leaflet, but a stable pore did not form until a significant amount of water (larger than a single wire of water) traversed the membrane. In simulations of the MGPC (22%)/POPC bilayer, a significant water pore did not form within the simulation time of 200 ns 14198

dx.doi.org/10.1021/la403919h | Langmuir 2013, 29, 14196−14203

Langmuir

Article

water density was similar for MGPC (22%)/POPC and MEPC (30%)/POPC. The MGPC (30%)/POPC bilayer had the largest DHH and the MEPC (30%)/POPC bilayer had the smallest DHH (Table 3). Both values were significantly smaller Table 3. Measures of Head-to-Head Spacing (DHH), Bilayer Thickness (DB), and Hydrophobic Thickness (2DC) in Å with ± Standard Errora

a

system

DHH

DB

2DC

MEPC (30%)/POPC MGPC (30%)/POPC MGPC (22%)/POPC

32.3 ± 0.2 34.2 ± 0.1 33.0 ± 0.1

32.0 ± 0.3 30.5 ± 0.1 31.0 ± 0.0

23.9 ± 0.0 24.0 ± 0.1 23.9 ± 0.1

Values of zero are smaller than 0.1.

than the experimental DHH value of 37.0 Å for pure POPC bilayers.37 DB, a measure of the Gibbs dividing surface of the water density, followed a trend inverse to that of DHH and was significantly smaller than that of pure POPC (36.8 Å).37 The hydrophobic thickness is invariant to the type of ester-modified lipid and lipid concentration (Table 3). The distribution of the ester groups was distinctly different in each of the three bilayers (Figure 4). MGPC (30%)/POPC resulted in a relatively flat distribution within |z| < 10 Å. Decreasing the amount of MGPC to 22% resulted in distinct peaks of density at z = ±9 Å, but never reaching zero at the midplane of the bilayer. The MEPC (30%)/POPC bilayers had peaks of density at z = ± 10.5−11.0 Å, and the density was nearly zero at the midplane of the bilayer. MEPC lipids lacked the ability to fully distribute the ester group across the hydrophobic region, preventing pore formation. However, MGPC, with its ester groups lower in the lipid chain, was able to distribute its ester groups across the hydrophobic region and even form pores if the concentration of MGPC was sufficiently high. 3.2. Water Pore Formation. As described above, only the MGPC (30%)/POPC bilayers formed a pore large enough to allow significant water transport across the bilayer. This pore was surrounded by an elevated concentration of ester-modified lipids (Figures 5B,C). In the snapshot of the water in run 2 (Figure 5C), Na+ was present at the center of the pore. The stable pore was lined with head groups (Figure 5A), which were solvated with water and sometimes ions. Although the ester groups on MGPC were needed for pore formation, the pores were lined and stabilized by head groups from both MGPC and POPC lipids. At a lower concentration of MGPC (22%), stable pores did not persist. However, short-lived (∼1 ns) connections of WPs in each leaflet enabled water movement across the bilayer (Figure S8). The lipid head groups primarily maintained their normal positions, but to facilitate these shortlived pores, head groups dipped into the inner portion of the membrane. On the 200 ns time scale, no Na+ ions were present at the center of the bilayer. Domain-like formation of the ester-modified lipids can be seen from the lateral view snapshots of the MGPC (30%)/ POPC bilayers (Figure 6A and Figure S9B). This was not limited to this lipid type and concentration level, as shown in Figure S9. MEPC (30%)/POPC formed a weblike structure, while the MGPC (22%)/POPC bilayer formed a mix of webs connected by nodules. The more hydrophilic ester groups tended to self-assemble within each leaflet to minimize contact with the hydrophobic chains.

Figure 3. Chain 1 deuterium order parameters (SCDs) from experiments on pure POPC bilayers38 and simulations of the three systems with MGPC and MEPC. The top panel shows the SCDs for POPC, and the bottom panel shows the SCDs for the MGPC and MEPC lipids. Error bars are smaller than symbol sizes and not shown, which are based on 1 ns block averages.

resulted in similar values for the SCDs with averages of 0.1 and 0.07 for C6−C18 on the sn-1 and sn-2 chains, respectively. The overall and component EDPs are shown in Figure 4 and Figures S5−S7. Clearly, the water density profile of the MGPC (30%)/POPC bilayer differed from those of the other two systems, as the former contained a water pore whereas the other two systems did not. MGPC (30%)/POPC was the only system with significant density at the center of the bilayer. The

Figure 4. Electron density profiles for run 1 of the MGPC (30%)/ POPC (solid), MGPC (22%)/POPC (dashed), and MEPC (30%)/ POPC (dotted) bilayers. The central COO group density is scaled by 3× for clarity. All are averages over the last 25 ns. 14199

dx.doi.org/10.1021/la403919h | Langmuir 2013, 29, 14196−14203

Langmuir

Article

Figure 5. Snapshots from the MGPC (30%)/POPC bilayer with a water pore. (A) Water is in blue, phosphorus is in purple, nitrogen is in green, and lipids are in transparent gray (run 1/150 ns). (B) Water is in blue, Na+ is in yellow, MGPC is in red, and POPC is in gray (run 1/150 ns). (C) Coloring is the same as in part B (run 2/149.3 ns).

Figure 6. (A) Top view of run 2 of the MGPC (30%)/POPC bilayer at 150 ns (primary cell plus portion of +/− in x/y-direction). MGPC is in red and POPC in gray. (B−F) Progression of water pore formation in run 1 at time points of 13.9, 47.3, 93.0, 116.0, and 142.6 ns. Water is in blue, phosphorus is in purple, nitrogen is in green, and lipids are not shown for clarity.

The size of the pore was calculated to further characterize the pore structure. As shown in the snapshots (Figure 6A and Figure S9B), the lipid head groups extended into and limited the size of the pore. For both bilayers, the pore diameter ranged from 5 to 7 Å with the most constricted area occurring near the bilayer’s center of mass. The mechanism for pore formation is shown in Figures 6B− F. Two prepores transiently occurred, with WPs with dipped head groups in a single leaflet (Figure 6B, below red line) or short-lived WWs that pierced the membrane (Figure 6C, highlighted in red). The WPs did not traverse the membrane unless there was an opposing pool on the other leaflet. The ester groups on MGPC facilitated the formation of short-lived WWs, allowing a pathway of water to form across the membrane if opposing MGPC lipids were present in the same corresponding positions in each leaflet. The stable pore did not start to form until 90 ns (Figure 6D); the pore began as a WW, which expanded and allowed lipid head groups to fill the pore (Figure 6E). However, WPs with headgroup protrusions into the membrane were also able to facilitate pore formation. For run 1, between 90 and 125 ns, the pore formed, destabilized, then re-formed again until the pore fully developed

to a diameter of 5−7 Å in size (Figure 6F). Long-term stability of the pore required the lipid head groups to line the pore surface. WWs and WPs also occurred at lower concentrations of MGPC (22%) but never formed extended pore structures as seen in Figures 6D−F. Since pore formation required lateral self-assembly of MGPC lipids, MD simulations should be long enough to reach the time scale to form these aggregates. As described above, pores structures tend to form at ∼100 ns, which corresponds well with the lateral arrangement of MGPC (Figure S10). Before 100 ns, MGPC clusters in a weblike manner and then forms a more globular assembly at 75 ns. This develops into a single ellipsoidal aggregate that expands to allow for WWs and full pores. This is in contrast with MEPC (30%)/POPC, which oscillates between more globular clusters and stretched out weblike structures (Figure S11). This oscillation of lateral selfassembly in MEPC (30%)/POPC and MGPC (22%)/POPC bilayers suggests that our MD simulation time scale of 150−200 ns is sufficient to sample lateral self-assembly of these lipids. 3.3. Dynamics of Ions within the Bilayer. To characterize the dynamics of ions at the center of the MGPC (30%)/ POPC bilayers, the amount of time each ion spent within the 14200

dx.doi.org/10.1021/la403919h | Langmuir 2013, 29, 14196−14203

Langmuir

Article

MGPC (22%)/POPC. However, at this concentration a broad peak was observed, and over a period of 0.5−13 h in a bilayer with 20% MGPC, two peaks formed. This was hypothesized to be Na+ trapped in pockets of aggregated MGPC. We did not observe these stable pockets in our short 200 ns simulations, but we did see aggregation of the MGPC lipids (Figure 6A and Figure S9). MEPC (30%)/POPC and MGPC (22%)/POPC bilayers formed weblike structures of these ester-modified lipids with some minor groupings at connections between the webs. The clustering could potentially accept partially to fully hydrated Na+ ions at the center of the bilayer, but our MD simulations were too short to give qualitative proof. Menger et al.22 hypothesized that ester-modified lipids like MEPC ideally intermix with a typical lipid, in this case POPC, and so special “ester-modified domains” do not form. In our MD simulations, bilayers with MEPC at 30% and MGPC at 22% still formed laterally structured domains (Figure S9). However, these weblike structures cannot form water pores. MEPC contains ester groups in the upper parts of its chains, preventing pores that traverse the membrane. In MGPC (22%) bilayers, the weblike structure appeared more stable, resulting in larger aggregates of this lipid and preventing the formation of a long-term stable pore. Analysis of bilayer properties reveals both differences and similarities between pore-forming and WP-forming lipids. One clear difference is the SA per lipid: MEPC (30%)/POPC and MGPC (22%)/POPC equilibrated at SAs of ∼81 Å2/lipid, whereas MGPC (30%)/POPC equilibrated at a SA of 86 Å2/ lipid. This resulted from structural differences in the bilayer, i.e., water pore formation. This change in SA per lipid did not influence the SCDs to any extent as the POPC SCDs were fairly similar for the sn-1 and sn-2 chains between the two mixed bilayers but differed from those of pure POPC bilayers (Figure 3 and Figure S4). The mixed bilayers exhibited a large increase in SA per lipid compared to the pure POPC bilayer likely due to the ester-modified lipids. The SCDs of these lipids indicate fairly disordered chains (Figure 3 and Figure S4) that result in an effectively larger SA per lipid. The increase in the SA per lipid is consistent with studies of monolayers of similar oxidized lipids, where such lipids increased the lateral monolayer area at oxidized lipid concentrations less than 40%.19 The structure of the water pores and mechanism of their formation in the MGPC (30%)/POPC bilayers are similar to what occurs in electroporation.4,14−16 These ester-modified lipids uniquely induce pore formation without an imposed electric field, charge gradient, or pore-forming peptides. Pores that form from electroporation are shaped like an hourglass with the smallest diameter at the center. These pores are lined with lipid head groups but tend to be larger in diameter compared to those from our simulations. The mechanism for pore formation from electroporation consists of three steps: (1) tilting of lipids, (2) formation of a prepore water wire (WW), and (3) completion of the fully developed pore lined by lipid head groups.14 The mechanism of pore formation in bilayers with estermodified lipids is schematically shown in Figure 7 and is similar to but slightly more complex than the mechanism of pore formation by electroporation. A random mixture of POPC and pore-forming ester-modified lipids, such as MGPC, initiates self-assembly of the ester-modified lipids. This can result in a WW, which would require assemblies of the ester-modified lipids from both leaflets to contact each other, or a WP. WWs and WPs are weakly stabilized by the existence of ester groups

hydrophobic region of the bilayer (|z| < 9.5 Å) was recorded. Figure S12 shows the results for the time tracking of each ion for each run. It is evident for both run 1 and run 2 that Na+ spent more time in the hydrophobic region than Cl−. For run 2, Cl− and Na+ resided in the center of the bilayer for 6.7 ± 1.8 and 16.2 ± 3.3 ps, respectively. However, there were no examples of Na+ or Cl− that translocated across the membrane; i.e., the ions entered and resided in the pore but then returned to their original sides. Although 23Na NMR clearly shows that ions move across the membrane on the submicrosecond time scale,22 our MD simulations do not confirm this. The lack of ion translocation may be due to the known artifact in the CHARMM force field that ions can bind too tightly to lipids;39 it is also possible that longer simulations are needed to observe ion exchange. Evidence from 23Na NMR suggests that sodium can be trapped within the bilayer at concentrations of 22% of MGPC. However, our simulations did not indicate that Na+ became trapped in or even transiently entered the center of the bilayer.

4. DISCUSSION AND CONCLUSIONS Menger et al.22 synthesized several ester-modified lipids and found that bilayers containing some of these lipids allowed exchange of Na+ ions between the inner and outer contents of the vesicle. Our bilayer simulations with MGPC (30%)/POPC mixtures clearly demonstrate why this ester-modified lipid and concentration level result in sodium exchange. At these conditions, MGPC lipids assembled and facilitated formation of a stable pore, allowing water and ions to pass through. MGPC and likely other lipids with a single Na+ signal reported by Menger et al.22 form pores. The ability of these lipids to form pores depends on chain length and ester-modified locations based on the Menger et al.22 study of seven lipid types. The lipids that allowed for ion exchange in experiment include (1) short chains (11−12 carbons) containing a single ester near the center of the chain, (2) medium-length chains (15−16 carbons) containing an ester above and an ester below the center of the chain, and (3) long chains (18−19 carbons) containing an ester near the center and an ester close to the end of the chain. The placement of these ester groups is crucial for pore formation across the membrane, which is clearly shown with the EDPs (Figure 4). MGPC (30%)/POPC exhibited a distribution that was relatively flat and significant across the center of the bilayer. Before the formation of a full pore, the distribution of the ester group in the MGPC (30%)/POPC simulations looked similar to that of MGPC (22%)/POPC in Figure 4 with its distinct peaks away from the center but significant density at the center of the bilayer. The distribution of the ester group of the MEPC (30%)/POPC bilayer had peaks farther from the center and a density that went to zero at the bilayer midplane. Therefore, based on our MD simulations and previous experiments, ester groups need to have significant density at the center of the bilayer to facilitate water pore formation. Other lipids like MEPC can form WPs to some depth in the bilayer but not deep enough to form water pores. Although the location of the ester groups is an important factor in pore formation, the ability to form pores also depends on the concentration of the pore-forming ester-modified lipids.22 At concentrations of MGPC ranging from 5 to 27.5%, two peaks in the 23Na spectra were observed 30 min after preparation of the vesicles, indicating no Na+ exchange and the lack of pores. This agrees with our simulations of 14201

dx.doi.org/10.1021/la403919h | Langmuir 2013, 29, 14196−14203

Langmuir

Article

hydrophilic group only in the upper part of the acyl chain is too great. It is not until there is a large majority of these aldehyde lipids (≥75%) that water pores form. We suspect that higher concentrations of MEPC lipids will also allow for EWWs. If there are enough peroxidized lipids (oxidized at varying positions on the chain) to self-assemble, the positions of these hydroperoxides would be similar to the ester-modified lipids and would likely result in bilayers similar to those in our simulations. Since oxidized lipids also increase membrane permeability and generally change the biophysical properties of the membrane,18 these changes are likely due to water pores similar to those described here. Lipid peroxidation has many negative effects on the cell and cell signaling,18 as the formation of water pores would be detrimental to the primary function of membranes to protect the cell and enclose its contents. In conclusion, our MD simulations have described previous 23 Na NMR results on ester-modified lipids by showing that certain lipid membranes can form water pores. The formation of these pores requires sufficiently high levels of ester-modified lipids to stabilize long-lived pores. Placement of the esters within the lipid chains is also crucial to the formation of EWW structures and consequently water pores. These pore-forming membranes offer a possible new route for liposomal delivery if the permeability can be controlled by physiological conditions, such as pH. Overall, our results not only reveal the mechanism of water pore formation in bilayers with ester-modified lipids, they also suggest that lipid peroxidation likely results in selfassembly of the oxidized lipids, water pore formation, and increased membrane permeability.

Figure 7. Proposed mechanism for water pore formation. RM = randomly mixed, WW = water wire, WP = water pool, EWW = expanded water wire, and P = pore. POPC is in gray, ester-modified lipids are in red, water is in blue, and Na+ is in yellow.

within the center of the bilayer. These short-lived structures can then interconvert or destabilize. The WW or WP can lead to an expanded WW (EWW), which is a larger pore of water without lipid head groups lining the interior. If this EWW remains open, the lipid head groups will begin to tilt and line the pore. Although MGPC lipids predominantly line the pore, POPC lipids also do so. Once this occurs, a fully stable pore (P) with an hourglass shape is formed and allows water and ions to pass. The minimum dimension of this pore is 5−7 Å in diameter. At low concentrations of MGPC, the ester-modified lipids can briefly connect from both leaflets and form a WW or WP. However, only at higher concentrations of MGPC do EWW and P structures result. In bilayers with MEPC and likely other ester-modified lipids with esters closer to the headgroup, a WP can form, but a WW, EWW, or P cannot due to the placement of the ester groups. By facilitating pore formation, ester-modified lipids offer a way to make a vesicle more permeable without the need to induce an electric field. Although they may have potential technological applications, these lipids may also serve as a model for studying the detrimental effects of peroxidized lipids, as there are several similarities between ester-modified lipids and byproducts of free-radical reactions of lipids and cholesterol. As stated in the Introduction, peroxidation of lipids can lead to hydrophilic moieties at the center of the membrane.17 Hydroperoxides form at the C8 to C11 positions on lipid chains with a single double bond and as deep as the C15 position on polyunsaturated lipid chains. 7-Ketocholesterol is a typical byproduct of peroxidation of cholesterol. These two forms of perioxidized lipids are similar to MGPC and MEPC because these contain hydrophilic moieties connected to hydrophobic chains/rings that can further extend into the bilayer. MD simulations of possible hydroperoxides have been performed by Wong-ekkabut et al.20 with 9- or 13-hydroperoxyoctadecadienoic acids. These simulations resulted in an increase in water permeation but lacked a pore formation with hydroperoxide concentrations