Skip to main content
NCBI home page
ACS Omega logoLink to ACS Omega
. 2020 Jul 17;5(29):18080–18090. doi: 10.1021/acsomega.0c01633

Interfacial Water Structure at Zwitterionic Membrane/Water Interface: The Importance of Interactions between Water and Lipid Carbonyl Groups

Hujun Shen †,‡,*, Zhenhua Wu ‡, Xuefeng Zou †
PMCID: PMC7391366  PMID: 32743182

Abstract

graphic file with name ao0c01633_0012.jpg

In this work, atomistic molecular dynamics (MD) simulations of palmitoyl-oleoyl-phosphatidylcholine (POPC) bilayer were carried out to investigate the effect of water models on membrane dipole potential, which is primarily associated with the preferential orientation of molecular dipoles at the membrane–water interface. We discovered that the overestimation of the dipole potential by the TIPS3P water model can be effectively reduced by the TIP4P water model. On the one hand, the TIP4P water model decreases the negative contribution of lipid to the dipole potential through influencing the orientation of lipid headgroups. On the other hand, the TIP4P water model reduces the positive contribution of water to the dipole potential by increasing the preference of H-down orientation (the water dipole orients toward the bilayer center). Interestingly, the TIP4P water model affects the orientation of interfacial water molecules more obviously than that of lipid headgroups, leading to the decrease in the dipole potential. Furthermore, the MD results revealed that the water close to the positively charged choline (namely, N-associated water) prefers the H-down orientation while the water around the negatively charged phosphate (namely, P-associated water) favors the H-up orientation, in support of recent experimental and MD studies. However, interfacial water molecules are more strongly influenced by the phosphate groups than by the choline groups, resulting in the net H-up orientation (the water dipole orients toward the bilayer center) in the region of lipid headgroups. In addition, it is intriguing that the preference of H-up orientation decreases when water molecules penetrate more deeply into the lipid bilayer. This is attributed to the counteracting effect of lipid carbonyl groups, and the effect varies with the lipid chains (oleoyl and palmitoyl chains), suggesting the important role of lipid carbonyl groups.

1. Introduction

Permeation of molecules into cell membranes is a fundamental process in the life cycle of cells. There are many different factors affecting dynamic behaviors of the permeated molecules through cell membranes. Among them, membrane dipole potential, which is defined as the difference of electrostatic potential between the bulk water region (having a high dielectric constant) and the membrane hydrophobic center (having a low dielectric constant), plays a critical role.1,2 The seminal experimental studies of Gawrisch et al.3 have revealed that the dipole potential is primarily associated with the preferential alignment of lipid headgroups and interfacial water molecules and the positive contribution of water to the dipole potential is significant because of the existence of ordered water molecules at membrane interfaces. For instance, in the case of zwitterionic phosphatidylcholine (PC) lipid bilayers, it has been demonstrated that the preferential orientation of lipid headgroups has a counteracting effect on that of interfacial water molecules. Because the negative contribution of lipid molecules to the dipole potential can be overcompensated by the positive contribution of interfacial water molecules, this would lead to a positive dipole potential inside the PC lipid membranes, in which different experiments and molecular dynamics (MD) simulations agree with each other.4−11

Nevertheless, it has been shown that fixed-charge water models tend to overestimate the values of the dipole potential compared to experiments.12−14 The overestimation is attributed to the mean-field treatment of many-body polarization in the membrane–water interfacial region.6,8 In our previous work,15 we attempted to investigate the effect of fixed-charge water models on the dipole potential by carrying out atomistic MD simulations of diphytanyl PC (ether–DPhPC) and diphytanoyl PC (ester–DPhPC) lipid bilayers. We discovered that the water and lipid contributions to the dipole potential vary with the water models employed. This is reasonable because the orientation of lipid headgroups and interfacial water molecules (with respect to the bilayer normal) should be influenced by water models. In addition, our work showed that the overestimation of the calculated dipole potential can be effectively improved using a fixed-charge water model with a better description of electrostatic interactions. However, the correlation between the interfacial water orientation and the dipole potential remains to be elucidated.

Recent progress on vibrational sum frequency generation (VSFG) spectroscopy16−19 enables us to investigate the preferential orientation of interfacial water molecules. Allen et al.20 employed phase-sensitive VSFG (PS-VSFG) spectroscopy to explore interfacial water structure at membrane–water interfaces, demonstrating that the negative charge of the phosphate group would exert a dominant impact on the interfacial water structure and the hydrogen atoms of interfacial water molecules orient toward the lipid tails. Tahara et al.21 used the heterodyne-detected VSFG (HD-VSFG) spectroscopy to investigate the orientation of water molecules at zwitterionic lipid–water interfaces, and they concluded that there are three distinct types of water orientations coexisting at the zwitterionic lipid–water interfaces: (1) the water around phosphate (the negatively charged group) prefers a “H-up orientation” (the hydrogen atoms of water orient toward the hydrophobic center); (2) the water near choline (the positively charged group) favors a “H-down orientation” (the hydrogen atoms of water orient toward the bulk water region); and (3) the water close to the hydrophobic region chooses a H-up orientation. Nevertheless, the orientation of interfacial water molecules is poorly understood at the microscopic level.

Molecular dynamics (MD) simulations were used by Nagata and Mukamel22 to investigate the properties of interfacial water molecules, showing that the orientation of interfacial water molecules is governed by lipid hydrophilic headgroups instead of lipid glycerol groups. Meanwhile, the MD work by Nagata and Mukamel22 suggested that the orientation of near-bulk water molecules is strongly influenced by the surrounding water molecules. Based on all-atom MD simulations, Sugita and co-workers23 reported the existence of a mosaic of interfacial water orientation at a membrane–water interface. The MD study by Sugita et al.23 revealed that the orientation of phosphate-associated water is opposite to that of the choline-associated one, in consistence with the observations revealed by the HD-VSFG experiment.21 Recently, MD simulations carried out by Morita et al.24 demonstrated that the interfacial water molecules associated with phosphate orient toward the lipid side (H-up orientation) and the interfacial water molecules in the vicinity of the choline groups orient toward the bulk water region (H-down orientation), in consistence with the VSFG experiments.20,21 However, as for these MD studies, it is known that the orientation of interfacial water molecules should be influenced by the water models used, which is less well understood.

In this study, we carried out MD simulations of palmitoyl-oleoyl-phosphatidylcholine (POPC) lipid bilayer using two different water models: CHARMM TIP3P (termed TIPS3P)25,26 and TIP4P.27 Please note that the TIPS3P water model implemented for the CHARMM force field28 is a slightly modified version of the traditional TIP3P water model.26 First, we explored the influence of water models on the dipole potential of the POPC bilayer. Our results revealed that the use of the TIP4P water model can considerably improve the overestimation of the dipole potential calculated with the TIPS3P water model. Second, we investigated the effect of water models on the orientation of lipid hydrophilic groups, finding that the orientation of P–N vector (connecting the P atom of phosphate group and the N atom of choline group) is sensitive to the water models used. Finally, we compared the impact of the TIPS3P and TIP4P water models on the orientation of interfacial water molecules. We discovered that the TIPS3P model magnifies the dipole potential relative to the TIP4P model owing to the excessive preference of the H-up orientation (water dipole orients toward the bilayer center).

2. Results and Discussion

2.1. Influence of Water Models on the Dipole Potential

In this work, atomistic MD simulations of POPC lipid bilayer were performed using the CHARMM36 force field with the two different water models (TIPS3P and TIP4P) respectively. Based on the MD simulation trajectories, we constructed the number density profiles for phosphate (PO4) and choline (CHOL) of lipid, which are shown in Figure 1. It appears that the choice of water models has an insignificant influence on the number density of PO4 along the bilayer normal. From the number density profile of PO4 (Figure 1A), it is straightforward to calculate the thickness of the POPC bilayer, which is defined as the phosphate-to-phosphate distance (dHH) in this work. For the comparison purpose, we also employed the grid-based method (GridMAT-MD)29 to determine the bilayer thickness. The calculated results are collected in Table 1, showing that the lipid bilayer thickness varies insignificantly with the water models used here. Meanwhile, it is seen that the experimental value30 is slightly overestimated by the MD simulations.

Figure 1.

Figure 1

Open in a new tab

Number density profiles for the (A) phosphate (PO4) and (B) choline (CHOL) groups of lipid, obtained from the CHARMM36 atomistic simulations of POPC lipid bilayer with two different water models: TIPS3P (black) and TIP4P (red).

Table 1. Physical Properties of the POPC Lipid Bilayer Calculated from the CHARMM36/TIPS3P and CHARMM36/TIP4P Simulations and Comparison between the Simulations and Experimentsa.

  TIPS3P TIP4P experiment
bilayer thickness dHH (nm) 3.91 ± 0.23 3.89 ± 0.22 3.730
3.93 ± 0.34b 3.93 ± 0.34b
area per lipid Ap (nm2) 0.64 ± 0.01 0.64 ± 0.01 0.6431
0.64 ± 0.01b 0.64 ± 0.01b
dipole potential ϕd (V) 0.85 ± 0.05 0.60 ± 0.04 0.3632
Open in a new tab
a

For the comparison purpose, we also employed the grid-based method (GridMAT-MD) to measure the values of bilayer thickness and area per lipid. The experimental value of POPC bilayer thickness was determined by Kučerka et al.,30 that for the area per lipid was obtained by Kučerka et al.,31 and the experimental dipole potential of POPC lipid bilayer was measured by Haldar et al.32

b

Using the grid-based method (GridMAT-MD).29

Piggot et al.33 have demonstrated that the bilayer thickness of POPC varies with the parameters used in the MD simulations, including the type of lipid force field, the size of simulation box, the cutoff value of neighbor list, the size of integration step, and so on. They have shown that the overestimation of bilayer thickness can be improved by decreasing the integration step size to 1 fs. Nevertheless, we found that other physical properties of lipid bilayer are influenced insignificantly by increasing the integration step to 2 fs. Figure 1B shows that the water models used here yield different results for the number density of CHOL. This implies that the preferential orientation of the P–N vector (connecting the P atom of PO4 to the N atom of CHOL) should be influenced by the water models used here, which will be discussed below.

The area per lipid Ap for the POPC bilayer can be calculated based on the following equation

2.1. 1

where Axy represents the area of bilayer surface (x–y plane) and N corresponds to the total number of lipids. Similarly, we also used the GridMAT-MD method to calculate the area per lipid Ap. The obtained results for Ap with the two methods are compared in Table 1, showing that both the methods give the same prediction of Ap and the simulated results nicely match the experimental one. Meanwhile, one can see from Table 1 that the calculated results for Ap are independent of the choice of water models.

The electrostatic potential ϕ(z) along the bilayer normal (z-axis) can be computed through the double integration of local charge density ρ(z)7

2.1. 2

where ε0 represents the vacuum permittivity. Figure 2 shows the electrostatic potential profile constructed from the MD simulations with two different water models. In this work, the dipole potential is defined as the difference of the electrostatic potential between the bilayer center (|z| = 0.0 nm) and the bulk water region (|z| > 3.0 nm). The calculated results are summarized in Table 1. It is seen in Table 1 that the dipole potential calculated with the TIPS3P water model is more than double of that measured from the experiment by Haldar et al.32 However, the calculated dipole potential decreases by 0.25 V when using the TIP4P water model, and this suggests that the overestimation of the dipole potential can be improved by employing a water model with a better description of electrostatic interactions, which is in agreement with our previous work.15

Figure 2.

Figure 2

Open in a new tab

Electrostatic potential profile for the POPC lipid bilayer simulated with two different water models: TIPS3P (black) and TIP4P (red).

Figure 3 illustrates the calculated results for the water and lipid contributions to the electrostatic potential along the bilayer normal of the POPC lipid bilayer. The TIPS3P and TIP4P simulations consistently reveal that the contribution to the dipole potential is positive from water and negative from lipid. However, it is seen in Figure 3 that these contributions are very sensitive to the water models used here. Specifically, the water and lipid contributions are both substantially overestimated by the TIPS3P water model compared to the TIP4P water model. Table 2 shows a comparison between the TIPS3P and TIP4P water models in the calculation of the dipole potential as well as the individual contributions from water and lipid. One can see from Table 2 that the impact of water models is more obvious on the water contribution than on the lipid contribution, explaining why the TIP4P water model can considerably improve the overestimation of the dipole potential by the TIPS3P water model. Meanwhile, the substantial change (more than 1.0 V) in the water and lipid contributions should reflect a change in the orientation of lipid headgroups and interfacial water molecules, which will be discussed in the following section

Figure 3.

Figure 3

Open in a new tab

Electrostatic potential profiles for the individual contributions of (A) water and (B) lipid, obtained from the CHARMM36/TIPS3P (black) and CHARMM36/TIP4P (red) simulations.

Table 2. Dipole Potential Obtained from the CHARMM36/TIPS3P and CHARMM36/TIP4P Simulations, Including the Individual Contributions from Water and Lipid as well as Their Differences.

  TIPS3P TIP4P absolute difference between the two models
contribution of water (V) 4.80 ± 0.05 3.49 ± 0.04 1.31 ± 0.05
contribution of lipid (V) –3.95 ± 0.05 –2.89 ± 0.04 1.06 ± 0.05
total dipole potential (V) 0.85 ± 0.05 0.60 ± 0.04 0.25 ± 0.05
Open in a new tab

2.2. Effect of Water Models on the Orientation of Lipid Headgroups

Figure 4A presents the number density profile for water in the POPC bilayer membrane, in which a comparison is made between the TIPS3P and TIP4P water models. It appears that more TIPS3P water molecules are allowed to penetrate into the membrane–water interface compared to TIP4P water molecules. In addition, the radial distribution functions (RDFs) obtained from the TIP4P simulation are compared to those from the TIPS3P simulation in Figure 4B–D, displaying that the positions of the peaks in the RDFs shift toward larger distances. This is probably due to the different strengths of interactions between lipid headgroups and water molecules, which consequently would influence the orientation of lipid headgroups and interfacial water molecules.

Figure 4.

Figure 4

Open in a new tab

(A) Number density profile for water in the POPC bilayer membrane, and radial distribution functions (RDFs) of (B) g(P–O), (C) g(C21–O), and (D) g(C31–O), obtained from the CHARMM36/TIPS3P (black) and CHARMM36/TIP4P (red) simulations. In the insets in (B, C), P represents the phosphorus atom of phosphate group, C21 represents the carbon atom of oleoyl chain, C31 corresponds to the carbon atom of palmitoyl chain, and O corresponds to the oxygen atom of water.

It has been established that the negative contribution of lipid to the dipole potential strongly correlates with the orientation of lipid headgroups. To measure the orientation of lipid headgroups, we calculated the tilt angle θPN of the P–N vector (see Figure 5A) with respect to the bilayer normal (z-axis). θPN is equal to 0 or 180° if the P–N vector is completely parallel to the bilayer normal, and θPN is 90° when the P–N vector is perpendicular to the bilayer normal. In this work, the N-down orientation is defined as θPN < 90° and the N-up orientation is defined as θPN < 90°, as shown in Figure S1 of the Supporting Information. Figure 5B displays the simulated probability distribution of θPN (normalized with a distribution from random orientation), demonstrating that the N-down orientation (the lipid headgroup dipole orients toward the bulk water region) is preferred over the N-up orientation, and Table 3 reveals the influence of water models on the orientation of lipid headgroup.

Figure 5.

Figure 5

Open in a new tab

(A) Definition of the tilt angles of the P–N and O=C vectors with respect to the bilayer normal (z-axis). Probability distributions of (B) θPN, (C) θOC of the oleoyl chain (unsaturated hydrophobic chain), and (D) θOC of the palmitoyl chain (saturated hydrophobic chain). The angle distributions were normalized with a distribution from random orientation.

Table 3. Percentage of the Population of the Up and Down Orientations, Calculated Based on the Probability Distributions of θPN and θOC Given in Figure 5.

  P–N vector
 O=C vector of oleoyl chain
 O=C vector of palmitoyl chain
  N-down (%) N-up (%) C-down (%) C-up (%) C-down (%) C-up (%)
TIPS3P 68.3 31.7 38.5 61.5 16.7 83.3
TIP4P 63.4 36.6 38.8 61.2 14.4 85.6
Open in a new tab

Meanwhile, the average tilt angle ⟨θPN⟩ of the P–N vector can be calculated based on the probability distribution ρ(θ)

2.2. 3

The final results are given in Table 4, showing that ⟨θPN⟩ is around 70°, and it is very close to the experimental result for the PC headgroup (about 72°).33 In addition, it is seen in Table 4 that ⟨θPN⟩ is increased by about 4° when using the TIP4P water model. Owing to an increase in ⟨θPN⟩ (see Table 4) or a decrease in the population of N-down orientation (see Table 3), the P–N vector would become more parallel to the bilayer surface (x–y plane), consequently decreasing the negative contribution of the lipid to the dipole potential (see Table 2). Furthermore, although the average tilt angle ⟨θPN⟩ of the P–N vector is not greatly influenced by the choice of water models, Table 2 shows that the TIP4P model substantially decreases the lipid contribution by 1.06 V relative to the TIPS3P model. This suggests that the dipole potential should be very sensitive to the orientation of lipid headgroups. The absolute values of headgroup order parameter for POPC have been measured by experiment and MD simulaitons.34,35 The order parameters (SCH) for the POPC headgroup segments (defined in Figure S2 of the Supporting Information) were respectively calculated from the TIPS3P and TIP4P simulations, showing that the CHARMM36 force field with the TIPS3P and TIP4P models can reproduce the experimental order parameters (given in Figure S2 of the Supporting Information). In particular, the order parameters for the β-carbon atom are slightly closer to experiments in the TIP4P simulation than in the TIPS3P simulation, indicating that the lipid headgroup tilt is more realistic in the TIP4P model.

Table 4. Average Tilt Angles <θ> of the P–N Vector and the O=C Vectors with Respect to the Bilayer Normal (z-axis), Calculated Based on the Probability Distributions of θPN and θOC Given in Figure 5.

  P–N vector (deg) O=C vector of oleoyl chain (deg) O=C vector of palmitoyl chain (deg)
TIPS3P 68.5 96.6 115.6
TIP4P 72.4 96.4 117.8
Open in a new tab

It has been shown by Harder et al.6 that lipid ester groups play a less important role in influencing the dipole potential compared to the phosphate and choline groups of lipid. However, it is observed in Figure 4C,D that the strength of interactions between the ester groups and water is differently influenced by the water models employed here. Thus, it would be interesting to investigate how the TIP4P model affects the orientation of O=C vector (given in Figure 5A) when comparing to the TIP3P model. Figure 5C,D illustrates the probability distributions of the tilt angles θOC of the oleoyl and palmitoyl chains. In this study, the C-down orientation is defined as θOC < 90° and the C-up orientation is defined as θOC < 90°, as shown in Figure S3 of the Supporting Information. It is shown in Figure 5 and Table 4 that the choice of water models has a limited impact on the orientation of the carbonyl groups of oleoyl and palmitoyl chains. Meanwhile, one can see from Tables 3 and 4 that the carbonyl groups of oleoyl and palmitoyl chains both favor the C-up orientation (the carbon atom of the carbonyl group orients toward the bilayer center).

2.3. Impact of Water Models on the Orientation of Interfacial Water Molecules

From the MD simulation trajectories, we attempted to investigate the impact of water models on the orientation of interfacial water molecules by measuring the tilt angle θSOL between the water dipole and the bilayer normal (shown in Figure 11A). In our calculations, the tilt angles θSOL (z) of water molecules were computed as a function of the z distance from the bilayer center (see Figure 11B). Similarly, the H-down orientation is defined as θSOL < 90° and the H-up orientation is defined as θSOL < 90°, as shown in Figure S4 of the Supporting Information. Figure 6 illustrates the probability distributions (normalized with a distribution from random orientation) of θSOL in five representative slabs (or at five representative distances), which were calculated from the TIPS3P and TIP4P simulations. Based on the probability distributions, we determined the orientation of water and the percentage of the H-up and H-down populations as a function of the z distance from the bilayer center, given in Figure 7. It is seen that the H-up orientation is more preferred than the H-down orientation in the region from |z| = 1.6 to 2.6 nm. Especially, the strongest H-up orientation is found around |z| = 2.1 nm, where the phosphate and choline groups are located (see Figure 1). This is reasonable because the favorable orientation of the lipid P–N vector is the N-down orientation in this region (see Figure 5 and Table 3) and the preferential N-down orientation would have a counteracting effect on the orientation of interfacial water molecules (illustrated in Figure S5 of the Supporting Information),36,37 leading to the strong H-up orientation in the region of lipid headgroups.

Figure 11.

Figure 11

Open in a new tab

(A) Tilt angle θSOL of water defined as the angle between the water dipole and the bilayer normal (z-axis). (B) The simulation box is divided into a number of slabs (labeled with different names, such as slab0, slab10, slab20, slab30, etc.) with equal thickness (0.1 nm) in the z direction.

Figure 6.

Figure 6

Open in a new tab

Probability distributions of θSOL at five different locations (distances from the bilayer center are 1.2, 1.7, 2.1, 2.5, and 3.0 nm), obtained from the (A) TIPS3P and (B) TIP4P simulations. θSOL is defined as the angle of water dipole with respect to the bilayer normal (given in Figure 11A). The H-down orientation is defined as θSOL < 90°, and the H-up orientation is defined as θSOL < 90°. The distributions were normalized with a distribution from random orientation.

Figure 7.

Figure 7

Open in a new tab

(A) Average orientation of water molecules and (B) the percentage of the H-up and H-down populations as a function of the z distance from the bilayer center, obtained from the TIPS3P (black) and TIP4P simulations. The H-down orientation is defined as θ < 90°, and the H-up orientation is defined as θ > 90°. <θ> represents the average tilt angle, and <θ0> represents the average angle (90°) from the distribution of random orientation.

On the bulk water side of lipid headgroups (|z| > 2.2 nm), one can see that the water orientation gets more random as the z distance increases. This is because the water orientation is strongly affected by bulk water molecules, which orient randomly. On the other side of lipid headgroups (|z| < 2.0 nm), it is observed that the preference of H-down orientation increases as the z distance decreases and the transition occurs between the H-down and H-up orientations. It is intriguing that the carbonyl groups of the oleoyl and palmitoyl chains are located around |z| = 1.5 nm (see Figure 8). This suggests that the switch between the H-up orientation and the H-down orientation should be related to the carbonyl groups. For instance, it is seen in Figure 5 and Table 3 that the carbonyl groups of lipid have a net C-up orientation. The preferential C-up orientation should have an offsetting effect on the neighboring water molecules, causing the preferential H-down orientation in this region (|z| < 1.5 nm), as illustrated in Figure S5 of the Supporting Information.

Figure 8.

Figure 8

Open in a new tab

Number density profiles for (A) the carbon atom C21 of the ester group of the oleoyl chain and (B) the carbon atom C31 of the palmitoyl chain, obtained from the CHARMM36 atomistic simulations of POPC lipid bilayer with two different water models: TIPS3P (black) and TIP4P (red).

It is seen in Figure 7 that the H-up orientation is more preferred by the TIPS3P water than by the TIP4P water. This indicates that the TIPS3P model amplifies the positive contribution of water to the dipole potential compared to the TIP4P model because of the excessive preference of the H-up orientation (water dipole orients toward the bilayer center). In addition, on the lipid tail side of the headgroups (|z| < 1.5 nm), it is shown that the TIP4P water model increases the population of H-down orientation more notably than the TIPS3P model. It is known that the preference of the H-down orientation would reduce the water contribution to the dipole potential.

Based on the first minimum positions of the RDFs given in Figures 4 and S6 of the Supporting Information, interfacial water molecules were divided into different types, shown in Figure S7 of the Supporting Information. For instance, P-associated water is defined as within 0.35 nm (dO–P ≤ 0.35 nm and dO–N > 0.40 nm) of the phosphorus atom of phosphate group, N-associated water as within 0.4 nm (dO–N ≤ 0.40 nm and dO–P > 0.35 nm) of the nitrogen atom of choline group, and PN-associated water as the bridging water between the phosphate and choline groups (dO–P ≤ 0.35 nm and dO–N ≤ 0.40 nm). Note that the P-associated and N-associated waters exclude the bridging waters between the phosphate and choline groups. Figure 9A,B presents the orientation of the N-associated water as a function of the z distance from the bilayer center, showing that more H-down-orientated waters are located on the bulk water side of lipid headgroups (|z| > 2.4 nm) and more H-up-orientated waters on the other side (|z| < 2.4 nm). In contrast, as for the P-associated water, the preferred regions of the H-up and H-down orientations are reversed, shown in Figure 9C,D. This suggests that the phosphate and choline groups should oppositely influence the water structure, in consistence with the MD work by Sugita et al.23 By integrating the populations of the H-down and H-up orientations, we found that the P-associated water favors a net H-up orientation (making positive contribution to membrane dipole potential), while the N-associated water prefers a net H-down orientation (making negative contribution to membrane dipole potential), in agreement with the HD-VSFG experiment performed by Tahara et al.21 It is observed in Figure 9E,F that the PN-associated water shows a similar pattern to that of the P-associated water, which strongly prefers the H-up orientation. This might imply that the negative charge of the phosphate group should have a predominant influence on the orientation of interfacial water molecules in the region of lipid headgroups, in support of the PS-VSFG experiment by Allen et al.20 This is reasonable because water molecules form stronger hydrogen bonds with the negatively charged phosphate groups compared to the positively charged choline group.21 In fact, the oxygen atoms of the phosphate group are the more probable sites to accept hydrogen bonds compared to the hydrogen atoms of the choline group. Furthermore, the influence of the lipid headgroups on the orientation of water should depend on the strength of hydrogen bonds. Thus, owing to the overpreference of H-up orientation at membrane–water interface, the negative contribution of the lipid headgroup dipole can be overcompensated by the positive contribution of the water dipole, leading to the positive dipole potential inside membrane.

Figure 9.

Figure 9

Open in a new tab

Average tilt angle <θ> of (A) N-associated water, (C) P-associated water, and (E) PN-associated water as a function of the z distance from the bilayer center. Percentage of the H-up and H-down populations for (B) N-associated water, (D) P-associated water, and (F) PN-associated water as a function of z distance from the bilayer center. The calculated results were obtained from the TIPS3P (black) and TIP4P simulations. The H-down orientation is defined as θ < 90°, and the H-up orientation is defined as θ > 90°. <θ> represents the average tilt angle, and <θ0> represents the average angle (90°) from the distribution of a random orientation.

Based on the first minimum positions of the RDFs given in Figure 4, OLE-associated and PAL-associated waters are defined as within 0.37 nm of the carbon atoms of oleoyl and palmitoyl chains, as shown in Figure S7 of the Supporting Information. It is seen in Figure 10 that the preference of H-up orientation decreases as the z distance increase. This is reasonable because the preferential C-up orientation (shown in Figure 5 and Table 3) of carbonyl groups would have a counteracting effect on the neighboring water molecules. However, the effect is different because the carbonyl groups of the oleoyl and palmitoyl chains have different orientations. As for the OLE-associated water, the H-down orientation is dominant on the hydrophobic side (|z| < 1.5 nm) while the H-up orientation is favored on the headgroup side (|z| > 1.5 nm), and the TIP4P water has stronger orientation selectivity than the TIPS3P water, as shown in Figure 10A,C. As for the PAL-associated waters, the H-up orientation is still favored by both the TIP4P and TIPS3P models (Figure 10B,D). In Table 5, we compared the preference of the H-up and H-down orientations for all OLE-associated and PAL-associated waters, showing that the H-up orientation is still dominant for the PAL-associated waters but the preference of H-up orientation is weak for the OLE-associated water. This suggests that lipid carbonyl groups are of great importance for the orientation of interfacial water molecules close to the hydrophobic region of lipid, which has been pointed out by Nagata et al.38

Figure 10.

Figure 10

Open in a new tab

Average tilt angle <θ> of (A) OLE-associated water and (B) PAL-associated water, and the percentage of the H-up and H-down populations for (C) OLE-associated water and (D) PAL-associated water as a function of the z distance from the bilayer center. The calculated results were obtained from the TIPS3P (black) and TIP4P simulations. The H-down orientation is defined as θ < 90°, and the H-up orientation is defined as θ > 90°. <θ> represents the average tilt angle, and <θ0> represents the average angle (90°) from the distribution of a random orientation.

Table 5. Percentage of the H-Up and H-Down Populations for All OLE-Associated and PAL-Associated Waters.

  OLE-associated water
 PAL-associated water
  H-down (%) H-up (%) H-down (%) H-up (%)
TIPS3P 43.5 56.5 35.6 64.4
TIP4P 51.5 48.5 40.2 59.8
Open in a new tab

3. Conclusions

In this study, we carried out MD simulations of POPC/TIPS3P and POPC/TIP4P bilayer systems using the CHARMM36 force field. The simulated results for bilayer thickness and area per lipid are comparable to experiment and show an insignificant influence of water models used here. However, the dipole potential calculated with the TIPS3P water model, which is more than double of the experimental one, can be considerably improved by the TIP4P water model. A comparison was made between the TIPS3P and TIP4P water models in the calculations of the individual contributions of water and lipid to the dipole potential, showing the substantial change (more than 1.0 V) in the water and lipid contributions. This suggests that the choice of water models would influence the orientation of lipid headgroups and interfacial water molecules since the dipole potential is sensitive to the alteration in the orientation of molecular dipoles at the interface.

The RDF profiles constructed from the MD simulations revealed the effect of water models on the strength of interactions between lipid headgroups and water molecules. The MD results showed that the TIP4P model increases the average tilt angle ⟨θPN⟩ of the P–N vector and decreases the population of N-down orientation compared to the TIPS3P model, consequently leading to the decrease in the lipid contribution to the dipole potential. In addition, we discovered that the C-up orientation is preferred for the carbonyl groups of both the oleoyl and palmitoyl chains and the choice of water models has a limited impact on the orientation of the carbonyl groups. This suggests that the TIP4P water model decreases the lipid contribution to the dipole potential by mainly influencing the orientation of lipid headgroups.

The preferential N-down orientation of lipid headgroups has a counteracting effect on interfacial water molecules, leading to the preferential H-up orientation of water dipole. However, on the lipid tail side of the headgroups, the preferential H-up orientation of water is reduced owing to the carbonyl groups of the oleoyl and palmitoyl chains, which is C-up-oriented. However, compared to the TIPS3P water model, the TIP4P water model increases the preference of the H-down orientation, which reduces the water contribution to the dipole potential. This suggests that the TIPS3P model enlarges the positive contribution of water to the dipole potential owing to the excessive preference of the H-up orientation.

Finally, our work revealed that the preferential orientation of the P-associated water and the N-associated water are opposite: the former favors the H-up orientation and the latter prefers the H-down orientation. This is consistent with recent experiments and MD studies. Interestingly, we discovered that the choice of water models has a limited impact on the preferential orientation of the P-associated and N-associated waters. In addition, the water models used here have a notable influence on the orientation of the OLE-associated and PAL-associated waters, suggesting that a reliable description of the interactions between water and carbonyl groups of lipid is of great importance for further understanding of the water structure at interface.

4. Methods

4.1. Atomistic MD Simulations

In this work, we carried out atomistic MD simulations of POPC lipid bilayer, in which the two water models (TIPS3P and TIP4P) were used, respectively. Each bilayer system, which contains 128 lipids and 5000 water molecules, was constructed using the PACKMOL program.39 All MD simulations of the POPC lipid bilayers were performed using the CHARMM36 force field28 in the simulation package GROMACS 4.6.7.40 For each simulation, the initial configuration was minimized with the conjugate gradient (CG) method41 and then with the steepest descent method. Subsequently, the minimized bilayer structure was equilibrated for 500 ps under NPT condition and then for 500 ps under NVT condition. Finally, an NPT production run was carried out for 500 ns and the last 400 ns MD simulation was used for final analysis. For all NPT production runs, an integration time step of 2 fs was used since the LINCS algorithm42 was employed for constraining all of the bonds between heavy and hydrogen atoms. The constant temperature of 303 K was controlled using the Nose–Hoover thermostat,43 and the semi-isotropic pressure (z and x–y planes) of 1 bar was maintained by employing the Parrinello–Rahman method.44 Electrostatic interactions were computed with the particle mesh Ewald (PME) algorithm,45 and van der Waals (vdW) interactions were calculated at a cutoff value of 1.4 nm.

4.2. Calculation of the Orientation of Interfacial Water Molecules

To measure the orientation of interfacial water molecules, we calculated the tilt angle θSOL of water, which is defined as the angle between the water dipole and the bilayer normal (z-axis), as shown in Figure 11A. Before the calculation of θSOL, each simulation box was divided into a number of slabs along the z-axis, each of which has an equal thickness (0.1 nm) in the z direction (see Figure 11B). All slabs were labeled with different names, such as slab0, slab10, slab20, slab30, etc. For example, slab10 is located at a distance of 1.0 nm from the bilayer center, slab20 is at the z distance of 2.0 nm from the bilayer center, and so on. Thus, the tilt angles θSOL (z) of water molecules in each slab were, respectively, calculated as a function of the z distance from the bilayer center.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (No. 21863002), the Science and Technology Foundation of Guizhou Province (Nos. QKHJC-[2020]1Y040 and QKHJC-[2016]1109), the start-up fund from the Guizhou Education University, and the construction project for Guizhou Provincial Key Disciplines (No. ZDXK[2015]10). The Shanghai Supercomputer Center (SSC) is gratefully acknowledged for providing the computational resources in MD simulations.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.0c01633.

  • Schematic illustration of the P–N vector orientation, the OC orientation of the oleoyl and palmitoyl chains, orientation of water dipole vector, counteracting effect on the interfacial water molecules from the lipid hydrophilic groups, radial distribution function (RDF) of g(N–O), different types of interfacial water molecules (PDF)

The authors declare no competing financial interest.

Supplementary Material

ao0c01633_si_001.pdf (619.2KB, pdf)

References

  1. Shen H.; Zhao K.; Wu Z. Effects of Ether Linkage on Membrane Dipole Potential and Cholesterol Flip-Flop Motion in Lipid Bilayer Membranes. J. Phys. Chem. B 2019, 123, 7818–7828. 10.1021/acs.jpcb.9b06570. [DOI] [PubMed] [Google Scholar]
  2. Shen H.; Wu Z.; Zhao K.; Yang H.; Deng M.; Wen S. Effect of Cholesterol and 6-Ketocholestanol on Membrane Dipole Potential and Sterol Flip-Flop Motion in Bilayer Membranes. Langmuir 2019, 35, 11232–11241. 10.1021/acs.langmuir.9b01802. [DOI] [PubMed] [Google Scholar]
  3. Gawrisch K.; Ruston D.; Zimmerberg J.; Parsegian V.; Rand R.; Fuller N. Membrane dipole potentials, hydration forces, and the ordering of water at membrane surfaces. Biophys. J. 1992, 61, 1213–1223. 10.1016/S0006-3495(92)81931-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Chiu S. W.; Jakobsson E.; Sott H. L. Combined Monte Carlo and Molecular Dynamics Simulation of Hydrated Dipalmitoyl-Phosphatidylcholine-Cholesterol Lipid Bilayers. J. Chem. Phys. 2001, 114, 5435. 10.1063/1.1349057. [DOI] [Google Scholar]
  5. Saiz L.; Klein M. L. Electrostatic Interactions in a Neutral Model Phospholipid Bilayer by Molecular Dynamics Simulations. J. Chem. Phys. 2002, 116, 3052–3057. 10.1063/1.1436077. [DOI] [Google Scholar]
  6. Harder E.; MacKerell A. D.; Roux B. Many-Body Polarization Effects and the Membrane Dipole Potential. J. Am. Chem. Soc. 2009, 131, 2760–2751. 10.1021/ja806825g. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Shinoda K.; Shinoda W.; Baba T.; Mikami M. Comparative Molecular Dynamics Study of Ether- and Ester-link Phospholid Bilayers. J. Chem. Phys. 2004, 121, 9648–9654. 10.1063/1.1806791. [DOI] [PubMed] [Google Scholar]
  8. Li H.; Chowdhary J.; Huang L.; He X.; MacKerell A. D.; Roux B. Drude Polarizable Force Field for Molecular Dynamics Simulations of Saturated and Unsaturated Zwitterionic Lipids. J. Chem. Theory Comput. 2017, 13, 4535–4552. 10.1021/acs.jctc.7b00262. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Peterson U.; Mannock D. A.; Lewis R. N. A. H.; Pohl P.; McElhaney R. N.; Pohl E. E. Origin of Membrane Dipole Potential: Contribution of the Phospholipid Fatty Acid Chains. Chem. Phys. Lipids 2002, 117, 19–27. 10.1016/S0009-3084(02)00013-0. [DOI] [PubMed] [Google Scholar]
  10. Clarke R. J. The Dipole Potential of Phospholipid Membranes and Methods for its Detection. Adv. Colloid Interface Sci. 2001, 89-90, 263–281. 10.1016/S0001-8686(00)00061-0. [DOI] [PubMed] [Google Scholar]
  11. Wang L.; Bose P. S.; Sigworth F. J. Using Cryo-EM to Measure the Dipole Potential of a Lipid Membrane. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 18528–18533. 10.1073/pnas.0608714103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Siu S. W. I.; Vacha R.; Jungwirth P.; Bockmann R. A. Biomolecular Simulations of Membranes: Physical Properties from Different Force Fields. J. Chem. Phys. 2008, 128, 125103 10.1063/1.2897760. [DOI] [PubMed] [Google Scholar]
  13. Shen H.; Deng M.; Zhang Y. Extension of CAVS CoarseGrained Model to Phospholipid Membranes: The Importance of Electrostatics. J. Comput. Chem. 2017, 38, 971–980. 10.1002/jcc.24770. [DOI] [PubMed] [Google Scholar]
  14. Shen H.; Deng M.; Wu Z.; Zhang J.; Zhang Y.; Gao C.; Cen C. Effect of Cholesterol on Membrane Dipole Potential: Atomistic and Coarse-Grained Molecular Dynamics Simulations. J. Chem. Theory Comput. 2018, 14, 3780–3795. 10.1021/acs.jctc.8b00092. [DOI] [PubMed] [Google Scholar]
  15. Shen H.; Wu Z.; Deng M.; Wen S.; Gao C.; Li S.; Wu X. Molecular Dynamics Simulations of Ether- and Ester-Linked Phospholipid Bilayers: A Comparative Study of Water Models. J. Phys. Chem. B 2018, 122, 9399–9408. 10.1021/acs.jpcb.8b06726. [DOI] [PubMed] [Google Scholar]
  16. Ostroverkhov V.; Waychunas G. A.; Shen Y. R. New Information on Water Interfacial Structure Revealed by PhaseSensitive Surface Spectroscopy. Phys. Rev. Lett. 2005, 94, 046102 10.1103/PhysRevLett.94.046102. [DOI] [PubMed] [Google Scholar]
  17. Stiopkin I. V.; Jayathilake H. D.; Bordenyuk A. N.; Benderskii A. V. Heterodyne-Detected Vibrational Sum Frequency Generation Spectroscopy. J. Am. Chem. Soc. 2008, 130, 2271–2275. 10.1021/ja076708w. [DOI] [PubMed] [Google Scholar]
  18. Mondal J. A.; Nihonyanagi S.; Yamaguchi S.; Tahara T. Structure and Orientation of Water at Charged Lipid Monolayer/Water Interfaces Probed by Heterodyne-Detected Vibrational Sum Frequency Generation Spectroscopy. J. Am. Chem. Soc. 2010, 132, 10656–10657. 10.1021/ja104327t. [DOI] [PubMed] [Google Scholar]
  19. Johnson C. M.; Baldelli S. Vibrational Sum Frequency Spectroscopy Studies of the Influence of Solutes and Phospholipids at Vapor/Water Interfaces Relevant to Biological and Environmental Systems. Chem. Rev. 2014, 114, 8416–8446. 10.1021/cr4004902. [DOI] [PubMed] [Google Scholar]
  20. Chen X.; Hua W.; Huang Z.; Allen H. C. Interfacial Water Structure Associated with Phospholipid Membranes Studied by PhaseSensitive Vibrational Sum Frequency Generation Spectroscopy. J. Am. Chem. Soc. 2010, 132, 11336–11342. 10.1021/ja1048237. [DOI] [PubMed] [Google Scholar]
  21. Mondal J. A.; Nihonyanagi S.; Yamaguchi S.; Tahara T. Three Distinct Water Structures at a Zwitterionic Lipid/Water Interface Revealed by Heterodyne-Detected Vibrational Sum Frequency Generation. J. Am. Chem. Soc. 2012, 134, 7842–7850. 10.1021/ja300658h. [DOI] [PubMed] [Google Scholar]
  22. Nagata Y.; Mukamel S. Vibrational Sum-Frequency Generation Spectroscopy at the Water/Lipid Interface: Molecular Dynamics Simulation Study. J. Am. Chem. Soc. 2010, 132, 6434–6442. 10.1021/ja100508n. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Re S.; Nishima W.; Tahara T.; Sugita Y. Mosaic of Water Orientation Structures at a Neutral Zwitterionic Lipid/Water Interface Revealed by Molecular Dynamics Simulations. J. Phys. Chem. Lett. 2014, 5, 4343–4348. 10.1021/jz502299m. [DOI] [PubMed] [Google Scholar]
  24. Ishiyama T.; Terada D.; Morita A. Hydrogen Bonding Structure at Zwitterionic Lipid/Water Interface. J. Phys. Chem. Lett. 2016, 7, 216–220. 10.1021/acs.jpclett.5b02567. [DOI] [PubMed] [Google Scholar]
  25. Mackerell A. D. Jr; Bashford D.; Bellott R. L.; Dunbrack R. L. Jr; Evanseck J. D.; Field M. J.; Fischer S.; Gao J.; Guo H.; Ha S.; Joseph-McCarthy D.; Kuchnir L.; Kuczera K.; Lau F. T. K.; Mattos C.; Michnick S.; Ngo T.; Nguyen D. T.; Prodhom B.; Reiher W. E. III; Roux B.; Schlenkrich M.; Smith J. C.; Stote R.; Straub J.; Watanabe M.; Wiorkiewicz-Kuczera J.; Yin D.; Karplus M. All-Atom Empirical Potential for Molecular Modeling and Dynamics Studies of Proteins. J. Phys. Chem. B 1998, 102, 3586–3616. 10.1021/jp973084f. [DOI] [PubMed] [Google Scholar]
  26. Bjelkmar P.; Larsson P.; Cuendet M. A.; Hess B.; Lindahl E. Implementation of the CHARMM Force Field in GROMACS: Analysis of Protein Stability Effects from Correction Maps, Virtual Interaction Sites, and Water Models. J. Chem. Theory Comput. 2010, 6, 459–466. 10.1021/ct900549r. [DOI] [PubMed] [Google Scholar]
  27. Jorgensen W. L.; Chandrasekhar J.; Madura J. D.; Impey R. W.; Klein M. L. Comparison of Simple Potential Functions for Simulating Liquid Water. J. Chem. Phys. 1983, 79, 926–935. 10.1063/1.445869. [DOI] [Google Scholar]
  28. 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 the CHARMM All-Atom Additive Force Field for Lipids: Validation on Six Lipid Types. J. Phys. Chem. B 2010, 114, 7830–7843. 10.1021/jp101759q. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Allen W. J.; Lemkul J. A.; Bevan D. R. GridMAT-MD: A Grid-based Membrane Analysis Tool for Use with Molecular Dynamics. J. Comput. Chem. 2009, 30, 1952–1958. 10.1002/jcc.21172. [DOI] [PubMed] [Google Scholar]
  30. Kučerka N.; Tristram-Nagle S.; Nagle J. F. Structure of Fully Hydrated Fluid Phase Lipid Bilayers with Monounsaturated Chains. J. Membr. Biol. 2006, 208, 193–202. 10.1007/s00232-005-7006-8. [DOI] [PubMed] [Google Scholar]
  31. Kučerka N.; Nieh M.-P.; Katsaras J. Fluid Phase Lipid Areas and Bilayer Thicknesses of commonly Used Phosphatidylcholines as A Function of Temperature. Biochim. Biophys. Acta, Biomembr. 1808, 2011, 2761–2771. 10.1016/j.bbamem.2011.07.022. [DOI] [PubMed] [Google Scholar]
  32. Haldar S.; Kanaparthi R. K.; Samanta A.; Chattopadhyay A. Differential Effect of Cholesterol and Its Biosynthetic Precursors on Membrane Dipole Potential. Biophys. J. 2012, 102, 1561–1569. 10.1016/j.bpj.2012.03.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Piggot T. J.; Pineiro A.; Khalid S. Molecular Dynamics Simulations of Phosphatidylcholine Membranes: A Comparative Force Field Study. J. Chem. Theory Comput. 2012, 8, 4593–4609. 10.1021/ct3003157. [DOI] [PubMed] [Google Scholar]
  34. Akutsu H.; Nagamori T. Conformational Analysis of the Polar Head Group in Phosphatidylcholine Bilayers: a Structural Change Induced by Cations. Biochemistry 1991, 30, 4510–4516. 10.1021/bi00232a020. [DOI] [PubMed] [Google Scholar]
  35. Ferreira T. M.; Coreta-Gomes F.; Ollila O. H. S.; Moreno M. J.; Vazb W. L. C.; Topgaard D. Cholesterol and POPC Segmental Order Parameters in Lipid Membranes: Solid State 1H–13C NMR and MD Simulation Studies. Phys. Chem. Chem. Phys. 2013, 15, 1976–1989. 10.1039/C2CP42738A. [DOI] [PubMed] [Google Scholar]
  36. Botan A.; Favela-Rosales F.; Fuchs P. F. J.; Javanainen M.; Kanduc M.; Kulig W.; Lamberg A.; Loison C.; Lyubartsev A.; Miettinen M. S.; Monticelli L.; Maatta J.; et al. Towards Atomistic Resolution Structure of Phosphatidylcholine Headgroup and Glycerol Backbone at Different Ambient Conditions. J. Phys. Chem. B 2015, 119, 15075–15088. 10.1021/acs.jpcb.5b04878. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Bohinc K.; Giner-Casares J. J.; May S. Analytic Model for the Dipole Potential of a Lipid Layer. J. Phys. Chem. B 2014, 118, 7568–7576. 10.1021/jp5050173. [DOI] [PubMed] [Google Scholar]
  38. Ohto T.; Backus E. H. G.; Hsieh C.-S.; Sulpizi M.; Bonn M.; Nagata Y. Lipid Carbonyl Groups Terminate the Hydrogen Bond Network of Membrane-Bound Water. J. Phys. Chem. Lett. 2015, 6, 4499–4503. 10.1021/acs.jpclett.5b02141. [DOI] [PubMed] [Google Scholar]
  39. Martínez L.; Andrade R.; Birgin E. G.; Martínez J. M. PACKMOL: A Package for Building Initial Configurations for Molecular Dynamics Simulations. J. Comput. Chem. 2009, 30, 2157–2164. 10.1002/jcc.21224. [DOI] [PubMed] [Google Scholar]
  40. Hess B.; Kutzner C.; van der Spoel D.; Lindahl E. GROMAS4: Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simulation. J. Chem. Theory Comput. 2008, 4, 435–447. 10.1021/ct700301q. [DOI] [PubMed] [Google Scholar]
  41. Hestenes M. R.; Stiefel E. Methods of Conjugate Gradients for Solving Linear Systems. J. Res. Natl. Bur. Stand. 1952, 49, 409–436. 10.6028/jres.049.044. [DOI] [Google Scholar]
  42. Hess B. P-LINCS: A Parallel Linear Constraint Solver for Molecular Simulation.. J. Chem. Theory Comput. 2008, 4, 116–122. 10.1021/ct700200b. [DOI] [PubMed] [Google Scholar]
  43. Evans D. J.; Holian B. L. The Nose-Hoover Thermostat. J. Chem. Phys. 1985, 83, 4069. 10.1063/1.449071. [DOI] [Google Scholar]
  44. Parrinello M.; Rahman A. Polymorphic Transitions in Single Crystals: A New Molecular Dynamics Method. J. Appl. Phys. 1981, 52, 7182–7190. 10.1063/1.328693. [DOI] [Google Scholar]
  45. Darden T.; York D.; Pedersen L. Particle Mesh Ewald: AnN·log(N) Method for Ewald Sums in Large Systems. J. Chem. Phys. 1993, 98, 10089–10092. 10.1063/1.464397. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ao0c01633_si_001.pdf (619.2KB, pdf)

Articles from ACS Omega are provided here courtesy of American Chemical Society

RESOURCES