Role of Cardiolipins in the Inner Mitochondrial Membrane: Insight

Department of Physics, Tampere University of Technology, P. O. Box 527, FI-33101 Temrpere, Finland, Department of Physical Chemistry, Barcelona Univer...
0 downloads 0 Views 1MB Size
J. Phys. Chem. B 2009, 113, 3413–3422

3413

Role of Cardiolipins in the Inner Mitochondrial Membrane: Insight Gained through Atom-Scale Simulations Tomasz Ro´g,† Hector Martinez-Seara,§ Nana Munck,#,⊥ Matej Oresˇicˇ,⊥ Mikko Karttunen,‡ and Ilpo Vattulainen*,†,∇,¶ Department of Physics, Tampere UniVersity of Technology, P. O. Box 527, FI-33101 Temrpere, Finland, Department of Physical Chemistry, Barcelona UniVersity, Spain, VTT Technical Research Centre of Finland, Espoo, FI-02044 VVT, Finland, Department of Applied Mathematics, UniVersity of Western Ontario, London (ON), Canada N6A 3K7, MEMPHYS-Center for Biomembrane Physics, UniVersity of Southern Denmark, DK-5230 Odense M, Denmark, and Department of Applied Physics, Helsinki UniVersity of Technology, P. O. Box 1000, FI-02015 TKK, Finland ReceiVed: August 30, 2008; ReVised Manuscript ReceiVed: January 20, 2009

Mitochondrial membranes are unique in many ways. Unlike other cellular membranes, they are comprised of two membranes instead of just one, and cardiolipins, one of the abundant lipid species in mitochondrial membranes, are not found in significant amounts elsewhere in the cell. Among other aspects, the exceptional nature of cardiolipins is characterized by their small charged head group connected to typically four hydrocarbon chains. In this work, we present atomic-scale molecular dynamics simulations of the inner mitochondrial membrane modeled as a mixture of cardiolipins (CLs), phosphatidylcholines (PCs), and phosphatidylethanolamines (PEs). For comparison, we also consider pure one-component bilayers and mixed PC-PE, PC-CL, and PE-CL membranes. We find that the influence of CLs on membrane properties depends strongly on membrane composition. This is highlighted by studies of the stability of CL-containing membranes, which indicate that the interactions of CL in ternary lipid bilayers cannot be deduced from the corresponding ones in binary membranes. Moreover, while the membrane properties in the hydrocarbon region are only weakly affected by CLs, the changes at the membrane-water interface turn out to be prominent. The effects at the interface are most evident in membrane properties related to hydrogen bonding and the binding phenomena associated with electrostatic interactions. Introduction Cardiolipins (CLs) constitute a class of lipids that is, in many ways, exceptional. Unlike most of the other lipid types, cardiolipins are divalent anionic lipids with four acyl chains. This unusual structure results from their dimeric nature, where two phosphatidyl moieties are linked together through a central glycerol group.1,2 Consequently, CLs have typically four acyl chains which are usually unsaturated, though the specific fatty acids found in CLs vary depending on the organisms. For example, in the mammalian heart, the fatty acids of CLs are mostly linoleic acid, while in some marine organisms, the most abundant type is docosahexaenoic acid.3 Given these unusual features, the structure of CL is characterized by a large hydrophobic region and a strongly charged relatively small head group, which together imply that CLs favor negative curvature and form different types of aggregates ranging from cylinders to inverted micelles and hexagonal structures.4-6 In biological systems, CLs are typically found in the inner bacterial and mitochondrial membranes with molar concentrations ranging from 5 to 20%.7-10 In addition, although there * To whom correspondence should be [email protected]. † Tampere University of Technology. § Barcelona University. ⊥ VTT Technical Research Centre of Finland. ‡ University of Western Ontario. ∇ University of Southern Denmark. ¶ Helsinki University of Technology. # Deceased.

addressed.

E-mail:

are a number of different anionic lipids in typical eukaryotic cell membranes, CLs are virtually the only charged lipid species in mitochondria.7 That is likely one of the reasons why cardiolipins may form lipid domains in bacterial membranes,11 which in turn are closely related to the membranes of mitochondria. CLs are involved in an exceptionally broad variety of functions, including stabilization of membrane proteins and respiratory complexes12,13 as well as electron and proton transfer.9,13,14 Also, due to their charged nature, CLs are involved in maintaining the electrochemical proton gradient across membranes, which enables ATP synthesis and ADP-ATP translocation.13 CLs also play an important role in programmed cell death,15 aging, and oxidative stress16,17 and in numerous metabolic illnesses such as the Barth syndrome or thyroid dysfunction.18 Moving on, it has been observed that cytochrome c interacts favorably with cardiolipins.19 This may stem from the fact that cardiolipin binding sites have been reported for a number of membrane proteins such as the ADP/ATP carrier, and a cardiolipin molecule has also been described in the structure of the cytochrome bc1 complex.20 Furthermore, there is reason to emphasize that the strongly charged nature of CLs makes them quite different from other lipids; since direct electrostatic interactions between charged species are strong, they can act as both stabilizers and destabilizers of membranes.21-23 Overall, the diverse roles of cardiolipin have been discussed by Schlame et al.24 Although there has been a lot of interest in CL, especially regarding its diverse roles in mitochondria, it is surprising that

10.1021/jp8077369 CCC: $40.75  2009 American Chemical Society Published on Web 02/19/2009

3414 J. Phys. Chem. B, Vol. 113, No. 11, 2009

Ro´g et al.

Figure 1. The molecular structures of (a) CL, (b) PC, and (c) PE molecules including the numbering of atoms. Charge groups have been marked separately.

by now there have been only two published computational studies focusing on CL.5,25 In ref 5, one employed a coarsegrained model to study the different morphological phases of CL derivatives; due to its structure, CL and its derivatives display a very rich phase behavior.5,6 The only reported atomistic simulations of lipid bilayers with cardiolipin were described in ref 25. These simulations showed strong bonding of counterions, especially to cardiolipin molecules. However, despite being

essential for understanding the properties of mitochondrial membranes, detailed atomistic simulation studies of how CL interacts with other lipids are completely lacking. The main purpose of this work is to investigate the detailed mechanisms of how the presence of CL molecules modifies membrane properties and influences membrane structure. That is essential for understanding stability, transport, and signaling

Role of CLs in the Inner Mitochondrial Membrane in mitochondria and is required to extend the current models to include new features such as charge transport involving CLs. The systems that we study here are constructed to reflect the natural lipid composition in the inner membrane of mitochondria.8 The experimental data for mitochondrial membrane structure available for comparison is limited; thus, we concentrate on a comparison with other membrane systems that have recently been investigated through simulations. For the reason of consistency and since this work is the first study of its kind for mitochondrial membranes, the model for the inner membrane is further compared in detail with several reference systems also simulated in the present study. We focus on the behavior at the membrane-water interface; other properties will be discussed in more detail elsewhere. Methods Section We have performed atomic-scale molecular dynamics (MD) simulations of six different membrane systems. Our model is based on the natural composition of the inner mitochondrial membrane,8 50 mol% of phosphatidylcholine (PC), 40 mol% of phosphatidylethanolamine (PE), and 10 mol% of cardiolipin (CL). The following lipid types with diunsaturated hydrocarbon chains were used for PC, PE, and CL: 1. PC (phosphatidylcholine, 2-18:2c9,12) 2. PE (phosphatidylethanolamine 2-18:2c9,12) 3. CL (cardiolipin 4-18:2c9,12) The chosen molar concentrations result in 54 PC molecules, 46 PEs, and 14 CLs, such that the total number of lipids in the bilayer is 114. As reference systems, pure PC (128 molecules), PE (128 molecules), mixed PC-PE (70 and 58 molecules, respectively), PC-CL (100 and 14 molecules, respectively), and PE-CL (100 and 14 molecules, respectively) bilayers were used. As the above indicates, the acyl chains of all lipids were unsaturated, including 18 carbons with double bonds at positions 9 and 12 (linoleic acid, 18:2c9,12). Figure 1 shows the structures and the numbering of the atoms in all lipid molecules used in this study. The charge of cardiolipins has remained somewhat uncertain, and the consensus on this matter remains to be found.22 The evidence seems to indicate, though, that CLs are either singly (-1e) or doubly charged (-2e), depending not only on pH but also on the environmental factors such as the concentration of CLs.22 In this work, we treat cardiolipins as molecules with two negative charges, assuming both acidic sites to be ionized. Considering the time scale of the simulations, about 100 ns, this choice is as justified as other possible ones since we inevitably model a transient state of the membrane. The other scenarios with different charge states will be considered in later studies. All bilayers were hydrated with ∼3500 water molecules. Counterions (Na+) were added to neutralize the negative charge (-2e) of the CL molecules. The initial structures of all bilayers were obtained by arranging the lipid molecules in a regular array in the bilayer (x, y) plane with an initial surface area of 0.32 nm2 per lipid chain. Prior to the actual MD simulations, the steepest-descent algorithm was used to minimize the energies of the initial structures.26,27 All of the simulations were performed using the GROMACS software package28,29 over a time scale of 130 ns. The first 40 ns were considered as an equilibration period, and the remaining period of 90 ns of each trajectory was analyzed. Equilibration was monitored by following the time development of the area per lipid, temperature, and potential energy, which settled to their equilibrium values during 34-36 ns (data not shown).

J. Phys. Chem. B, Vol. 113, No. 11, 2009 3415 TABLE 1: Average Area Per Hydrocarbon Chain and Membrane Thickness membrane

area [nm2]

thickness [nm]

PC PC-CL PC-PE PC-PE-CL PE PE-CL

0.340 ( 0.001 0.318 ( 0.001 0.318 ( 0.001 0.308 ( 0.001 0.299 ( 0.003 0.296 ( 0.001

4.1 4.1 4.2 4.2 4.4 4.3

TABLE 2: Number of Hydrogen Bonds between Any Oxygen in Lipids and Watera lipid membrane

PC

PC PC-CL PC-PE PC-PE-CL PE PE-CL

7.54 7.56 7.58 7.40

PE

CL 13.99 (6.99)b

9.63 9.85 9.66 9.56

13.34 (6.67)b 13.63 (6.81)b

a Errors are less than 0.05. b Number of H-bonds assuming that one CL molecule is equivalent to two PC molecules.

To parametrize the lipid molecules, we used the all-atom OPLS (optimized parameters for liquid simulations) force field.30-33 Partial charges on the PC head group were taken from Takaoka et al.,34 and those of PE were from Murzyn et al.35 Both sets of charges were derived in a manner that is compatible with the OPLS methodology. The same partial charges were used for the phosphate and carbonyl groups of CL, and the standard OPLS charges were used for the hydroxyl group on the glycerol backbone and in the hydrocarbon chains. Each methylene or methyl group was treated as a separate charge group. Charge groups are marked on chemical structures of lipids in Figure 1. In our previous paper, we have shown that this parametrization correctly reproduces the properties of lipid bilayers composed of PC, PE, and glycolipids.36 For water, we employed the TIP3P model, which is compatible with the OPLS parametrization.37 The lipid parameters are available at the web site www.softsimu.org/downloads.shtml. Periodic boundary conditions with the usual minimum image convention were used in all three directions. The LINCS algorithm38 was used to preserve the hydrogen covalent bond lengths. The time step was set to 2 fs, and the simulations were carried out at constant pressure (1 bar) and temperature (310 K), the latter choice implying that the systems are in the fluid phase. The temperature and pressure were controlled using the Nose´-Hoover39,40 and the Parrinello-Rahman41 methods, respectively. The temperatures of the solute and solvent were controlled independently. For pressure, we used semi-isotropic control. The Lennard-Jones interactions were cut off at 1.0 nm, and for the electrostatic interactions, we employed the particle mesh Ewald method42,43 with a real space cutoff of 1.0 nm, β-spline interpolation (order of 6), and direct sum tolerance of 10-6. The list of nonbonded pairs was updated every 10th time step. The simulation protocol used in this study has been successfully applied in many previous MD simulation studies of lipid bilayers.23,36,42,44 Errors were calculated by using the block analysis method as described in ref 45. Results Characteristics of the Membrane Systems. The surface area of membranes in MD studies is usually given per lipid molecule, typically calculated by dividing the total area of the bilayer by

3416 J. Phys. Chem. B, Vol. 113, No. 11, 2009

Ro´g et al. used the distance between points at which the water and membrane densities match as a measure for membrane thickness for all of the systems. The resulting values are given in Table 1. The values obtained are in agreement with the area per chain given (having an inverse relation, as expected). To characterize the order of the acyl chains, the profiles of the molecular order parameter Smol are shown in Figure 3. The order parameter for the nth segment of an acyl chain, Smol, was calculated using47

Smol ) (1/2)〈3 cos2 θn - 1〉

Figure 2. Partial density profiles along the bilayer normal. (a) PC and PE bilayers; (b) PC-PE-CL bilayer. The coordinate z ) 0 corresponds to the membrane center.

the number of lipids in a single leaflet. In this paper, however, the area was calculated by dividing the total surface area by the number of acyl chains in one leaflet (the number of chains per leaflet is constant in all systems). This approach was chosen due to the fact that a CL molecule has four hydrocarbon chains whereas PC and PE molecules have only two. The values obtained for the surface areas per chain are given in Table 1. As the data show, the addition of CL in a PC membrane condenses the system by an amount of 0.02 nm2, while in the PC-PE matrix, the effect is smaller, about 0.01 nm2, and in the PE bilayer, cardiolipins express no observable effect. The observed differences in the surface area imply that the membrane thicknesses are also slightly different. To illustrate this, Figure 2a shows the membrane density of the lipids along the bilayer normal in pure PC and PE bilayers; these two bilayers represent the systems with the largest and the smallest area, respectively. Membrane thickness is not a uniquely defined parameter; instead, different definitions are available. For the reference systems consisting of pure PC or PE bilayers, one could use the distance between the average positions of the phosphate atoms in the opposite layers since that approach is known to work well in single-component systems.46 In mixtures, however, this definition seems to be inadequate. To illustrate this, we calculated the P-P distance for the three lipids separately in the mixed PC-PE-CL system; the distances are 4.30 ( 0.01, 4.10 ( 0.01, and 4.06 ( 0.02 nm for PC, PE, and CL, respectively. This indicates that the CL molecules are located further away from the water-membrane interface region than the PCs and PEs. Figure 2b shows the density profiles of PC, PE, and CL molecules in the mixed PC-PE-CL bilayer. As can be seen, PE and CL are inserted deeper in the membrane than PC. Similar trends are observed in all systems and are likely related to the sizes of the polar head groups; larger head groups (PC) tend to protrude more toward the water phase than small ones (CL). Similarly, PEs, with head groups smaller than those of PCs, are located deeper in the membrane than PCs and more in the water phase than CLs. Due to the above difficulties, we

(1)

where θn is the instantaneous angle between the nth segmental vector, that is, the (Cn-1, Cn+1) vector linking n - 1 and n + 1 carbon atoms in the acyl chain and the bilayer normal; 〈...〉 denotes both ensemble and time averages. As Figure 3 shows, the presence of CL molecules increases the order of the PC chains but has practically no influence on the PE acyl chains. This is in agreement with the changes in the surface area. In the PC-PE-CL mixture (Figure 3d), the ordering of the chains of all lipid species is close to each other, though PCs are slightly more ordered than PEs and CLs and PEs are somewhat more ordered than CLs. These small differences are in agreement with the slightly different positions of these lipids along the bilayer normal (Figure 2b). Head Group Dynamics. The rotational autocorrelation functions (RACFs) of the P-N vectors of PCs and PEs are shown in Figure 4. The effect of CLs on the head group dynamics is clear from Figure 4; CLs restrict the rotation of both the PC and PE head groups. The effect is, however, much stronger for PCs. It is interesting that there is no strong correlation between the rotation of the P-N vector and the surface area accessible for the PC and PE head groups. Although one can find that rotation rates follow trends observed for surface area in the given bilayers, a similar change of area can lead to very large or moderate effects. The lack of simple correlation results likely from the intermolecular interactions whose number and strength depend on the particular composition (discussed in more detail in the following sections). Interaction between Head Groups and Water. The number of hydrogen bonds (H-bonds) between water and any H-bond acceptor or donor atom of lipid molecules, the number of water bridges between lipid head groups, and the number of nearestneighbor water molecules around lipid oxygens are given in Tables 2-4, respectively. In addition, the number of water molecules hydrating the choline groups is given in Table 4 for the PC molecules. For these purposes, we employed the same geometrical definitions as that in our previous papers.25,48 The number of H-bonds between the water molecules and lipids is similar in all bilayers; the observed differences are in the range of 2-5%. The differences between the lipid species are likely related to the head group positions along the bilayer normal (CLs are located deeper in the membrane and are, thus, less hydrated) and the differences between the surface areas of the membrane systems (larger areas correlate with higher hydration). On average, a PC binds about 7.5, a PE 9.5 (including H-bonds with the amino group), and a CL about 13.5 water molecules. The number of water molecules hydrating the choline groups is similar in each of the bilayers, about 11 water molecules per PC. Water bridges, that is, water molecules simultaneously H-bonded to two lipid molecules, are most common between PE molecules. The presence of CLs, and especially of PCs, decreases the number of water bridges (Table 3).

Role of CLs in the Inner Mitochondrial Membrane

J. Phys. Chem. B, Vol. 113, No. 11, 2009 3417

Figure 3. Profiles of the molecular order parameter (Smol) calculated for sn - 1 chains of (a) PC, (b) PE, and (c) CL molecules, comparison in different bilayers. (d) PC, PE, and CL molecules in the PC-PE-CL bilayer. Results for segments 2-3 are not included due to the carbonyl group; the description in that case is not well defined.

Figure 4. Rotational autocorrelation function of the P-N vector of (a) PC and (b) PE molecules in the different bilayers considered.

To describe the dynamics of hydration, we calculated the autocorrelation function for hydrogen-bonded water as

ACF(t') ) 〈hw(t + t')hw(t)〉 / 〈hw(t)〉

(2)

where the function hw(t) adopts the value of unity if a water molecule is bonded with any lipid molecule and zero if not. This analysis follows the approach of Rapaport,49 Chandra,50 and Balasubramanian et al.51 for the dynamics of hydrogen bonding. The computed autocorrelation functions shown in Figure 5 indicate that the presence of CLs stabilizes water bonding. PEs are found to bond the water molecules stronger than PCs.

Direct Hydrogen Bonding. Unlike PCs, both PEs and CLs are capable of participating in H-bonding as hydrogen donors due to the presence of an ammonium group in PE and a hydroxyl group in CL. The numbers of H-bonds between PEs and other lipids are given in Table 5. The highest number of H-bonds per PE molecule is observed in the mixed PC-PE-CL bilayer, which is surprising since in a binary mixture of PEs with PCs or CLs, the number of H-bonds decreases or remains constant, in respective order. This observation emphasizes the fact that the properties of ternary mixtures cannot be simply deduced from simpler systems. PE molecules also form a significant number of intramolecular H-bonds, and their number follows the values of the corresponding surface areas (when the area is larger, the head groups are more extended, and the number of intramolecular H-bonds is lower). To better characterize the H-bonding patterns in different bilayers, the percentage of H-bonds per each oxygen type in the acceptor molecule, with PEs as the donor, is given in Table 6. We find that the phosphate oxygens O13 and O14 are the predominant H-bond acceptors (80-90% of H-bonds). H-bonds between the ammonium groups and the carbonyl oxygens (O22 and O32) are rare. This pattern is practically the same for all membrane and lipid types. The CL hydroxyl group does not participate in intermolecular H-bonding (Table 7). This group is mainly involved in interactions with water and rarely in intramolecular bonding. To better describe the dynamics of H-bonds between lipids, we calculated a number of relevant autocorrelation functions similarly as in the case of the above-discussed hydration (see eq 2). The autocorrelation functions shown in Figure 6 depict that the same H-bond-lipid pair can express different behavior depending on the system; the behavior of PE-PC pairs in the PE-PC bilayer is different from their behavior in the PEPC-CL bilayer. As a general trend, one finds that the PE-PE bonds are the most stable ones, the next strongest being the PE-CL bonds, while the PE-PC bonds are the least stable. Charge Pairing. While PC molecules do not participate in H-bonds as hydrogen donors, they can be connected via charge pairs due to electrostatic interactions between the positively charged choline groups and the negatively charged oxygen

3418 J. Phys. Chem. B, Vol. 113, No. 11, 2009

Ro´g et al.

TABLE 3: Number of Intermolecular Water Bridges between Lipid Moleculesa pairs membrane PC PC-CL PC-PE PC-PE-CL PE PE-CL a

PC-PC per PC

PC-PE per PC (PE)

1.02 1.15 0.58 0.48

0.74 (0.89) 0.35 (0.42)

PC-CL per PC (CL)

PE-PE per PE

PE-CL per PE (CL)

CL-CL per CL

0.00 (0.01) 0.02 (0.06)

0.24 1.96 1.57 4.46 3.81

0.51 (1.67)

0.08

0.02 (0.16)

0.27

Errors are on the order of or less than 5%; the results for the two different leaflets were also consistent within margins of error.

TABLE 4: Number of Nearest-Neighbor Water Molecules around Lipid Oxygens Per Lipida lipid membrane

PC

PC PC-CL PC-PE PC-PE-CL PE PE-CL

7.24/11.6 6.75/11.3 7.03/11.2 6.28/10.9

PE

CL 12.10

8.00 7.49 7.43 7.11

11.78 11.25

a

For PC, also the number of water molecules hydrating the choline group is included. Errors are less than 0.04 (and 0.1 in the first column for the latter numbers shown).

Figure 5. Autocorrelation function of water bonding with the membrane surface in PC, PE, PC-PE, PC-CL, PE-CL, and PC-PE-CL bilayers.

atoms.52 The numbers of inter- and intramolecular charge pairs given in Table 8 show that the number of charge pairs per PC molecule is reduced in mixed bilayers as compared to that in the pure PC bilayer. The strongest reduction is observed in the three-component bilayer and is not compensated by new charge pairs with other lipids. The trend regarding the total number of intermolecular charge pairs can be rationalized by its inverse relation with the P-N angle with respect to the membrane normal direction calculated separately for both leaflets (the average P-N angles are 99.7° for PC, 102.5° for PC-CL, 105.9° for PC-PE, and 108.9° for the PC-PE-CL system). That is, the deeper the -N(CH3)3 group penetrates to the water phase, the less accessible are the negatively charged oxygens to form charge pairs. In the case of bilayers including PEs, another contribution arises from the competition for donors with stronger H-bonds. Strong preference for charge pairs between PCs and CLs is observed, especially in the three-component bilayer. To show the patterns of charge pairing in different bilayers, the percentages of charge pairs per each oxygen type in the acceptor molecule are given in Table 9. The main acceptors of charge pairs are the phosphate oxygen atoms O13 and O14 and the carbonyl oxygen atoms O22 and O32.

To describe the dynamics of charge pairs, we have calculated the autocorrelation functions as in the cases of hydration and H-bonding between lipids. The autocorrelation functions shown in Figure 7 illustrate that charge pairs are more stable in mixed bilayers, and in particular, the presence of CL increases the stability of charge pairs. The most stable ones are charge pairs between PCs and CLs, while the stability of PC-PC and PC-PE charge pairs is similar. Interaction with Ions. In order to preserve overall charge neutrality, cationic counterions (Na+) were added to the systems to compensate for the negative charge of the CL molecules. In agreement with previous simulations of charged lipids,23,53 the sodium counterions were strongly bonded with lipid head groups. In our analysis, we considered the ions bonded with lipid oxygens if the distance between them was less than 0.3 nm, the position of the first minimum in the radial distribution functions of sodium ions with respect to the lipid oxygen atoms (data not shown). An ion is considered to be bridging two lipids if it is simultaneously bonded with their oxygen atoms. The numbers of ions bonded with lipids are given in Table 10, and the number of ion bridges is given in Table 11. Both tables bring about the view that the sodium ions are preferably bonded with CLs due to their net charge and, hence, create numerous bridges between CLs and PEs. Discussion and Conclusions In this article, we have built a detailed atomistic model for inner mitochondrial membranes containing cardiolipin, PC, and PE molecules and characterized its properties by large-scale atomistic molecular dynamics simulations. Pure and mixed PE and PC bilayers and bilayers with CLs mixed with either PEs or PCs were used as reference systems. The number of reference systems was larger than usual since we wanted (a) to characterize the interactions of CLs in typical two-component systems and (b) to study a corresponding ternary system to see which, if any, of its properties can be deduced from those of the twocomponent systems. Also, since this is the first atom-scale simulation study of membranes including cardiolipins and the number of experimental model membrane studies with CL is very limited, we preferred to focus on trends that result from cardiolipins in particular. The results show that the effect of CL on all of the three matrices (PC, PE, and mixed PE-PC) was different and cannot be easily derived from the effects of CL on either of the PC or PE systems alone. The structure of a CL molecule is unusual, and its effect on the interactions seems to be nonadditive. This feature is particularly strong in the case of PCs. The addition of CL molecules into a PC bilayer leads to a small condensation in the surface area and a related increase in the order of acyl chains. This result agrees with other computational and experimental data, which show a decrease in membrane permeability

Role of CLs in the Inner Mitochondrial Membrane

J. Phys. Chem. B, Vol. 113, No. 11, 2009 3419

TABLE 5: Average Number of Hydrogen Bonds between the PE Ammonium Group and Other Lipids Per PE (PC/CL) Moleculea membrane composition (mol%)

PE-PE intramol

PE-PE intermol

PE 100% PE-CL 78:22% PC-PE 55:45% PC--PE-CL 42:36:22%

1.78 1.63 1.61 1.39

2.97 [100%] 2.67 [88%] 1.80 [70%] 1.37 [36%]

a

PC-PE

PE-CL

total inter 2.97 3.03 2.69 3.44

0.36 (2.59) [12%] 0.89 (0.74) [30%] 1.56 (1.33) [45%]

0.51 (1.67) [19%]

The percentage of the total intermolecular number of H-bonds is given in square brackets. Errors are less than 0.05.

TABLE 6: Percentage of Intermolecular Hydrogen Bonds between the Ammonium Group of PE and Individual Acceptor Oxygen Atomsa pairs

PE-PE

PC-PE

PE-CL

bilayer

PE

PE-CL

PC-PE

PC-PE-CL

PC-PE

PC-PE-CL

PE-CL

PC-PE-CL

O13 O14 O22 O32 O12 O11 O21 O31

41 41 1 1 8 8 0 0

39 41 3 1 6 10 1 0

42 36 2 1 9 9 1 0

44 42 2 0 3 7 2 0

43 42 3 3 4 4 1 0

39 38 4 1 9 9 0 0

45 45 1 2 2 5 0 0

40 40 2 2 9 7 0 0

a

Errors are less than 2 on the same scale.

TABLE 7: Average number of hydrogen bonds between CL hydroxyl groups and lipids per CL molecules. Errors are less than 0.02. membrane

CL-CL intramol

CL-CL intermol

PC-CL PE-CL PC-PE-CL

0.21 0.18 0.29

0.00 0.01 0.00

CL-PC

CL-PE

0.00

0.00 0.00 0.00

and increased stability when CL is present in PC bilayers.54 At the interfacial region, we observed PCs to form charge pairs preferentially with CLs rather than with other lipids, 0.84 PC-CL charge pairs per PC. These pairs are also considerably more stable than PC-PC pairs and are likely to be responsible for the observed slowing down of the rotational motion of the PC head groups. The total number of charge pairs is, nevertheless, lower in a mixed than that in a pure bilayer, though it is

compensated by bonding via counterions. It should also be mentioned that here, we used sodium as the counterions. Multivalent counterions such as Ca2+ have been demonstrated to be able to destabilize CL membranes, which can be stabilized again by addition of POPC.54,55 The rich morphological behavior4-6 and the sensitivity of membranes containing CLs have been suggested to have wide biological consequences, for example, in ion transport and regulation of the transmembrane voltage as well as formation of complexes particularly in mitochondria.55,56 Complex formation, possibly mediated by ions, may be an important factor in mitochondrial apoptosis as it involves transfer of cardiolipins from the inner to the outer membrane.57 Work is in progress to address the effects of multivalent ions. In the case of PE-based bilayers, the CL molecules had, surprisingly, virtually no influence on membrane condensation

TABLE 8: Number of Inter- And Intramolecular Charge Pairs Per PC (PE or CL) Molecule in PC, PC-CL, PC-PE, and PC-PE-CL bilayersa membrane % composition

PC-PC intramol

PC-PC intermol

PC 100% PC-CL 78:22% PC-PE 55:44% PC-PE-CL 42:36:22%

0.34 0.32 0.32 0.25

3.01 [100%] 2.04 [72%] 1.56 [69%] 0.64 [32%]

a

PC-PE

PC-CL

total 3.01 2.83 2.25 2.01

0.79 (5.67) [28%] 0.69 (0.84) [31%] 0.46 (0.54) [23%]

0.91 (3.51) [45%]

The percentage of the total intermolecular number of charge pairs is given in square brackets. Errors are less than 0.04.

TABLE 9: Percentage of Intermolecular Charge Pairs between the Choline Group of PC and Individual Acceptor Oxygen Atomsa PC-PC O13 O14 O22 O32 O12 O11 O21 O31 a

PC-PE

PC-CL

PC

PC-CL

PC-PE

PC-PE-CL

PC-PE

PC-PE-CL

PC-CL

PC-PE-CL

22.5 21.6 17.7 14.6 8.7 11.9 2.0 0.9

21.8 20.1 18.9 15.9 9.0 11.3 1.9 0.9

23.8 20.5 18.4 13.8 9.1 11.6 1.9 1.0

19.7 23.9 13.9 20.0 10.7 9.0 1.8 1.1

20.8 18.2 22.7 18.6 4.6 12.6 1.5 1.0

14.1 15.9 18.3 36.9 2.3 9.1 1.6 1.8

20.4 23.7 14.0 15.8 11.8 12.1 1.8 0.3

23.5 22.1 13.5 13.6 13.2 12.4 1.4 0.3

Errors are less than 0.6.

3420 J. Phys. Chem. B, Vol. 113, No. 11, 2009

Ro´g et al.

Figure 6. Autocorrelation function describing the dynamics of H-bonds; (a) PE-PE hydrogen bonds, (b) PE-PC H-bonds, and (c) PE-CL H-bonds (comparison in the panels a-c is for the lipid pairs in different bilayers); (d) H-bonds in the PE-CL bilayer, (e) H-bonds in the PC-PE bilayer, and (f) H-bonds in PC-PE-CL bilayer (where the panels d-f show a comparison of the different lipid pairs present in mixed bilayers).

TABLE 10: Number of Bonded Ions Per Lipid Moleculesa

a

membrane

PC

PC-CL PC-PE-CL PE-CL

0.28 0.39

PE

CL

0.56 0.28

0.97 0.64 0.70

Errors are less than 0.03.

or hydrocarbon chain order. H-bonds between PEs and CLs were observed to be less frequent and less stable than those between PE molecules (there should be 0.66 per PE, and we observed only 0.36). The total number of H-bonds is similar in pure and mixed bilayers, and head group rotation is almost unaffected, which suggests that the intramolecular H-bonds dominate over intermolecular ones. The only additional interactions at the water-membrane interface are ion bridges. This behavior is very different from that observed in a mixture of anionic POPG (phosphatidylglycerol) and POPE, where the observed significant (30%) increase in hydrogen bonding is mostly due to H-bonds between the ammonium groups of PEs and phosphodiester oxygens of PGs.53 The behavior of mixed PC-PE bilayers is very interesting as PC and PE molecules influence each other’s properties. This

can be observed from the fact that the observed surface area of 0.635 ( 0.001 nm2 is lower than the simple average of 0.643 nm2 for a mixture of 70 PC and 58 PE molecules. The number of PE-PE hydrogen bonds (1.80) is higher than the number of PE-PC hydrogen bonds (0.89), and the former are also more stable. Also, PC-PC charge pairs are more numerous than the PC-PE ones, but their lifetime is longer; see Figure 7e. When CL is added to a mixed PC-PE bilayer membrane, the surface area remains unaffected, while the interface region becomes less dynamic; rotations of both PC and PE head groups are slowed down. Rotations of the PC head groups in the PC-PE-CL bilayer are the slowest among all three bilayer systems, while the rotational motions of the PE head groups are similar to those found in pure PE and PE-CL bilayers. In agreement with experiments,58 we find that the tilt of the P-N vector of PCs in the ternary systems increases by 9° as compared to a pure PC bilayer. In the experiments of Pinheiro et al., the change in the P-N dipole orientation was ∼6° and increased for increasing CL concentration.58 In comparison with the mixed POPG-POPE bilayer,53 the addition of CL in a PE-PC bilayer leads to more complex

TABLE 11: Number of ion bridges per lipid molecule. Errors are less than 0.02 PC-CL PC-PE-CL PE-CL

PC-PC per PC

PC-PE per PC (PE)

PC-CL per PC (CL)

0.21 0.12

0.35 (0.41)

0.00 (0.00) 0.00 (0.00)

PE-PE per PE

PE-CL per PE (CL)

CL-CL per CL

0.30 0.23

0.19 (0.62) 0.10 (0.70)

0.16 0.00 0.14

Role of CLs in the Inner Mitochondrial Membrane

J. Phys. Chem. B, Vol. 113, No. 11, 2009 3421

Figure 7. Autocorrelation function describing the dynamics of charge pairs; (a) PC-PC pairs, (b) PC-PE, (c) PC-CL, (d) pairs in the PC-CL bilayer, (e) pairs in the PC-PE bilayer, and (f) pairs in the PC-PE-CL bilayer.

behavior which cannot be deduced from the binary systems or the behavior of the mixture of anionic POPG and POPE. At the interaction level, we observed that the PE-PE H-bonds are more common and stable than the PE-PC and PE-CL ones and that the PC-CL charge pairs are the most frequent and stable. It should be stressed that in the ternary systems, the total number of H-bonds is the highest among all membranes, while the number of charge pairs is the lowest. This indicates stronger interlipid interactions in the PC-PE-CL bilayer than in any of the other bilayers. In conclusion, we observed that the effects of CL in ternary membrane systems are complex and cannot be easily deduced from its behavior or the behavior of other anionic lipids in binary mixtures. CL also has a tendency to form complex morphological structures which can be stabilized by, for example, PCs.54,55 Interestingly, CLs had virtually no effect on PEs in a binary mixture. These properties are also likely important in the role of CLs in mitochondria since the stabilization-destabilization of the inner membrane is one of the most important issues in mitochondrial apoptosis. It would be of interest to study CLs with varying chain lengths to investigate how the asymmetry in lengths affects its behavior. In nature, both in humans and other organisms, there is remarkable uniformity in hydrocarbon chain lengths. Significant compositional variations of CLs have been observed in the cases of serious diseases such as the Barth Syndrome59 and, to a lesser extent, in early stages of diabetes.60 Acknowledgment. This work was carried out under the HPC-EUROPA Project (RII3-CT-2003-506079), with the support of the European Community - Research Infrastructure

Action under the FP6 “Structuring the European Research Area” Programme. Computational resources were provided by the Barcelona Supercomputing Center, The Finnish IT Centre for Science (CSC), and the SharcNet grid computing facility (www.sharcnet.ca). We wish to thank the Academy of Finland, the Emil Aaltonen Foundation, and the Natural Sciences and Engineering Research Council of Canada (NSERC) for financial support. Partial financial support was provided by SEID through the Project FIS200603525 and by DURSI through the Project 2005-SGR000653. Note Added after ASAP Publication. This article was published ASAP on February 19, 2009. Due to a production error, one of the authors was omitted in the original version. The corrected version was reposted on February 23, 2009. References and Notes (1) Pangborn, M. C. J. Biol. Chem. 1942, 142, 247. (2) Le Cocq, J.; Ballou, C. E. Biochemistry 1964, 3, 976. (3) Kraffe, E.; Soudant, P.; Marty, Y.; Kervarec, N.; Jehan, P. Lipids 2002, 37, 507. (4) Alessandrini, A.; Valdre, G.; Valdre, U.; Muscatello, U. Chem. Phys. Lipids 2007, 146, 111. (5) Dahlberg, M. J. Phys. Chem. B 2007, 111, 7194. (6) Powell, G. L.; Marsh, D. Biochemistry 1985, 24, 2902. (7) Hovius, R.; Lambrechts, H.; Nicolay, K.; DeKruijff, B. Biochim. Biophys. Acta 1990, 1021, 217. (8) Daum, G. Biochim. Biophys. Acta 1985, 822, 1. (9) Hoch, F. L. Biochim. Biophys. Acta 1992, 1113, 71. (10) Gomez, B., Jr.; Robinson, N. C. Anal. Biochem. 1999, 267, 212. (11) Matsumoto, K.; Kusaka, J.; Nishibori, A.; Hara, H. Mol. Microbiol. 2006, 61, 1110.

3422 J. Phys. Chem. B, Vol. 113, No. 11, 2009 (12) Krause, F.; Reifschneider, N. H.; Vocke, D.; Seelert, H.; Rexroth, S.; Dencher, N. A. J. Biol. Chem. 2004, 279, 48369. (13) Lenaz, G.; Genova, M. L. Am. J. Physiol. Cell. Physiol. 2007, 292, 1221. (14) Gomez, B.; Robinson, N. C. Anal. Biochem. 1999, 267, 212. (15) Gonzalvez, F.; Gottlieb, E. Apoptosis 2007, 12, 877. (16) Sanz, A.; Pamplona, R.; Barja, G. Antiox. Redox Signal. 2006, 8, 582. (17) Sastre, J.; Pallardo´, F. V.; Garcı´a De La Asuncio´n, J.; Vin˜a, J. Free Radical Res. 2000, 32, 189. (18) Schlame, M.; Ren, M. FEBS Lett. 2006, 580, 5450. (19) Pinheiro, T. J. T.; Watts, A. Biochemistry 1994, 33, 2459. (20) Palsdottir, H.; Hunte, C. Biochim. Biophys. Acta 2004, 1666, 2. (21) Shoemaker, S. D.; Vanderlick, K. T. Biophys. J. 2002, 83, 2007. (22) Nichols-Smith, S.; Kuhl, T. Colloids Surf., B 2005, 41, 121. (23) Zhao, W.; Ro´g, T.; Gurtovenko, A. A.; Vattulainen, I.; Karttunen, M. Biophys. J. 2007, 92, 1114. (24) Schlame, M.; Rua, D.; Greenberg, M. L. Prog. Lipid Res. 2000, 39, 257. (25) Dahlberg, M.; Maliniak, A. J. Phys. Chem. B 2008, 112, 11655. (26) Murzyn, K.; Ro´g, T.; Jezierski, G.; Takaoka, Y.; PasenkiewiczGierula, M. Biophys. J. 2001, 81, 170. (27) Patra, M.; Karttunen, M.; Hyvo¨nen, M. T.; Falck, E.; Lindqvist, P.; Vattulainen, I. Biophys. J. 2003, 84, 3636. (28) van der Spoel, D.; Lindahl, E.; Hess, B.; Groenhof, G.; Mark, A. E.; Berendsen, H. J. C. J. Comput. Chem. 2005, 26, 1701. (29) Lindahl, E.; Hess, B.; van der Spoel, D. J. Mol. Model. 2001, 7, 306. (30) Jorgensen, W. L.; Maxwell, D. S.; Tirado-Rives, J. J. Am. Chem. Soc. 1996, 118, 11225. (31) Rizzo, R. C.; Jorgensen, W. L. J. Am. Chem. Soc. 1999, 121, 4827. (32) Price, M. L.; Ostrovsky, D.; Jorgensen, W. L. J. Comput. Chem. 2001, 22, 1340. (33) Kaminski, G. A.; Friesner, R. A.; Tirado-Rives, J.; Jorgensen, W. L. J. Phys. Chem. B 2001, 105, 6474. (34) Takaoka, Y.; Pasenkiewicz-Gierula, M.; Miyagawa, H.; Kitamura, K.; Tamura, Y.; Kusumi, A. Biophys. J. 2000, 79, 3118. (35) Murzyn, K.; Pasenkiewicz-Gierula, M. Acta Biochim. Pol. 1999, 46, 631. (36) Rog, T.; Vattulainen, I.; Bunker, A.; Karttunen, M. J. Phys. Chem. B 2007, 111, 10146.

Ro´g et al. (37) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R.; Klein, M. L. J. Chem. Phys. 1983, 79, 926. (38) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. J. Comput. Chem. 1997, 18, 1463. (39) Nose´, S. J. Chem. Phys. 1984, 81, 511. (40) Hoover, W. G. Phys. ReV. A 1985, 31, 1695. (41) Parrinello, M.; Rahman, A. J. Appl. Phys. 1981, 52, 7182. (42) Essman, U.; Perera, L.; Berkowitz, M. L.; Darden, H. L. T.; Pedersen, L. G. J. Chem. Phys. 1995, 103, 8577. (43) Patra, M.; Karttunen, M.; Hyvonen, M. T.; Falck, E.; Vattulainen, I. J. Phys. Chem. B 2004, 108, 4485. (44) Murzyn, K.; Zhao, W.; Karttunen, M.; Kurdziel, M.; Ro´g, T. Biointerphases 2006, 1, 98. (45) Hess, B. J. Chem. Phys. 2002, 116, 209. (46) Martinez-Seara, H.; Ro´g, T.; Pasenkiewicz-Gierula, M; Vattulainen, I.; Karttunen, M.; Reigada, R. J. Phys. Chem. B 2007, 111, 11162. (47) Davis, J. H. Biochim. Biophys. Acta 1983, 737, 117. (48) Pasenkiewicz-Gierula, M.; Takaoka, Y; Miyagawa, H.; Kitamura, K.; Kusumi, A. J. Phys. Chem. A 1997, 101, 3677. (49) Rapaport, D. C. Mol. Phys. 1983, 50, 1151. (50) Chandra, A. Phys. ReV. Lett. 2000, 85, 768. (51) Balasubramanian, S.; Pal, S.; Bagchi, B. Phys. ReV. Lett. 2002, 89, 115505. (52) Pasenkiewicz-Gierula, M.; Ro´g, T.; Kitamura, K.; Kusumi, A. Biophys. J. 2000, 78, 1376. (53) Zhao, W.; Ro´g, T.; Gurtovenko, A. A.; Vattulainen, I.; Karttunen, M. Biochimie 2008, 90, 930. (54) MacDonald, P. M.; Seelig, J. Biochemistry 1987, 26, 6298. (55) Allegrini, P. R.; Pluschke, G.; Seelig, J. Biochemistry 1984, 23, 6425. (56) Horobin, R. W.; Trapp, S.; Weissig, V. J. Controlled Release 2007, 121, 125. (57) Epand, R. F.; Schlattner, U.; Wallimann, T.; Lacombe, M.-L.; Epand, R. M. Biophys. J. 2007, 92, 126. (58) Pinheiro, T. J. T.; Duralski, A. A.; Watts, A. Biochemistry 1994, 33, 4896. (59) Schlame, M.; Ren, M.; Xu, Y.; Greenberg, M. L.; Haller, I. Chem. Phys. Lipids 2005, 138, 38. (60) Han, X.; Yang, J.; Yang, K.; Zhao, Z.; Abendschein, D. R.; Gross, R. W. Biochemistry 2007, 46, 6417–6428.

JP8077369