Distinct membrane properties are differentially influenced by cardiolipin content and acyl chain composition in biomimetic membranes

Abstract

Cardiolipin (CL) has a critical role in maintaining mitochondrial inner membrane structure. In several conditions such as heart failure and aging, there is loss of CL content and remodeling of CL acyl chains, which are hypothesized to impair mitochondrial inner membrane biophysical organization. Therefore, this study discriminated how CL content and acyl chain composition influenced select properties of simple and complex mitochondrial mimicking model membranes. We focused on monolayer excess area/molecule (a measure of lipid miscibility), bilayer phase transitions, and microdomain organization. In monolayer compression studies, loss of tetralinoleoyl [(18:2)₄] CL content decreased the excess area/molecule. Replacement of (18:2)₄CL acyl chains with tetraoleoyl [(18:1)₄] CL or tetradocosahexaenoyl [(22:6)₄] CL generally had little influence on monolayer excess area/molecule; in contrast, replacement of (18:2)₄CL acyl chains with tetramyristoyl [(14:0)₄] CL increased monolayer excess area/molecule. In bilayers, calorimetric studies showed that substitution of (18:2)₄CL with (18:1)₄CL or (22:6)₄CL lowered the phase transition temperature of phosphatidylcholine vesicles whereas (14:0)₄CL had no effect. Finally, quantitative imaging of giant unilamellar vesicles revealed differential effects of CL content and acyl chain composition on microdomain organization, visualized with the fluorescent probe Texas Red DHPE. Notably, microdomain areas were decreased by differing magnitudes upon lowering of (18:2)₄CL content and substitution of (18:2)₄CL with (14:0)₄CL or (22:6)₄CL. Conversely, exchanging (18:2)₄CL with (18:1)₄CL increased microdomain area. Altogether, these data demonstrate that CL content and fatty acyl composition differentially target membrane physical properties, which has implications for understanding how CL regulates mitochondrial activity and the design of CL-specific therapeutics.

Graphical abstract

1.0 Introduction

Cardiolipin (CL) ¹ is a unique anionic phospholipid that consists of two phosphatidic acid moieties linked by a central glycerol. CL is predominately localized and synthesized in the inner mitochondrial membrane where it plays an important role in mitochondrial fusion/fission, apoptosis, cellular signaling, and electron transport [1-4]. CL interacts with proteins of the inner mitochondrial membrane, especially respiratory chain complexes and substrate carriers during oxidative phosphorylation [5-7]. Furthermore, it is hypothesized that CL, due to its ability to form non-bilayer structures, forms local microdomains, which may be essential for the formation of highly curved cristae [8;9]. Therefore, CL has a critical role in regulating mitochondrial membrane structure and thereby function.

In conditions like aging or pathologies such as cancers, diabetes, cardiac diseases, Parkinson's, and Barth syndrome, CL undergoes considerable changes [10-21]. Three well-identified pathological alterations to CL include the loss of CL content, peroxidation, and remodeling of acyl chains [10;12;15;22]. Loss of CL content could arise from either enhanced CL degradation, due to an up-regulation in phospholipase activity, or by a decrease in de novo synthesis as a result of compromised functions of the enzymes involved in CL synthesis [10;11]. Many of the pathological consequences associated with changes in CL content are hypothesized to be driven by a loss of CL, particularly tetralinoloeyl [(18:2)₄] CL that is highly abundant in tissues such as the heart [10;11]. For instance, in cardiac ischemia-reperfusion injury, CL concentration is decreased by up to 50% [23].

Aberrant CL acyl chain remodeling may arise from decreased availability of (18:2)₄CL and/or abnormal alterations in the specific activity of remodeling enzymes such as tafazzin [10;12;13]. CL remodeling events are highly diverse. For example, in the streptozotocin model of diabetes, a loss of (18:2)₄CL acyl chains is accompanied by an increase in the abundance of docosahexaenoyl (DHA, 22:6) acyl chains, which could presumably render the mitochondrial membrane more susceptible to oxidative damage [22]. Similarly, in aging, DHA acyl chains are elevated in rat heart CL in addition to increased abundance of arachidonic acid and decreased availability of linoleic acid [24].

The aforementioned changes in CL abundance and/or acyl chain composition would presumably influence the biophysical organization of lipid molecules in the inner mitochondrial membrane. Indeed, several studies show CL content regulates microviscosity, packing, morphology, and phase behavior in isolated mitochondria and model membranes [25-34]. In contrast, no lab has completely discriminated the effects of CL concentration compared to CL acyl chain remodeling on membrane organization. This is essential since there is debate in the field about the importance of CL acyl chain composition on mitochondrial function [35]. Thus, the objective of this study was to investigate select membrane properties of monolayers and bilayers in response to lowering of (18:2)₄CL content and replacement of 18:2 acyl chains with other fatty acids. The approach relied on model membranes, which provide the advantage of tightly controlling lipid composition [36].

2.0 Materials and Methods

2.1 Materials

Bovine heart cardiolipin [(18:2)₄CL)], 1,1′,2,2′-tetratetradecanoyl cardiolipin [(14:0)₄CL], 1,1′,2,2′-tetra-(9Z-octadecenoyl) cardiolipin [(18:1)₄CL], 11,2-dihexadecanoyl-sn-glycero-3-phosphocholine (DPPC), 1,2-di-(9Z-octadecenoyl)-sn-glycero-3-phosphocholine (DOPC), 1,2-di-(9Z-octadecenoyl)-sn-glycero-3-phosphoethanolamine (DOPE), 1,2-di-(9Z-octadecenoyl)-sn-glycero-3-phospho-L-serine (DOPS) and cholesterol (Chol) were purchased from Avanti Polar Lipids (Alabaster, AL). 1,1′,2,2′-tetra-(4Z,7Z,10Z,13Z,16Z,19Z-docosahexaenoyl) cardiolipin [(22:6)₄CL] was a custom synthesis from Avanti Polar Lipids. All solvents were HPLC grade and purchased from Fisher Scientific (Hampton, NH). Fresh stocks of lipid mixtures were used to minimize oxidation and handling of lipids occurred under low light conditions and a gentle stream of nitrogen gas.

2.2 Generation of monolayers

Pressure-area isotherms were generated on a Mini Langmuir-Blodgett Trough (KSV NIMA, Biolin Scientific, Paramus, NJ) using a Wilhelmy plate at a compression rate of 3.0 mm/min. Lipid mixtures were spotted using a Hamilton syringe and spread on a subphase of 10 mM sodium phosphate buffer (pH 7.4). Trough barriers were compressed and expanded prior to spotting lipids until the surface pressure remained constant (<0.3 mN/m). Compression began 10 minutes after spotting the lipids to ensure evaporation of chloroform. The trough was washed three times with 70% ethanol, Milli-Q water, and subphase prior to collecting pressure-area isotherms.

Lipid monolayers containing differing levels of CL were composed of either a binary mixture of DPPC and CL, or heterogeneous mixtures of DOPC, DOPE, CL, DOPS, and Chol. The amounts of PC, PE, CL, PS, and Chol approximated ratios found in cardiac inner mitochondrial membranes in health and disease [37]. All lipid mixtures were acquired multiple times to ensure reproducibility.

2.3 Analysis of lipid miscibility, elasticity modulus, and Gibbs free energy of mixing in monolayers

The ideal mean molecular area of multi-component mixed monolayers was calculated at a constant surface pressure (π) by:

$$A_{\mathit{\text{ideal}}}=X_1{(A_1)}_{\pi }+X_2{(A_2)}_{\pi }+\dots X_n{(A_n)}_{\pi }$$ (Eq. 1)

where X_n represents the mol fraction of each individual component and A_n represents the mean molecular area of each component [38]. The excess area/molecule (A_ex), a measurement of lipid miscibility, was calculated at a given surface pressure (π) by [39-41]:

$$A_{\mathit{\text{ex}}}={(A_{12\dots n})}_{\pi }-{(X_1{A}_1+X_2{A}_2+\dots X_n{A}_n)}_{\pi }$$ (Eq. 2)

where X_n represents the mol fraction of each individual component and A_n represents the mean molecular area of each component. (A_12…n) represents the mean molecular area of the mixed monolayer of interest at a given surface pressure (π). A negative excess area/molecule indicates attractive forces and a positive excess area/molecule indicates repulsive forces [39]. The surface pressure-area isotherms were also used to calculate the surface elasticity modulus ( $C_s^{-1}$):

$$C_s^{-1}=(-A){(d\pi /dA)}_{\pi }$$ (Eq. 3)

where A is the mean molecular area of the lipid mixture of interest at the indicated surface pressure (π). Gibbs free energy of mixing (ΔG_mix) was determined by the following relation:

$$\Delta G_{\mathit{\text{mix}}}=\Delta G_{\mathit{\text{ex}}}+\Delta G_{\mathit{\text{ideal}}}$$ (Eq. 4)

where ΔG_ex represents the excess Gibbs free energy of mixing and ΔG_ideal is the ideal Gibbs free energy of mixing. ΔG_ex was calculated from the pressure-area isotherms using the following expression [39]:

$$\Delta G_{\mathit{\text{ex}}}={\int }_0^{\pi }[A_{12\dots n}-(X_1{A}_1+X_2{A}_2+X_n{A}_n)]d\pi$$ (Eq. 5)

where A_n and X_n are the mean molecular area and mol fraction of each component, respectively, at a given surface pressure (π). ΔG_ideal was calculated from:

$$\Delta G_{\mathit{\text{ideal}}}=RT(X_{1}ln{X}_1+X_{2}ln{X}_2+X_{n}ln{X}_n)$$ (Eq. 6)

where R represents the ideal gas constant, T represents the temperature in Kelvin, and X_n represents the mol fraction of each component.

2.4 Differential scanning calorimetry (DSC)

Lipid mixtures of DPPC and CL (5.0 mg total lipid) were co-dissolved in chloroform and dried under nitrogen gas. Excess chloroform was removed by vacuum pumping in the dark overnight. Each sample was hydrated for one hour with frequent vortex mixing in 10 mM sodium phosphate buffer (pH 7.4) at ∼55 °C. The lipid suspension was subjected to three freeze-thaw cycles and then extruded using a mini extruder (Avanti Polar Lipids, Alabaster, AL). Each sample was extruded at ∼55 °C for a total of 15 passes across a 1.0 micron polycarbonate membrane (Whatman Nuclepore, Clifton, NJ).

A total of 0.30 mg of each lipid mixture was placed in hermetically sealed aluminum T-zero pans (TA Instruments, New Castle, DE). DSC curves were obtained using a TA Instruments Q2000 DSC (TA Instruments, New Castle, DE). Samples were subjected to one heating cycle (20 °C to 60 °C) immediately followed by one cooling cycle (60 °C to 20 °C) at a scan rate of 2 °C per minute. Each sample was analyzed against a reference pan containing 10 mM sodium phosphate buffer (pH 7.4). Raw DSC curves were analyzed using TA Universal Analysis 2000 software as previously described (TA Instruments, New Castle, DE) [42].

2.5 Construction of giant unilamellar vesicles (GUVs)

GUVs were constructed by electroformation as described [43;44]. Lipids (37.5 mol% DPPC, 27.4 mol% DOPC, 25 mol% Chol, 10 mol% CL) and the fluorescent probe Texas Red DHPE (0.1 mol%; Life Technologies, Carlsbad, CA) were co-dissolved in chloroform to achieve a total concentration of 0.5 mg/ml [45]. Next, 7.5 μg of total lipid was spread onto the conductive side of an indium tin oxide coated glass slide. The lipid-coated slide was subjected to vacuum pumping in the dark for 3 hours to remove excess solvent. Once the lipid film was dried, a GUV electroformation chamber was assembled as described [44].

GUV electroformation was performed at ∼55 °C, in the dark, using a 250 mM sucrose solution. GUV electroformation frequency and peak-to-peak voltage was controlled using a HP3324A programmable function generator. Initially, a sine waveform with a frequency of 10 Hz and a peak-to-peak voltage ranging from 0.2 V to 1.4 V was applied and linearly increased over a 30 minute period. Next, a sine waveform with a frequency of 10 Hz and a peak-to-peak voltage of 1.4 V was applied and held constant for 120 minutes. For the detachment of GUVs, a square waveform with a frequency of 4.5 Hz and peak-to-peak voltage of 2.1 V was applied and held constant for 30 minutes. Once electroformation was completed, GUVs were extracted from the chamber using a 20½-gauge needle and refrigerated (4 °C) overnight prior to imaging. GUVs were drawn into a rectangular micro-capillary tube, mounted onto a microscope slide, and imaged at room temperature (23 °C).

2.6 Confocal microscopy and image analysis

Imaging was conducted using an Olympus FV1000 Confocal Microscope using a 60× 1.35NA oil immersion objective (Olympus, Waltham, MA). Analysis of microdomains was conducted with NIH ImageJ as previously described [46;47]. After background subtraction, membrane microdomains were quantified as the area occupied by the fluorophore Texas Red DHPE, which reports on disordered microdomains [48-50]. A range of microdomain areas were measured and are presented as a frequency distribution. A Gaussian fit (Graph Pad Prism v5.0b) was applied to the frequency distributions. Larger areas reflect greater coverage of the fluorophore on the perimeter of the GUV. The diameter of each GUV was quantified using NIH ImageJ.

2.7 Statistics

Data were analyzed with GraphPad Prism v5.0b. Data were ensured to display normalized distributions. Statistical significance relied on one-way ANOVA followed by a post-hoc Bonferroni test compared to (18:2)₄CL. For select GUV studies, statistical analyses were conducted with a one-way ANOVA followed by a post-hoc Bonferroni multiple comparison test. P values less than 0.05 were considered significant.

3.0 Results

3.1 Monolayers of differing CL species in DPPC/CL and DOPC/DOPE/CL/DOPS/Chol mixtures

The behavior of four different CL species (Fig. S1A-D, Supporting Material) was studied in DPPC/CL and DOPC/DOPE/CL/DOPS/Chol monolayers. The rationale for using the DPPC/CL model system was to allow for comparison with previous studies with binary mixtures containing CL [33;34;51]. The five-component mixture (PC/PE/CL/PS/Chol) modeled what is more biologically relevant for mitochondrial inner membranes. The first objective was to specifically understand how loss of (18:2)₄CL content and replacement of (18:2)₄CL with other CL species would influence monolayer properties.

We present the raw monolayer data, which were then used to generate quantitative values on excess area/molecule, elasticity modulus, and Gibbs free energy of mixing. Sample monolayers of DPPC/CL with increasing levels of (18:2)₄CL, (14:0)₄CL, (18:1)₄CL, and (22:6)₄CL are depicted in Figure 1A-D. All four CL species alone occupied a much greater mean molecular area compared to DPPC alone, which was in agreement with the notion that CL is a larger structure than DPPC. At high surface pressures, (18:2)₄CL notably displayed a long plateau region suggesting possible expulsion of (18:2)₄CL from the monolayer (Fig. 1A). (14:0)₄CL displayed a unique isotherm with a plateau region around 20 mN/m (Fig. 1B). In contrast, (18:1)₄CL (Fig. 1C) and (22:6)₄CL (Fig. 1D) did not have any plateau regions. Decreasing the levels of any of the CLs in DPPC/CL monolayers, which occurred at the expense of increasing DPPC, decreased the mean molecular area at all surface pressures (Fig. 1A-D).

Figure 1. Monolayers containing differing CL species mixed with DPPC.

Pressure-area isotherms of DPPC/CL containing differing levels of (A) (18:2)₄CL, (B) (14:0)₄CL, (C) (18:1)₄CL, and (D) (22:6)₄CL. Isotherms were acquired at 23 °C using a 10 mM sodium phosphate buffer (pH 7.4). Isotherms were used for calculating excess area/molecule, elasticity modulus, and Gibbs free energy of mixing. Data are average ± S.E.M. from 3 independent experiments.

Figure 2A-D respectively presents sample monolayers of DOPC/DOPE/CL/DOPS/Chol with differing mol percent of (18:2)₄CL, (14:0)₄CL, (18:1)₄CL, and (22:6)₄CL. CL concentration was increased at the expense of the other lipids (for simplicity referred to as ‘compensated loss’); thus, as CL levels increased, the levels of DOPC/DOPE/DOPS/Chol were reduced. Pressure-area isotherms for all four CL species showed one clear trend; decreasing the levels of CL in the DOPC/DOPE/CL/DOPS/Chol system decreased the mean molecular area (Fig. 2A-D).

Figure 2. Monolayers containing differing CL species in model mitochondrial membranes.

Pressure-area isotherms of DOPC/DOPE/CL/DOPS/Chol at varying mol percent of (A) (18:2)₄CL, (B) (14:0)₄CL, (C) (18:1)₄CL, and (D) (22:6)₄CL. Isotherms were acquired at 23 °C using a 10 mM sodium phosphate buffer (pH 7.4). Isotherms were used for calculating excess area/molecule, elasticity modulus, and Gibbs free energy of mixing. Data are average ± S.E. from 3-9 independent experiments.

Monolayers of DPPC/(18:2)₄CL (80/20 mol%) were also generated in which (18:2)₄CL concentration was reduced by 50%, referred to as an ‘absolute loss’ of CL (Fig. S2A, Supporting Material). As expected, an absolute loss of (18:2)₄CL content in DPPC/(18:2)₄CL monolayers decreased the mean molecular area relative to DPPC/(18:2)₄CL (80/20 mol%) (Fig. S2A, Supporting Material). Similarly, monolayers of DOPC/DOPE/(18:2)₄CL/DOPS/Chol (40/30/20/5/5 mol%) were generated in which (18:2)₄CL concentration was lowered by 50%, again referred to as an ‘absolute loss’ of CL (Fig. S2B, Supporting Material). The absolute loss of (18:2)₄CL content in DOPC/DOPE/(18:2)₄CL/DOPS/Chol monolayers did not have a dramatic effect on the pressure-area isotherms, displaying a small decrease in the mean molecular area (Fig. S2B, Supporting Material).

3.2 Loss of (18:2)₄CL content promotes a decrease in the excess area/molecule with modest effects on the elasticity modulus

The aforementioned pressure-area isotherms (Figs. 1-2 & Fig. S2 Supporting Material) were used to calculate the excess area/molecule, the elasticity modulus (C_s^-1), and the Gibbs free energy of mixing (ΔG_mix) in response to a loss of (18:2)₄CL content and changes in CL acyl chain composition. All analyses were conducted at a physiologically relevant surface pressure of 30 mN/m [52;53].

DPPC/(18:2)₄CL (80/20 mol%) and DOPC/DOPE/(18:2)₄CL/DOPS/Chol (40/30/20/5/5 mol%) were defined as the control (Fig. 3A, Table 1). Compensated loss of (18:2)₄CL (i.e. increase in DPPC at the expense of lowering CL) had no influence on the excess area/molecule (Fig. 3A left panel, Table 1). An absolute loss of (18:2)₄CL content (i.e. 50% loss of CL concentration) decreased the excess area/molecule by 11.0 Ų in DPPC/(18:2)₄CL monolayers relative to the control (Fig. 3A left panel, Table 1). Similarly, in DOPC/DOPE/(18:2)₄CL/DOPS/Chol monolayers, compensated loss of (18:2)₄CL (i.e. increase in DOPC/DOPE/DOPS/Chol at the expense of lowering CL) had no influence on excess area/molecule (Fig. 3A right panel, Table 1). An absolute loss of (18:2)₄CL content (i.e. 50% loss of CL concentration) in DOPC/DOPE/(18:2)₄CL/DOPS/Chol monolayers decreased the excess area/molecule by 6.0 Ų relative to the control (Fig. 3A right panel, Table 1).

Figure 3. Loss of (18:2)₄CL content decreases the excess area/molecule in monolayer models.

(A) Excess area/molecule, (B) elasticity modulus, and (C) Gibbs free energy of mixing for DPPC/(18:2)₄CL (left panel) and DOPC/DOPE/(18:2)₄CL/DOPS/Chol (right panel) monolayers. Values were calculated from pressure-area isotherms at a physiologically relevant surface pressure of 30 mN/m. Control samples are defined as either DPPC/(18:2)₄CL (80/20 mol%) or DOPC/DOPE/(18:2)₄CL/DOPS/Chol (40/30/20/5/5 mol%). Compensated loss is defined as a 50% loss of (18:2)₄CL content with a compensated increase in DPPC content or DOPC/DOPE/DOPS/Chol content. Absolute loss is defined as a 50% loss of (18:2)₄CL concentration. Data are average ± S.E.M. from 3-6 independent experiments. Asterisks indicate significant from the control: *P<0.05, **P<0.01, ***P<0.001.

Table 1. Excess area/molecule, elasticity modulus, and Gibbs free energy of mixing in response to decreasing CL content in simple and complex lipid mixtures.

Values were calculated from pressure-area isotherms at a physiologically relevant surface pressure of 30 mN/m. Data are average ± S.E.M. from 3-6 independent experiments.

Lipid Mixture	Excess Area/Molecule (Å2)	Cs-1 (mN/m)	ΔGmix (kJ/mol)
DPPC/(18:2)4CL (80/20 mol%)a	-0.2 ± 0.8	69.1 ± 0.7	-1.8 ± 0.0
DPPC/(18:2)4CL (90/10 mol%)b	-1.2 ± 0.9	70.9 ± 1.7	-0.6 ± 0.0
DPPC/(18:2)4CL (50% loss of CL concentration)c	-11.2 ± 0.7	58.0 ± 1.9	-0.9 ± 0.0
DOPC/DOPE/(18:2)4CL/DOPS/Chol (40/30/20/5/5 mol%)a	1.8 ± 1.1	87.7 ± 3.1	-3.5 ± 0.4
DOPC/DOPE/(18:2)4CL/DOPS/Chol (46/33/10/5.5/5.5 mol%)b	0.5 ± 1.5	94.7 ± 1.2	-3.5 ± 0.0
DOPC/DOPE/(18:2)4CL/DOPS/Chol (50% of CL concentration)c	-4.2 ± 0.9	79.0 ± 2.9	-3.9 ± 0.1

^arepresents the control.

^brepresents the ‘compensated loss’ of CL, which is defined as decreasing CL content by 50 mol% at the expense of increasing other phospholipids in the mixture.

^crepresents the ‘absolute loss’ of CL, which is defined as 50% loss of CL concentration relative to the control.

Elasticity modulus (C_s^-1) values were not affected by compensated loss of (18:2)₄CL but were lowered by 11.1 mN/m with the absolute loss of (18:2)₄CL content in DPPC/(18:2)₄CL monolayers (Fig. 3B left panel, Table 1). Compensated or absolute loss of (18:2)₄CL content had no influence on C_s^-1 values of DOPC/DOPE/(18:2)₄CL CL/DOPS/Chol monolayers (Fig. 3B right panel, Table 1). The formation of both DPPC/(18:2)₄CL (Fig. 3C left panel, Table 1) and DOPC/DOPE/(18:2)₄CL/DOPS/Chol monolayers (Fig. 3C right panel, Table 1) showed negative values for the Gibbs free energy of mixing. Gibbs free energy of mixing was lowered with the compensated and absolute loss of (18:2)₄CL compared to the control in DPPC/(18:2)₄CL monolayers (Fig. 3C left panel, Table 1).

3.3 Replacement of (18:2)₄CL with (18:1)₄CL or (22:6)₄CL has no effect on excess area/molecule in model mitochondrial monolayers with the exception of (14:0)₄CL

CL undergoes remodeling of its acyl chains, particularly to 22:6, in conditions such as heart failure or aging [12;22;24;54]. Therefore, the different CL species were compared in the DPPC/CL (80/20 mol%) (Fig. 4, left panel) and DOPC/DOPE/CL/DOPS/Chol (40/30/20/5/5 mol%) (Fig. 4, right panel) monolayer models. (14:0)₄CL showed a significant increase in the excess area/molecule in both model systems by approximately 4.1 Ų and 8.9 Ų, respectively, compared to (18:2)₄CL (Fig. 4A left and right panels, Table 2). The (22:6)₄CL species decreased the excess area/molecule by 6.2 Ų compared to (18:2)₄CL in the DPPC/CL system (Fig. 4A left panel, Table 2) but had no effect in the model mitochondrial monolayers of DOPC/DOPE/CL/DOPS/Chol (Fig. 4A right panel, Table 2). Replacement of (18:2)₄CL with (18:1)₄CL had no effect on area/molecule in either model system (Fig. 4A left and right panels, Table 2).

Figure 4. Replacement of (18:2)₄CL with other CLs differentially influences monolayer excess area/molecule, elasticity modulus, and Gibbs free energy of mixing.

(A) Excess area/molecule, (B) elasticity modulus, and (C) Gibbs free energy of mixing for DPPC/CL (left panel) and DOPC/DOPE/CL/DOPS/Chol (right panel) monolayers. Values were calculated from pressure-area isotherms at a surface pressure of 30 mN/m. (18:2)₄CL species is defined as the control. DPPC/CL levels were maintained at 80/20 (mol%) and DOPC/DOPC/CL/DOPS/Chol levels were kept at 40/30/20/5/5 (mol%). Data are average ± S.E.M. from 3-9 independent experiments. Asterisks indicate significance from (18:2)₄CL: *P<0.05, **P<0.01, ***P<0.001.

Table 2. Excess area/molecule, elasticity modulus, and Gibbs free energy of mixing for CLs of differing acyl chain composition in simple and complex lipid mixtures.

Values were calculated from pressure-area isotherms at a physiologically relevant surface pressure of 30 mN/m. Data are average ± S.E.M. from 3-9 independent experiments.

Lipid Mixture	Excess Area/Molecule (Å2)	Cs-1 (mN/m)	ΔGmix (kJ/mol)
DPPC/(18:2)4CL (80/20 mol%)a	-0.2 ± 0.8	69.1 ± 0.7	-1.8 ± 0.0
DPPC/(14:0)4CL (80/20 mol%)	3.9 ± 0.4	194.2 ± 16.1	-1.1 ± 0.0
DPPC/(18:1)4CL (80/20 mol%)	0.4 ± 0.7	71.1 ± 4.1	-1.1 ± 0.1
DPPC/(22:6)4CL (80/20 mol%)	-6.4 ± 0.8	47.9 ± 0.6	-0.8 ± 0.1
DOPC/DOPE/(18:2)4CL/DOPS/Chol (40/30/20/5/5 mol%)a	1.8 ± 1.1	87.7 ± 3.1	-3.5 ± 0.4
DOPC/DOPE/(14:0)4CL/DOPS/Chol (40/30/20/5/5 mol%)	10.7 ± 0.6	96.3 ± 0.7	-4.1 ± 0.0
DOPC/DOPE/(18:1)4CL/DOPS/Chol (40/30/20/5/5 mol%)	-0.5 ± 0.9	98.0 ± 1.3	-3.3 ± 0.2
DOPC/DOPE/(22:6)4CL/DOPS/Chol (40/30/20/5/5 mol%)	-0.3 ± 0.5	74.4 ± 2.9	-4.1 ± 0.1

^arepresents the control.

C_s^-1 values were significantly elevated by 2.8-fold with (14:0)₄CL compared to (18:2)₄CL in the DPPC/CL mixture (Fig. 4B left panel, Table 2). (18:1)₄CL and (22:6)₄CL had no effect on C_s^-1 of DPPC/CL monolayers (Fig. 4B left panel, Table 2). In the DOPC/DOPE/CL/DOPS/Chol monolayers, (14:0)₄CL and (18:1)₄CL very modestly increased C_s^-1 values, whereas (22:6)₄CL decreased C_s^-1 compared to (18:2)₄CL (Fig. 4B right panel, Table 2). Gibbs free energy of mixing values were all negative for DPPC/CL (Fig. 4C left panel, Table 2) and DOPC/DOPE/CL/DOPS/Chol (Fig. 4C right panel, Table 2) monolayers. Replacement of (18:2)₄CL with the other CL species in the DPPC/CL monolayers led to a decrease in the Gibbs free energy of mixing, but overall the quantity remained negative (Fig. 4C left panel, Table 2).

3.4 At low levels of CL, replacement of (18:2)₄CL with (18:1)₄CL or (22:6)₄CL has no effect on excess area/molecule in model mitochondrial monolayers with the exception of (14:0)₄CL

We next addressed how exchange of (18:2)₄CL with the other CL species would influence area/molecule, elasticity modulus, and Gibbs free energy at a low CL concentration. In DOPC/DOPE/CL/DOPS/Chol (48/35/5/6/6 mol%) (Fig. S3A left panel, Supporting Material), (14:0)₄CL modestly increased the excess area/molecule with no statistically significant effect with the other CL species compared to (18:2)₄CL. In DOPC/DOPE/CL/DOPS/Chol (46/33/10/5.5/5.5 mol%) monolayers (Fig. S3A, right panel, Supporting Material), there were no differences between the CL species on excess area/molecule. Both (18:1)₄CL and (22:6)₄CL lowered C_s^-1 values by 14.0 mN/m and 22.6 mN/m, respectively, in DOPC/DOPE/CL/DOPS/Chol (48/35/5/6/6 mol%) compared to the (18:2)₄CL control (Fig. S3B left panel, Supporting Material). In DOPC/DOPE/CL/DOPS/Chol (46/33/10/5.5/5.5 mol%) monolayers containing (18:1)₄CL and (22:6)₄CL, C_s^-1 values were lowered by 10.5 mN/m and 18.4 mN/m, respectively (Fig. S3B right panel, Supporting Material). Gibbs free energy of mixing was negative for all lipid mixtures (Fig. S3C, left and right panels, Supporting Material).

3.5 In Bilayers, Replacement Of (18:2)₄Cl With (18:1)₄Cl Or (22:6)₄Cl Decreases The Phase Transition Temperature Of Dppc With The Exception Of (14:0)₄Cl

We further investigated if differing CL species, upon addition to DPPC bilayers, had differential effects on the phase transition temperature, cooperativity, and enthalpies of DPPC using DSC. DSC heating scans of lipid vesicles showed that the addition of increasing amounts of (18:2)₄CL (5-20 mol%) (Fig. 5A) decreased the main phase transition (T_m, centered at 41.6 °C), cooperativity (T_m1/2), and enthalpy (ΔH) for DPPC (Table 3). In contrast, addition of (14:0)₄CL to DPPC had no major effect on the main phase transition temperature (Fig. 5B). Enthalpy values were lowered with increasing amounts of (14:0)₄CL to DPPC to a much smaller degree compared to (18:2)₄CL (Fig. 5B, Table 3). Both (18:1)₄CL (Fig. 5C) and (22:6)₄CL (Fig. 5D) qualitatively and quantitatively (Table 3) showed similar effects as (18:2)₄CL on the main phase transition temperature, cooperativity, and enthalpy of DPPC. No major hysteresis was observed given that analyses of DSC cooling scans showed similar results (data not shown).

Figure 5. (14:0)₄CL does not lower the main phase transition temperature of DPPC as effectively as other CL species.

DSC heating scans for DPPC with differing levels of (A) (18:2)₄CL, (B) (14:0)₄CL, (C) (18:1)₄CL, and (D) (22:6)₄CL. Values to the right indicate the mol percent of DPPC/CL. No hysteresis was observed since nearly identical results were observed for cooling scans. Data are single scans representative of 3 independent experiments.

Table 3. Thermodynamic profiles of DPPC lipid bilayers in the presence of differing CL species.

Data are from heating scans of DPPC with increasing mol percent CL. Data are average ± S.E.M. from 3 independent experiments.

CL	DPPC/CL (mol%)	Tm (°C)	Tm1/2 (°C)	ΔH (kJ/mol)
----	100/0	41.6 ± 0.0	0.24 ± 0.0	46.1 ± 0.7
(18:2)4	95/5	40.0 ± 0.2	5.3 ± 0.1	40.1 ± 3.1
	90/10	38.6 ± 0.2	8.9 ± 0.5	34.7 ± 0.9
	85/15	37.1 ± 0.3	6.8 ± 0.2	22.3 ± 0.9
	80/20	34.8 ± 0.1	7.2 ± 0.8	19.1 ± 5.3
(14:0)4	95/5	41.7 ± 0.1	0.8 ± 0.0	71.3 ± 2.5
	90/10	41.4 ± 0.1	1.9 ± 0.2	77.0 ± 1.6
	85/15	41.1 ± 0.1	3.2 ± 0.1	69.4 ± 1.2
	80/20	41.1 ± 0.1	3.7 ± 0.1	64.3 ± 3.4
(18:1)4	95/5	39.9 ± 0.1	5.5 ± 0.1	52.5 ± 1.0
	90/10	38.7 ± 0.2	7.8 ± 0.5	31.3 ± 3.5
	85/15	35.9 ± 0.1	6.6 ± 0.2	16.8 ± 1.8
	80/20	32.7 ± 0.2	10.3 ± 1.8	30.9 ± 7.2
(22:6)4	95/5	40.9 ± 0.1	6.0 ± 0.1	43.2 ± 1.5
	90/10	40.2 ± 0.2	6.8 ± 0.3	33.2 ± 2.8
	85/15	38.2 ± 0.1	6.5 ± 0.1	21.4 ± 0.8
	80/20	36.7 ± 0.2	6.9 ± 0.3	20.4 ± 3.3

3.6 Loss of (18:2)₄CL content and remodeling to (14:0)₄CL have a stronger influence on lipid microdomain organization than remodeling to (18:1)₄ or (22:6)₄CL

Finally, we assessed how CL content and acyl chain composition influenced lipid microdomain organization, a key biophysical parameter of membranes. In order to understand how different CL species would organize lipid microdomains, we employed a model membrane system of DPPC/DOPC/Chol that is well known to form phase separated microdomains [45]. The fluorescent probe Texas Red DHPE was used to image the organization of microdomains. Microdomains were visible in GUVs made of DPPC/DOPC/Chol/(18:2)₄CL (37.5/27.4/25/10 mol%) with a broad Gaussian distribution of microdomain areas (Fig. 6A).

Figure 6. Loss of (18:2)₄CL content and CL acyl chain composition have differential effects on GUV lipid microdomain organization.

Sample images of DPPC/DOPC/Chol/CL (37.5/27.4/25/10 mol%) containing either (A) (18:2)₄CL, (B) (14:0)₄CL, (C) (18:1)₄CL, or (D) (22:6)₄CL. Plots to the right of the images indicate the frequencies of domains as a function of domain area (μm²). (E) Sample GUV image of DPPC/DOPC/Chol/(18:2)₄CL with a 50% loss of CL concentration. (F) The average microdomain area is plotted for the differing CL species and upon the loss of CL content. (G) Average diameter of the GUVs analyzed. Data are from a total of 40-60 GUVs analyzed from 2-3 independent experiments. Data are average ± S.E.M. Asterisks indicate significance from (18:2)₄CL, unless otherwise noted: **P<0.01, ***P<0.001.

Replacement of (18:2)₄CL with (14:0)₄CL in the DPPC/DOPC/Chol GUVs led to an increased frequency of distinct microdomains (Fig. 6B). Frequency analysis (Fig. 6B) showed a tightening distribution with smaller microdomain areas at higher frequencies with (14:0)₄CL compared to (18:2)₄CL. (18:1)₄CL generally showed less phase separated microdomains compared to (18:2)₄CL, with domains that covered the entire perimeter of the GUVs (Fig. 6C). This was reflected by a shift to the right in the domain area distribution compared to (18:2)₄CL (Fig. 6C). (22:6)₄CL also influenced microdomain organization, although not as pronounced as (14:0)₄CL when compared to (18:2)₄CL (Fig. 6D). We also measured microdomain areas upon the lowering of (18:2)₄CL concentration by 50% (Fig. 6E). The frequency distribution was tight with the majority of microdomains being relatively small (Fig. 6E) compared to (18:2)₄CL (Fig. 6A).

The average microdomain area for each CL species in the DPPC/DOPC/Chol/CL GUVs was calculated. Relative to (18:2)₄CL, both (14:0)₄CL and (22:6)₄CL had decreased average microdomain areas by 3.4-fold and 1.7-fold respectively, whereas (18:1)₄CL had an increased microdomain area by 1.3-fold (Fig. 6F). A 50% loss of (18:2)₄CL concetration decreased the average microdomain area relative to (18:2)₄CL, (18:1)₄CL, and (22:6)₄CL by 2.5-fold, 3.3-fold, and 1.5-fold, respectively (Fig. 6F). During the course of this study, we also observed that GUVs made of (22:6)₄CL or those that had a 50% loss of (18:2)₄CL concentration had a smaller diameter compared to (18:2)₄CL (Fig. 6G).

4.0 Discussion

4.1 Biological consequences for lowering (18:2)₄CL content and replacement of (18:2)₄CL with other CLs in monolayer studies

CL undergoes changes in concentration and acyl chain composition in a range of diseases. For instance, individuals with Barth Syndrome display a dramatic decrease in lymphoblast (18:2)₄CL content accompanied by remodeling of acyl chains [13;55;56]. A wide range of acyl chains are remodeled in human and mouse models of Barth Syndrome, which include (but not limited to) changes in 16:0, 16:1, 18:0, 18:1, 18:2, 20:2, 20:3, and 22:6 [56-58]. The aforementioned changes would presumably influence key biophysical properties of membranes that regulate mitochondrial function. Therefore, we studied CL content and acyl chain composition on select membrane properties.

In the differing model systems employed, (14:0)₄CL consistently displayed a different behavior when compared to (18:2)₄CL, (18:1)₄CL, and (22:6)₄CL. In monolayers, replacement of (18:2)₄CL with (14:0)₄CL led to a positive deviation in the excess area/molecule within DPPC/CL or DOPC/DOPE/CL/DOPS/Chol. A positive value for the excess area/molecule suggests that (14:0)₄CL does not ideally mix with the other lipids [39]. Calorimetry studies confirmed that the addition of small amounts of (14:0)₄CL to DPPC lipid bilayers did not completely abolish the phase transition temperature of DPPC as effectively as the other CLs. This again suggested (14:0)₄CL was not mixing with DPPC compared to other CLs.

The results with (14:0)₄CL were not in complete agreement with X-ray diffraction studies to show that (14:0)₄CL is highly miscible with DPPC [59]. The discrepancy between the studies could reflect differences in the buffers used or the surface pressures used for analyses. We relied on a surface pressure of 30 mN/m whereas the aforementioned study focused on lower surface pressures. Given that (14:0)₄CL is found in select bacteria but not in mammalian mitochondria [60], the biological significance is limited for human physiology. There is precedence to show the accumulation of short chain fatty acids besides (14:0)₄CL in yeast mutant models [61]. Perhaps, a significant increase in (14:0)₄CL could occur in response to dietary intervention with short chain saturated fatty acids, such as those found in coconut oil. To date, this possibility has not been investigated.

Exchange of (18:2)₄CL with either (18:1)₄CL or (22:6)₄CL did not strongly influence monolayer excess area/molecule and elasticity modulus. The only exception was with DPPC/CL mixtures, where (22:6)₄CL promoted a decrease in excess area/molecule but this effect was diminished in the more relevant DOPC/DOPE/CL/DOPS/Chol monolayers. The data with (22:6)₄CL were particularly surprising since we expected (22:6)₄CL to have a strong influence on elasticity modulus, given that DHA acyl chains, due to extensive polyunsaturation, are extremely flexible [62]. Our rationale for selecting (22:6)₄CL, as opposed to a CL with just one or two DHA acyl chains, was to test the most extreme example of a DHA-containing CL. Overall, (22:6)₄CL behaved similar to the control, whereas (14:0)₄CL did not, in the complex mitochondrial biomimetic monolayers.

There is precedence to show that n-3 polyunsaturated fatty acids can incorporate into CL, which may be beneficial or detrimental for mitochondrial function. In pre-clinical models, DHA and eicosapentaenoic acid (EPA, 20:5) incorporate into CL and are generally beneficial for improving metabolic outcomes associated with cardiovascular disease [63-66]. Paradoxically, DHA levels are also increased in the diabetic heart, which may have negative consequences since the fatty acid is highly oxidizable [12;22]. Therefore, future studies will determine how different heteroacid CL species containing EPA or DHA influence lipid miscibility. Furthermore, there is a need to address how mono-lysocardiolipin, saturated and monounsaturated CLs, which are associated with a range of diseases including Barth Syndrome, regulate lipid miscibility [4;67].

4.2 Loss of (18:2)₄CL content and remodeling to (14:0)₄CL have a stronger influence on microdomain organization than remodeling to (18:1)₄CL and (22:6)₄CL

It is hypothesized that distinct CL membrane microdomains may regulate protein activity and there is emerging evidence for their existence [9;68]. In E. coli, CL molecules form macroscopic domains in regions of high curvature, which likely support the recruitment of select proteins for regulation of bacterial division [9]. There is also some discussion of CL domains as having properties akin to lipid rafts (although this is highly debatable since cholesterol levels are low in the inner mitochondrial membranes) [69;70]. The importance of CL microdomains is of relevance during initiation of apoptosis, where caspase 8 is thought to localize to CL microdomains that would be spatially distributed at contact sites between the inner and outer mitochondrial membranes [71]. In another study, fission proteins were also reported to be associated with CL microdomains [72]. However, any conclusions about the existence of CL microdomains must account for the methodology used for isolating these membrane fractions, which may promote artifacts [73].

Given the potential importance of mitochondrial microdomains, we studied how loss of (18:2)₄CL content and replacement of (18:2)₄CL species influenced the organization of microdomains visualized with Texas Red DHPE [48-50]. We acknowledge the limitation of our experiments, which relied on raft/non-raft associated lipids and did not specifically measure CL-enriched domains. However, this model has previously been employed for assaying phase separation and thus the influence of CL on microdomain organization [72]. The data qualitatively and quantitatively demonstrate that lowering (18:2)₄CL content led to the formation of smaller domains on a micron scale compared to higher (18:2)₄CL levels. This suggests that domain size is tunable by CL levels analogous to studies on cholesterol and lipid rafts [74]. It is important to note that the loss of (18:2)₄CL content on microdomain size was equivalent to the effect of replacing (18:2)₄CL with (14:0)₄CL. An interpretation of these data are that replacement of (18:2)₄CL with (14:0)₄CL resembles the loss of (18:2)₄CL concentration.

Replacement of (18:2)₄CL with (14:0)₄CL or (22:6)₄CL also promoted the formation of smaller domains. In the case of (18:1)₄CL, there was no evidence for the formation of microdomains in the DPPC/DOPC/Chol/(18:1)₄CL GUVs. We speculate that this was likely due to (18:1)₄CL localizing to both ordered and disordered regions as reported for oleic acid containing phosphatidylethanolamines, thereby preventing macroscopic phase separation between ordered and disordered domains [75].

This study did not address the composition and properties of lipid microdomains. We tested the effects of the concentration and composition of CL on microdomain organization, which is one of many biophysical properties of membranes. Future imaging studies with a range of fluorescent probes will determine the nature of mitochondrial microdomains and their significance for mitochondrial function.

4.3. CL-protein interactions may be uniquely modified in response to changes in CL content and acyl chain composition

It is plausible that lowering of CL content and remodeling of its acyl chains would influence CL-protein interactions [55;76]. This would be relevant for numerous functions such as oxidative phosphorylation enzyme activity and regulating the degradation of CL. To exemplify, Xu et al. recently demonstrated that CL degradation in lymphoblasts was protected by associating with select inner membrane mitochondrial proteins whereas loss of this protection, as observed in subjects with Barth syndrome, led to increased CL degradation [55]. Moreover, monounsaturated fatty acids such as oleic acid decreased the rate of CL degradation and improved supercomplex formation [55]. Therefore, subsequent studies will need to evaluate how CL content and unsaturation regulate CL-protein interactions.

4.4 Implications for the design of mitochondrial CL-specific therapeutics

Many diseases are associated with the loss of (18:2)₄CL and aberrant remodeling of CL acyl chains, which leads to the loss of cristae and ultimately mitochondrial function [10]. A number of pharmaceutical approaches aimed at restoring mitochondrial bioenergetics are under pre-clinical development [77-79], and many of these implicate CL. Several cell-permeable peptides known as SS-31/Bendavia/Elamipretide and SS-20 bind CL and have shown efficacy in improving electron transport and mitochondrial function in animal models of heart and kidney disease [80-82]. In large-animal models of heart and kidney disease, treatment with the CL-targeting peptide Elamipretide improved mitochondrial energetics, lowered ROS, and restored CL content back to non-diseased levels [83;84]. Thus, development of therapeutics directed to restore CL levels and acyl chain composition has enormous potential across numerous disorders characterized by CL abnormalities.

4.4 Conclusion

The results demonstrate that (18:2)₄CL content and acyl chain composition differentially influence select membrane properties in biomimetic membranes. These studies set the foundation for future experiments on how CL concentration and acyl chain composition in model mitochondrial membranes influence domain size, stability, and in turn, diffusion and clustering of mitochondrial membrane proteins. Furthermore, the results have implications for the design of CL specific therapeutics and ultimately in the study of mitochondrial proteins that depend on the surrounding lipid environment for optimal function.

Highlights.

• In several diseases, cardiolipin (CL) content is lowered and acyl chains remodeled.

• CL content has a strong impact on physical properties of biomimetic membranes.

• Remodeling of (18:2)₄CL to (14:0)₄CL has a strong influence on membrane properties.

• (18:2)₄CL, (18:1)₄CL, and (22:6)₄CL behave similarly in complex monolayers.

• CL content and acyl chain composition differentially target membrane organization.