Introduction

α,ω-Dioic acids are formed in the liver cells as a result of ω-oxidation of the corresponding monocarboxylic fatty acids (Sanders et al. 2005; Reddy and Rao 2006; Wanders et al. 2011). Under normal physiological conditions, only 5–10 % of monocarboxylic fatty acids are metabolized through this pathway. However, the level of α,ω-dioic acids in liver cells substantially increases with some pathologies (obesity, starvation, diabetes mellitus), as well as various disorders of the monocarboxylic fatty acid metabolism (Reddy and Rao 2006; Wanders et al. 2011). In Reye’s syndrome, which is characterized by acute encephalopathy and infiltration of fat into liver cells (Tonsgard 1986), the level of α,ω-dioic acids in the blood serum of children rises to 55 % of total fatty acids. Most of these α,ω-dioic acids (up to 90 %) are long-chain acids (C16–C18), and their concentration in the serum can reach 0.5 mM (Tonsgard 1986). It has also been shown that ω-oxidation of free fatty acids (FFAs) in liver cells is significantly enhanced under the effect of some xenobiotics and chronic influence of ethanol (Kundu et al. 1991; Orellana et al. 1998). In addition, accumulation of α,ω-dioic acids in liver cells is known to occur in ground squirrels at the late stages of their hibernation (Nelson et al. 2009), although nothing is known about the biological role of the phenomenon and its consequences for the hibernating animal. It is also unclear how much the increase in the level of α,ω-dioic acids in the liver tissue, which is observed at various pathologies, is associated with the functional disorders of this vital organ.

At present, biologists consider a few types of cell death, which differ according to both morphological manifestations and biochemical mechanisms. In all cases, however, the central role is played by the mitochondria, which integrate many intracellular signal pathways leading to cell death (Malhi et al. 2010; Kroemer et al. 2007; Rasola and Bernardi 2011). One of the mitochondria-related steps in the complex mechanism of cell death is the transition in the nonspecific permeability of the inner mitochondrial membrane, which is induced by Ca2+. Following the permeability transition is the impairment of the energy-transducing functions of mitochondria (first of all, ATP synthesis) as well as the release of the so-called apoptogenic proteins—cytochrome c, apoptosis-inducing factor, endonuclease G, etc.—from the intermembrane space into the cytoplasm (Kroemer et al. 2007; Rasola and Bernardi 2011).

It is well known that free monocarboxylic fatty acids are effective natural inductors of nonspecific permeability transition in the inner mitochondrial membrane (Sultan and Sokolove 2001a, b; Mironova et al. 2004; Belosludtsev et al. 2005, 2006; Di Paola and Lorusso 2006). The “classical” mechanism of permeability transition implies that fatty acids interact with one of the components of the putative pore complex: the ADP/ATP antiporter. In this case, the effect of fatty acids manifests itself only in the presence of Ca2+ (other divalent cations are not effective) and is suppressed with CsA (Sultan and Sokolove 2001a). At the same time, it is established that palmitic and some other saturated fatty acids are able to open a nonspecific pore in liver mitochondria that is insensitive to CsA. In this case, Ca2+ can be replaced with other divalent cations. Unsaturated fatty acids cannot open this pore (Sultan and Sokolove 2001a, b; Mironova et al. 2004; Belosludtsev et al. 2005, 2006). The formation of such a pore was also demonstrated on liposomal membranes (Agafonov et al. 2003, 2007). This enabled us to suggest that the CsA-insensitive pore induced by palmitic and other saturated fatty acids in the inner mitochondrial membrane had a lipid nature. The opening of this pore seems to occur by the mechanism of chemotropic phase transition in the lipid bilayer; it is based on the formation and subsequent segregation of Ca2+ or Sr2+ complexes with fatty acids into solid membrane domains (Agafonov et al. 2007).

We have recently shown that α,ω-dioic acids are able to open a “non-classical” CsA-insensitive pore in Ca2+- or Sr2+-loaded liver mitochondria, with α,ω-hexadecanedioic acid (HDA) being the most effective inducer. Necessary requirements for the pore opening is energization of the organelles and presence of Ca2+ or Sr2+ in their matrix (Dubinin et al. 2013, 2014).

In the present work, we continue to study the role of α,ω-dioic acids as inducers of the CsA-insensitive Ca2+-dependent nonspecific permeability transition in liver mitochondria. In addition, to elucidate the molecular mechanism of this phenomenon, we examine the effect of HDA on the membrane of liposomes in the presence of Ca2+. The effects of HDA are compared to the effects of α,ω-tetradecanedioic acid (TDA), which has a shorter acyl chain. The data obtained are considered evidence of the ability of α,ω-dioic acids to induce Ca2+-dependent CsA-insensitive permeability transition in both mitochondrial and liposomal membranes.

Materials and methods

Materials

Medium components, all inorganic chemicals, fatty acids, sulforhodamine B, tetraphenyl phosphonium, cyclosporn A and lecithin were purchased from Sigma-Aldrich; DPPC was purchased from Avanti Polar Lipids.

Preparation of mitochondria

Experiments were conducted on mitochondria isolated from the liver of Wistar rats (animal weight was 210–250 g) using a convenient technique of differential centrifugation (Markova et al. 1999). The isolation medium contained 250 mM sucrose, 1 mM EGTA and 5 mM Hepes-Tris buffer (pH 7.4). The mitochondrial protein concentration was determined by the biuret method with BSA used as standard.

Mitochondrial swelling was detected by changes in the optical density of the mitochondrial suspension (A) at 540 nm using a KFK-3-01 spectrometer (ZOMZ, Russia) in a 4-ml cell and at 25 °C. The incubation medium contained 200 mM sucrose, 20 mM KCl, 20 µM EGTA, 5 mM succinic acid and 10 mM Hepes-Tris buffer (pH 7.4). Also, 2 μM rotenone was added to the cuvette just before measurements.

The concentration of Ca2+ ions in the incubation medium was determined with a Ca2+-selective electrode. Δψ was estimated by following the distribution of tetraphenylphosphonium (TPP+) across the mitochondrial membrane; in this case, the incubation medium was supplemented with 1.6 μM TPP+. The concentration of TPP+ was measured with a TPP+-sensitive electrode (Kamo et al. 1979). The changes in TPP+ and Ca2+ concentrations were recorded simultaneously in a 1.5-ml cell at constant stirring and thermostating (25 °C) using a Record-4 USB original multi-channel electrometrical system.

Preparation of large unilamellar vesicles

Large unilamellar vesicles (LUVs) were obtained by a conventional extrusion technique. Dry egg phosphatidylcholine (lecithin) (0.75 mg) was hydrated in 0.75 ml of buffer for several hours with periodic stirring in a vortex mixer. The buffer contained 10 mM Tris-HCl (pH 8.5), 40 mM KCl and 50 µM EGTA. After five cycles of freezing/thawing at −20/+30 °C, the suspension of multilamellar liposomes was pressed 11 times through a 0.1-µm polycarbonate membrane using an Avanti microextruder (Avanti Polar Lipids, Birmingham, AL). All operations (excluding the freezing/thawing procedure) were carried out at room temperature.

Measurement of permeabilization of SRB-loaded liposomes

SRB-loaded LUVs were prepared from egg phosphatidylcholine by a procedure similar to that described above except that (1) the buffer for lipid hydration contained 50 mM SRB instead of 40 mM KCl and (2) after extrusion liposomes were applied on a Sephadex G-50 column to remove the external SRB. The buffer for gel filtration contained 10 mM Tris-HCl (pH 8.5), 50 µM EGTA and 40 mM KCl.

The release of SRB from LUVs was detected by the increase in fluorescence due to the dissociation of SRB excimers after the dilution of dye in the external medium. SRB fluorescence was measured at 25 °C using an Ocean Optics USB 2000 spectrofluorimeter (excitation wavelength, 565 nm; emission wavelength, 586 nm). The fluorescence of SRB was found to be stable for hours and had a linear dependence on the SRB concentration in the range 10−8–10−6 M. In our experiments, SRB-loaded LUVs were added to 2 ml of buffer (10 mM Tris-HCl, 50 μM EGTA and 40 mM KCl, pH 8.5), and their fluorescence was measured before and after various additions. Each experiment was concluded with the final estimation of maximal fluorescence observed upon the total release of SRB from LUVs, the latter being achieved by the addition of 0.1 % Triton X-100 (TX-100). The quantity of LUVs to be added to a sample was adjusted so that the maximal fluorescence was always at the same level. The molar phospholipid concentration in samples measured by the modified Bartlett technique (Bartlett 1959; Kates 1972) was 30–50 μM. Free fatty acids were added to samples as ethanol solutions, with the final concentration of ethanol never exceeding 1 %. At this concentration, ethanol did not cause any release of SRB from LUVs. The release of SRB from LUVs upon the effect of an acting factor (i.e., a chemical or a combination of chemicals) was calculated as a percentage of total SRB entrapped in the LUVs: \(R = \frac{{k_{\text{F}} F_{\text{F}} - F_{\text{B}} }}{{0.88 \times F_{\text{T}} - F_{\text{B}} }} \times 100\,\%\), where R is the release of SRB from the LUV induced by an acting factor; F B is the base fluorescence level observed after the addition of LUVs; F F is the fluorescence after the effect of an acting factor; F T is the fluorescence upon the total release of SRB from LUVs after the addition of TX-100; k F is the correction coefficient for F F—it was used when the fluorescence of SRB was affected by the acting factor itself (in most cases, kF was equal to 1); 0.88 is a value of the correction coefficient for TX-100, which was found to increase SRB fluorescence.

Acoustic measurements

The parameters of the main phase transition in the membrane of dipalmitoylphosphatidylcholine (DPPC) LUVs were measured with a temperature-scanning differential ultrasonic fixed-length interferometer (Kharakoz et al. 2007; Astashev et al. 2014). The measured quantity was the relative change in sound velocity and absorption caused by the dispersed lipid of the concentration, c: \(\left[ {\text{U}} \right] \, = \, \left( {u - u_{0} /u_{0} c} \right)\) and \(\left[ {\text{A}} \right] \, = (\alpha \lambda - \alpha_{0} \lambda_{0} )/c\). Scan rates were ∼0.3 K/min on cooling runs. LUVs were obtained as described above except that (1) DPPC was used instead of egg phosphatidylcholine, (2) the concentration of DPPC in samples was higher (4.07 mM), and (3) all operations were carried out at 45 °C. The medium contained 10 mM Tris-HCl buffer (pH 8.5), 50 µM EGTA and 40 mM KCl. HDA was added to the dispersion of DPPC as an ethanolic solution.

Dynamic light scattering

Particle size of the LUV was measured by dynamic light scattering at 25 °C Zetasizer Nano ZS (Malvern Instruments Ltd.). The back-scattered light from a 4-mW He/Ne laser (632.8 nm) was collected at an angle of 173°. LUVs were prepared as described above for the preparation of LUV particles. The molar phospholipid concentration in samples was 50 μM. The medium contained 10 mM Tris-HCl buffer (pH 8.5), 50 μM EDTA and 40 mM KCl. HDAs were added to samples as ethanol solutions, with the final concentration of ethanol never exceeding 1 %. The acquisition time for a single autocorrelation function was 100 s. The resulting autocorrelation functions are averaged values from ten measurements. The volume-weighted size distributions were calculated using the following parameters: the refractive index of the solution, 1.330; refractive index of particles, 1.459 (refractive index of lecithin); a viscosity value of the solution, 0.8882 cP.

Statistical analysis

The data were analyzed using the Graph Pad Prizm 5 and Excel software and presented as mean ± SEM of 3–5 experiments. Statistical differences between data points were determined by a two-tailed t test.

Results and discussion

In the experiments, we used HDA as a representative of the homologous series of α,ω-dioic acids. It is known that HDA is formed in the cells of liver as a result of ω-oxidation of palmitic acid (Ferdinandusse et al. 2004), one of the most abundant natural fatty acids (Wojtczak and Schönfeld 1993). As was shown earlier (Dubinin et al. 2013) and is confirmed in the present work (Fig. 1), the addition of HDA to the suspension of Ca2+-loaded mitochondria leads to a substantial decrease of absorbance in both the presence and absence of CsA. This indicates swelling of the organelles as a result of the CsA-insensitive transition in the nonspecific permeability of the inner membrane (Sultan and Sokolove 2001a, b). In addition, the permeability transition induced by HDA in Ca2+-loaded liver mitochondria leads to the efflux of Ca2+ from the mitochondrial matrix and the fall of the membrane potential (Fig. 2).

Fig. 1
figure 1

Effect of HDA (a), Ca2+ (b) and their combined effect on the kinetics of light-scattering changes of the liver mitochondrial suspension in the presence (c) or absence (d) of 1 µM CsA. The incubation medium contained 200 mM sucrose, 20 mM KCl, 20 µM EGTA, 5 mM succinic acid and 10 mM Hepes-Tris (pH 7.4). The concentration of mitochondrial protein was 0.8 mg/ml. Additions: 200 µM calcium chloride (Ca2+); 20 µM HDA; 1 µM CsA (not shown). Similar results were obtained in three independent experiments

Full size image
Fig. 2
figure 2

Changes in the Δψ (dashed line) and Ca2+ concentration (solid line) in response to the addition of Ca2+ and HDA. The concentration of mitochondrial protein is 1.5 mg/ml. The medium composition is the same as in Fig. 1

Full size image

It was shown earlier that, apart from inducing a Ca2+-dependent permeability transition in mitochondria (Belosludtsev et al. 2005), palmitic and some other saturated monocarboxylic fatty acids are able to permeabilize artificial membranes (Agafonov et al. 2003, 2007). In this work, we present data showing that, like its predecessor palmitic acid, HDA can induce a leak of the fluorescent dye SRB from unilamellar liposomes upon the addition of Ca2+ (Fig. 3). However, the kinetics of the HDA-induced SRB release differ from those observed in the case of palmitic acid. As shown in Fig. 3, the addition of PA and Ca2+ to liposomes results in a sharp increase in the SRB fluorescence. Fluorescence jumps to a new stable level immediately after the addition of Ca2+. In case of HDA-containing liposomes, the addition of Ca2+ leads to a slow growth of fluorescence, and it takes a few minutes until it reaches the plateau. It should be noted that alone neither fatty acids nor Ca2+ caused a leak of SRB from liposomes (Fig. 3).

Fig. 3
figure 3

Release of SRB from lecithin liposomes upon the addition of fatty acids (FAs) and Ca2+. Medium composition: 10 mM Tris-HCl; 50 μM EGTA; 40 mM KCl (pH 8.5). Common additions: 0.1 % TX-100. Dashed line 50 μM PA + 1 mM CaCl2 + 0.1 % TX-100, solid line 50 μM HDA + 1 mM CaCl2 + 0.1 % TX-100, dotted line 1 mM CaCl2 + 0.1 % TX-100

Full size image

The dependences of the SRB release from liposomes on the concentration of HDA and Ca2+ are given in Figs. 4 and 5 respectively. As seen in the figures, the release depends on the concentration of both HDA and Ca2+, reaching maximal values at concentrations of HDA about 300 µM and Ca2+ of about 2 mM. Thus, as an inducer of permeability changes, HDA is second to palmitic acid. We also compared the effect of HDA with that of TDA, which is shorter than HDA by two carbons. As seen in Fig. 4, TDA is far less effective than HDA regarding the ability to induce Ca2+-dependent membrane permeabilization. Hence, as the number of carbons in its acyl chain decreases, the effectiveness of an α,ω-dioic acid drops significantly. Earlier, we showed a similar dependence on the acyl chain length of α,ω-dioic acids in regard to their effectiveness as inducers of the permeability transition in liver mitochondria (Dubinin et al. 2013). The longer the acyl chain of a fatty acid is, the higher its lipid/water partition coefficient. One can, therefore, suppose that the effectiveness of Ca2+-dependent permeabilization of both liposomal and mitochondrial membranes will depend on how fast and easy the process of distribution of a fatty acid in the lipid bilayer is.

Fig. 4
figure 4

Dependence of SRB release from lecithin liposomes on the concentration of FA. Medium composition: 10 mM Tris-HCl; 50 μM EGTA 40 mM KCl (pH 8.5). Common additions: 0.1 % TX-100; 1 mM CaCl2. Dashed line PA; solid line HDA; dotted line TDA. The data are represented as mean ± SEM (n = 3)

Full size image
Fig. 5
figure 5

Dependence of SRB release from lecithin liposomes on the concentration of Ca2+. Medium composition: 10 mM Tris-HCl; 50 μM EGTA; 40 mM KCl (pH 8.5). Common additions: 0.1 % TX-100; 50 μM PA (or HDA). Dashed line PA; solid line HDA. The data are represented as mean ± SEM (n = 3)

Full size image

As is known, the interaction of fatty acids with Ca2+ depends on the pH of the buffer. In an alkalescent medium, fatty acids are presumably in the anion form, which will facilitate their interaction with Ca2+ (Mironova et al. 2001). As shown earlier, palmitic acid, which binds Ca2+ with high affinity, mainly causes Ca2+-dependent membrane permeabilization in the alkalescent region (pH 7.5–8.5) (Agafonov et al. 2003). As seen in Fig. 6, the HDA/Ca2+-induced release of SRB from liposomes also depends on the pH of the medium. At the same time, the maximal HDA/Ca2+-induced SRB release was observed in the alkaline region (pH 9.0–10.0). The difference in the optimal pH value for the effects of mono- and dioic fatty acids is probably determined by the difference in their pKa.

Fig. 6
figure 6

pH-dependence of SRB release from liposomes induced by the sequential addition of HDA and Ca2+. Medium composition, common additions and the addition order are as in Fig. 3. Concentration of HDA-50 μM, CaCl2-1 mM. The data are represented as mean ± SEM (n = 3)

Full size image

The data obtained on liposomes and mitochondria indicate that, interacting with Ca2+ ions, HDAs are able to permeabilize membranes, and these permeability changes have a lipid nature. As in the case with monocarboxylic fatty acids, the pH optimum of membrane permeabilization lies in the alkaline region, and the effectiveness of dioic fatty acids also grows as their chain length increases. On the other hand, different kinetics of palmitate/Ca2+-and HDA/Ca2+-induced permeabilization of liposomes and mitochondria allow us to suppose that the mechanisms of permeabilization are different for mono- and dioic fatty acids.

As was shown in the earlier works (Agafonov et al. 2003, 2007), palmitate/Ca2+-induced permeabilization of mitochondria and liposomes is based on the phenomenon of chemotropic phase transition. It was demonstrated that, being added to the suspension of DPPC liposomes, palmitic acid increased the temperature of the lipid phase transition, whereas the subsequent addition of Ca2+ returned the phase transition point to the initial value (Agafonov et al. 2007; Belosludtsev et al. 2014). In the present work, we examined the effect of HDA and Ca2+ on the phase state of liposomal membranes using the method of ultrasound interferometry. Figure 7 shows the temperature dependence of specific sound velocity and specific sound absorption by DPPC in the presence of HDA and Ca2+. The sharp increase in absorption and decrease in velocity reflect a change in the phase state of lipid in the membrane (raising temperature results in the transition from gel to liquid-crystalline state). The S-shaped region of the specific sound velocity curve (Fig. 7b) indicates the main (temperature) phase transition. For DPPC, the temperature of gel-to-liquid-crystal phase transition is 41.5 °C. HDA increases the temperature of the lipid phase transition to 43.5 °C, making the membrane more rigid. This effect of HDA is similar to that of palmitic acid. At the same time, the addition of Ca2+ to HDA-containing liposomes does not cause changes in the phase state of the membrane—in contrast to what was observed for palmitic acid. After the addition of Ca2+, the shapes of the curves remain the same as they were in the presence of HDA alone. Hence, it is unlikely that complexes of HDA with Ca2+ form solid domains in the membrane. Therefore, HDA/Ca2+-induced permeabilization of mitochondria and liposomes seems to be based on a mechanism that differs from the mechanism of chemotropic phase transition described earlier for palmitic acid.

Fig. 7
figure 7

Effect of Ca2+ on the phase behavior of a DPPC/HDA mixture. a Temperature dependence of specific sound absorption in the cell containing DPPC liposomes. b Temperature dependence of specific sound velocity in the cell containing DPPC liposomes. The points of solidus (S) and liquidus (L) are defined as the points of maximal curvature. Solid line liposomes (4.07 mM DPPC); dashed line liposomes (4.07 mM DPPC) + 1 mM HDA; dotted line liposomes (4.07 mM DPPC) + 1 mM HDA + 1 mM Ca2+

Full size image

Permeabilization of lipid membranes by HDA and Ca2+ could be explained by the detergent effect of HDA, since its working concentration is relatively high. In this case, a part of the liposomes should be dispersed. To test this supposition, we examined the effect of HDA and Ca2+ on the size of vesicles by the method of dynamic light scattering. As seen in Fig. 8, the average hydrodynamical diameter of control liposomes under those conditions was 140 nm. The addition of HDA to the suspension of liposomes did not result in the change of the average size of vesicles. However, the subsequent addition of Ca2+ led to a substantial (up to 520 nm) increase of the average hydrodynamical diameter of liposomes. It should be noted that the addition of 1–2 mM Ca2+ to the suspension of control liposomes (without HDA) did not cause changes in their size, which corresponds to the literature data on the effect of Ca2+ on fusion and aggregation of lipid vesicles (Papahadjopoulos et al. 1990). The addition of the non-ionic surfactant Triton X-100 resulted in the appearance of small vesicles (about 5 nm) (data not shown).

Fig. 8
figure 8

The distribution of liposomes over their size. Medium composition: 10 mM Tris-HCl; 50 μM EGTA; 40 mM KCl (pH 8.5). Dotted line liposomes (50 μM lecithin); dashed line liposomes (50 μM lecithin) + 75 μM HDA; solid line liposomes (50 μM lecithin) + 75 μM HDA + 1 mM Ca2+

Full size image

The data obtained indicate that the cause of HDA/Ca2+-induced SRB release from liposomes may be aggregation and/or fusion of vesicles. It is known that some amphiphilic compounds can initiate fusion and aggregation of liposomes in the presence of Ca2+. For example, unsaturated fatty acids were shown to stimulate fusion of liposomes in the presence of high Ca2+ concentrations (Wilschut et al. 1992; Belosludtsev et al. 2014). The results of the present work indicate that HDA is able to induce Ca2+-dependent aggregation/fusion and permeabilization of liposomes as well. It is known that the processes of fusion and aggregation depend on the lipid composition of the membrane and are often accompanied by destabilization of the bilayer and the appearance of pre-fusion structures (a semi-fusion structure, fusion pore, etc.) (Chernomordik and Kozlov 2003, 2008; Lentz et al. 2000; Jackson and Chapman 2008). It is possible that these rearrangements of the liposomal membrane lead to the release of SRB from liposomes.

It can be suggested that the mechanism of HDA/Ca2+-dependent permeability transition described above is realized in liver mitochondria. Earlier we established (Dubinin et al. 2013, 2014) that the necessary condition for the induction of CsA-insensitive permeability transition in mitochondria by HDA is the accumulation of Ca2+ in the matrix of the organelles. It seems that the formation of HDA/Ca2+ complexes in the inner monolayer of the inner membrane may lead to its spontaneous bending and protrusion toward the outer membrane. As a result, the two membranes may come into close contact and fuse, with the formation of a fusion pore and the subsequent swelling of mitochondria. It is possible that HDA and Ca2+ would induce fusion of the outer and inner mitochondrial membranes in the region of contact sites, where the two membranes are close to each other.

Conclusions

Summarizing the data obtained, one can make the following conclusions.

  1. 1.

    The experiments on liposomes showed that, like PA, HDA and TDA are able to induce a Ca2+-dependent permeabilization of liposomes loaded with SRB. One can, therefore, suppose that the Ca2+-dependent, CsA-insensitive pore induced by α,ω-dioic acids in the inner mitochondrial membrane has a lipid nature. At the same time, different kinetics and different concentration dependences of PA/Ca2+- and HDA/Ca2+-induced permeabilization of liposomes and mitochondria suggest that the mechanisms of permeabilization for mono- and dioic fatty acids are also different.

  2. 2.

    Using the method of ultrasound interferometry, we found that HDA/Ca2+-induced permeabilization of liposomes was not accompanied by changes in the phase state of the membrane—in contrast to what was observed in the case of PA/Ca2+-induced permeabilization. Therefore, the mechanism that underlies HDA/Ca2+-induced permeabilization is different from the mechanism of chemotropic phase transition described for PA.

  3. 3.

    Using the method of dynamic light scattering, we revealed that the combined action of HDA and Ca2+ resulted in the increase of the hydrodynamic diameter of liposomes, which was apparently accompanied by the formation of short-living lipid pores (fusion pores). It seems that in liver mitochondria, an HDA/Ca2+-induced permeability transition may be underlain by the fusion of the outer and inner mitochondrial membranes, which would result in the formation of fusion pores.