Effect of Alkyl Chain Length on Translocation of Rhodamine B n-Alkyl Esters across Lipid Membranes

Abstract

Voltage-dependent translocation of a series of cationic rhodamine B derivatives differing in n-alkyl chain length (ethyl, butyl, octyl, dodecyl, octadecyl) from one lipid monolayer to another was studied by measuring electrical current relaxation after a voltage jump on a planar bilayer phosphatidylcholine (PC) membrane. The rate of the translocation decreased in the following series of lipids: diphytanyl-PC > dioleyl-PC > diphytanoyl-PC > dierucoyl-PC. For all the lipids studied, the rate increased with lengthening of the hydrocarbon chain of the rhodamine derivatives, with the increase being most pronounced for the compounds having a short alkyl chain. The results could be well explained by involvement of molecule reorientations in the process of transmembrane flip-flop of the hydrophobic membrane-bound compounds. However, an impact of membrane dipole potential on the translocation rate could not be excluded, because the dipole potential could contribute to the energy barrier for translocation of the compounds located at different depths in the water-membrane interface. Based on the data obtained, a difference in the dipole potential of ester diphytanoyl-PC membranes with respect to ether diphytanyl-PC was estimated to be 108 mV, highlighting the contribution of a layer of oriented carbonyl groups of the lipids to the membrane dipole potential.

Introduction

Passive permeability of drugs through cellular membranes is an important characteristic of their mechanism of action that affects pharmacokinetics. The permeability of a solute through a lipid membrane from one aqueous phase to another includes two major steps: 1) a binding/release to the membrane and 2) a transmembrane diffusion. In the case of solutes having a high affinity to the lipid-water interface, the latter step is frequently called a flip-flop. The first step is well studied for a wide range of pharmacologically relevant substances and is determined by their hydrophobicity. The step is generally considered to determine the whole process of the permeability, which manifests itself in a good correlation between the permeability and a partition coefficient between octanol and water (1, 2, 3), or even better, correlation with the partition coefficient between 1,9-decadiene and water (4).

Unfortunately, a study of the transmembrane diffusion through lipid membranes is more difficult, and it is usually hard to separate this step from the whole process of permeability for a major part of the solutes. However, it was shown that the rate of the transmembrane diffusion decreased upon the increase in the cross-sectional area of the solutes (5) and upon the increase of the surface density of the lipids (6). An important practical question is the prediction of a rate of the transmembrane diffusion on the alkyl chain length for several classes of compounds, including fatty acids. It was shown in one series of works that the flip-flop rate of fatty acids is high (the characteristic time was ∼10 ms) and did not depend on the hydrocarbon chain length (7). In contrast to that, the rate of the flip-flop of the fluorescently labeled analogs of fatty acids decreased substantially upon an increase in the chain length (8). Although the reason for this apparent contradiction is not clear, it could be ascribed to the complexity of the system, including the presence of protonated and deprotonated forms of fatty acids, which could hinder the interpretation of the experimental data.

In this work, we applied a method of electrical current relaxation for the study of the rate of translocation of homologous series of penetrating cations based on rhodamine B having a different alkyl chain (C_nRB, n = 2–18, Fig. 1 A). The ethyl ester of rhodamine B (C₂RB), n-butyl ester of rhodamine B (C₄RB), n-octyl ester of rhodamine B (C₈RB), and n-dodecyl ester of rhodamine B (C₁₂RB) were synthesized, and n-octadecyl ester of rhodamine B (C₁₈RB) was commercially available. It was shown in our previous work that rhodamine 19 alkyl esters diffuse through lipid membranes in protonated and deprotonated forms, leading to proton shuttling through the membrane and a pronounced pH dependence of the translocation rate of its protonated form (9, 10). Compared to rhodamine 19, rhodamine B has two additional ethyl groups at two nitrogen atoms that prevent the protonation process and protonophoric activity (9). It has been shown in this work that even the most hydrophilic C₂RB (rhodamine B ethyl ester) binds effectively to lipid membranes, and its translocation can be easily measured via current relaxation upon voltage jump. This method was used extensively for the study of the transport of tetraphenylborate and other penetrating anions (11, 12, 13, 14). The translocation of C₁₈RB on planar lipid bilayer was studied previously in (15, 16), and the electrical relaxation correlated well with the fluorescent transients measured in the presence of asymmetrically added fluorescent quencher (15).

Figure 1.

(A) Chemical structures of C_nRB. (B) Time courses of electrical current through planar BLMs after application of a voltage jump from 0 to 100 mV (at t = 0, i.e., “on” response) and from 100 to 0 mV (at t = 26 s, i.e., “off” response) in the presence of 0.1 μM of C₈RB and their best fits by a monoexponentinal function with τ = 3.4 s and τ = 6.5 s are shown (red curves). Bilayers were formed from DPhPC. The solution was 10 mM MES, TRIS, 100 mM KCl (pH 7.4). To see this figure in color, go online.

Materials and Methods

Chemicals

C₁₈RB (rhodamine B octadecyl ester perchlorate) was purchased from Sigma-Aldrich (St. Louis, MO). A series of n-alkyl esters of rhodamine B (C_nRB) was prepared by acid-catalyzed esterification using method analogous to one described earlier (17).

All derivatives were synthesized by general procedure as indicated for the n-octyl ester of rhodamine B (C₈RB). In brief, rhodamine B (0.718 g, 1.5 mmol) was dissolved in octanol (6 mL, 3.8 mol). Concentrated sulfuric acid (0.5 mL) was added, and the mixture was heated at 130°C for 24 hr. The resulting solution was allowed to cool, and aqueous ethanol (100 mL) was added to precipitate the crude product, which contained a mixture of the required ester and unchanged rhodamine. The cooled mixture was decanted, and the residual solution was evaporated to dryness under reduced pressure. The octyl ester of rhodamine B was purified by column chromatography on silica gel (mesh 70–230) in solvent mixture containing dichloromethane and ethanol at a 3:1 ratio. Fractions containing pure target compound were collected and concentrated by rotary evaporation. After evaporation, n-octyl ester of rhodamine B (C₈RB) as red powder was obtained (383 mg, 46%). Liquid chromatography-mass detection (LC-MS): found 555.1, required for C₃₆H₄₇N₂O₃ 555.8.

The following esters of rhodamine B were synthesized by the abovementioned method: C₂RB, yield 57%, LC-MS: found 471,9, required for C₃₀H₃₅N₂O₃ 471.6; C₄RB, yield 61%, LC-MS: found 500.2, required for C₃₂H39N₂O₃ 499,7; C₁₂RB, yield 58%, LC-MS: found 612.1, required for C₄₀H₅₅N₂O₃ 611.9.

Analytical thin layer chromatography was made on aluminum plates precoated with silica gel (Silica Gel 60 F₂₅₄; Merck, Kenilworth, NJ). Column chromatography was performed on silica gel (Kieselgel 60; Merck), eluting with solvent mixture containing dichloromethane and ethanol at a 3:1 ratio. Molecular masses were measured by the LC-MS method on a Waters Acquity chromatograph with a tandem quadrupole detector (Waters, Milford, MA). All the solvents were distillated by the standard methods before use.

Electrical current across planar lipid bilayers

Planar bilayer lipid membranes (BLMs) were formed from a solution of 1,2-diphytanoyl-sn-glycero-3-phosphocholine (DPhPC), 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1,2-dierucoyl-sn-glycero-3-phosphocholine (DEPC), or 1,2-di-O-phytanyl-sn-glycero-3-phosphocholine (DPhytanylPC) in n-decane (all from Avanti Polar Lipids, Alabaster, AL). The membranes (18) were spread from a lipid solution in n-decane across a circular aperture (500 or 800 μm) in a polytetrafluorethylene septum, which separated two aqueous phases of a polytetrafluorethylene chamber. Membrane thinning was observed optically and electrically via the determination of membrane capacitance. The electric currents (I) were recorded under voltage-clamp conditions. Voltages were applied to BLMs with Ag-AgCl electrodes connected via agar bridges. The currents measured by means of a patch-clamp amplifier (OES-2; OPUS, Moscow, Russia) were digitized using NI-DAQmx (National Instruments, Austin, TX) and analyzed with a personal computer with the use of WinWCP Strathclyde Electrophysiology Software designed by J. Dempster (University of Strathclyde, Glasgow, UK). In these relaxation experiments, the voltage was switched from zero to some particular value of U at t = 0, and the current across the membrane (I(t)) started to decrease from the initial level I(0) to steady-state level I(∞). At the beginning of each experiment, we recorded the capacitance response of the unmodified membrane (control record of the current after voltage-jump). Hydrophobic cations were added from stock solutions in ethanol to the bathing solutions at both sides of the BLM and incubated for at least 10 min with constant stirring. The record in the presence of a hydrophobic cation was analyzed after subtraction of the control record. In most of the experiments, the solution contained 100 mM KCl, 10 mM Tris (pH = 7.4). All experiments were carried out at room temperature (23–25°C).

Results

Fig. 1 B shows a record of the current transients on BLM made from DPhPC in the presence of 0.1 μM C₈RB (top part of Fig. 1 B) and the applied voltage on the bottom part of the figure. Shown is a current curve after the subtraction of a control capacitance current in the absence of C₈RB, which has a substantial contribution only at t < 0.1 s. The transients are well fitted by monoexponential kinetics (red curves in Fig. 1 B) I(t) = I_∞ + (Ι₀ − I_∞) × exp(−t/τ) (I₀ is initial current, I_∞ is steady-state current) with characteristic time τ_on = 3.4 s (voltage jump from 0 to 100 mV) and τ_off = 6.5 s (voltage jump from 100 to 0 mV). The total charge transferred (integral under exponent) for the switching on and switching off of the voltage jump was equal 0.7 × 10⁻⁹ C and 0.68 × 10⁻⁹ C, correspondently.

Fig. 2 A shows C_nRB current transients upon application of V = 100 mV for the case of membranes made from 1,2-di-O-phytanyl-sn-glycero-3-phosphocholine (DPhytanylPC), i.e., a lipid having substantially lower dipole potential, which facilitates a translocation of cationic species (19, 20, 21). Of note, the dipole potential arising from the glycerol-ester region of the lipids, the hydrated polar head groups, and the dipoles of the terminal methyl groups of aliphatic chains can reach several hundreds of millivolts (22). The latter contribution to dipole potential is likely to vanish in lipid bilayers. The curves are fitted well by the equation I(t) = I_∞ + (Ι₀ − I_∞) × exp(−t/τ) with characteristic times 0.12, 0.063, 0.034, 0.032, and 0.026 s for C₂RB, C₄RB, C₈RB, C₁₂RB, and C₁₈RB, respectively. The dependences of τ on the applied voltage (up to 150 mV) for these compounds are shown in Fig. 2 B. The kinetics can be used to calculate the translocated charge from one side of the membrane to the other using the equation ΔQ = I(0)×τ. Fig. 2 C shows the dependences of ΔQ on the applied voltage for C_nRB at a concentration of 0.2 μM.

Figure 2.

(A) Time courses of electrical current through planar BLMs after application of a voltage jump from 0 to 100 mV (at t = 0) in presence of 0.2 μM C₂RB (blue curve), 0.2 μM C₄RB (black curve), 0.25 μM C₈RB (red curve), 0.2 μM C₁₂RB (gray curve), and 0.25 μM C₁₈RB (green curve). The solution was 10 mM MES, TRIS, 100 mM KCl (pH 7.4). (B) Corresponding voltage dependences of the characteristic time τ (mean ± SD for three independent experiments) are shown. Solid curves are best fits according to Eq. 2. (C) Potential dependences of translocated charge of C_nRB in the presence of 0.2 μM of the hydrophobic cations are shown. Solid lines are best fits according to Eq. 1. Bilayers were formed from DPhytanylPC. To see this figure in color, go online.

The experimental data are in reasonable accord with existing models of hydrophobic ion transport between two energy minima near the aqueous interfaces of the lipid membrane (11, 12, 23). These models predict that the equilibrium charge displacement and the translocation time should depend on the externally applied membrane potential V in the following manner:

$$\mathit{\Delta }Q=Q_0\tanh \left(z\beta u/2\right)$$ (1)

and

$$\tau =\frac{1}{k+2k_i\cosh \left(z\beta u/2\right)}\text{,}$$ (2)

where u = FV/RT is dimensionless voltage (F and R are Faraday and gas constants, T is absolute temperature); Q₀ is a total charge absorbed in each energy minimum; k is a rate constant of a binding/release of C_nRB to/from membrane surface; k_i is a translocation rate constant; β is a portion of the applied voltage that is involved in the translocation process; and z is an ion charge, z = 1. Nonunity value of the parameter β means that the ions are located at a certain distance from the membrane-water interface. The values of β and Q₀ were calculated by fitting curves on Fig. 2 C by Eq. 1, and rate constants k_i and k were derived from fitting curves on Fig. 2 B by Eq. 2. The fitting parameters are summarized in Table 1 (mean ± SD for three–five experiments). For all voltages studied, the characteristic times decreased upon the increase in the number of atoms in the hydrocarbon chain (n) for n = 2–8 and were practically independent of n in the range of n = 8–18. The parameters of Eq. 2 were sensitive to the changes in n, i.e., the hydrophobicity of the compounds. For n ≥ 8, I₀/I_∞ ≥ 50, meaning k ≪ k_i, and thus k can be neglected in Eq. 2. For C₂RB and C₄RB, I₀/I_∞ ≤ 50, the ratio decreased upon the increase in the C_nRB concentrations in aqueous solution, suggesting that the rate constants k and k_i are close to each other.

Table 1.

Parameters of C_nRB Transport in DPhytanylPC Membranes

n, Number of Hydrocarbon Atoms in CnRB	ki, s−1	k, s−1	β	Q0, nC/cm2
2	1.1 ± 0.3	1.8 ± 0.5	0.94 ± 0.02	34 ± 0.5
4	2.8 ± 0.4	1.0 ± 0.3	0.84 ± 0.04	53 ± 1
8	6.0 ± 1.0	ND	0.73 ± 0.09	56 ± 2
12	7.6 ± 0.9	ND	0.79 ± 0.04	57 ± 2
18	7.7 ± 1.2	ND	0.85 ± 0.09	15 ± 0.2

ND, not determined.

According to Table 1, the increase in n led to the increase in Q₀ only at small n (∼1.5-fold for n = 2 and n = 4), whereas further increases in n did not affect the binding under our experimental conditions (except for C₁₈RB, which had severalfold lower Q₀). It has been shown earlier in our laboratory that C18-alkyl ester of rhodamine 19 (C₁₈R1) forms aggregates and/or micelles in aqueous solution, which considerably hinders its binding to liposomes (24). Owing to structural similarity between C₁₈R1 and C₁₈RB, the decrease in the binding of C₁₈RB to the BLM was a result of its aggregation in aqueous solution. Of note, the translocation of C₁₈RB across the BLM has been studied earlier (15) via its addition directly into the membrane-forming solution. Similar values of Q₀ for C₄RB, C₈RB, and C₁₂RB can hardly be attributed to saturating concentrations of the hydrophobic cations in aqueous solution, because Q₀ increased with the increase in the bathing concentration of these rhodamine derivatives (data not shown). The adsorption constant γ = Q₀/(c × q × N_A) can be estimated as 1.8 × 10⁻³ cm for C₂RB, 2.8 × 10⁻³ cm for C₄RB, and ∼3 × 10⁻³ cm for C₈RB under these conditions (N_A is the Avogadro constant, q is the elementary charge, and c is the concentration in water solution). Our data can hardly be applied for the estimation of γ for C₁₂RB and C₁₈RB because of the marked decrease of their aqueous concentration owing to aggregation and the adsorption to the chamber. We observed this process by measuring the absorption spectra of aliquots of the solution bathing the BLM (data not shown).

Five compounds of the C_nRB series were also studied on membranes made from DPhPC, DOPC, or DEPC. Fig. 3 shows the dependence of τ on n upon the application of voltage jump from 0 to 100 mV for these four lipids. The dependences were similar to that observed for DPhytanylPC, i.e., the rate of the translocation increased with the increase in hydrophobicity exhibiting saturation in the range of n = 12–18. The characteristic time of relaxations increased 35–100 times in membranes made from lipids having ester bonds (DOPC, DPhPC, and DEPC) compared to ether bonds (DPhytanylPC), apparently due to substantially different value of the dipole potential at the membrane-water interface (Δφ_d), suggesting the impact of a layer of oriented carbonyl groups of the lipids in the dipole potential. The data shown in Fig. 3 enables us to estimate the Δφ_d values of DPhPC membranes compared to DPhytanylPC according to the following equation Δφ_d = (RT/F)log(τ_l/τ_ref), where τ_l and τ_ref are relaxation times for DPhPC and for a reference lipid (DPhytanylPC), respectively. Using the data for C₁₂RB and C₁₈RB, Δφ_d was 108 ± 7 mV (mean ± SD, n = 5). The values of the dipole potential for membranes formed from ester and ether lipids were measured previously by voltage-jump technique (19) and charge relaxation method (25, 26), and the estimated differences of ∼100 mV in the dipole potential were reported. Relaxation kinetics for all C_nRB (Fig. 3) decelerated with the increase in the chain length of lipid fatty acids from C₁₆ (DOPC) to C₂₂ (DEPC). In particular, τ_100mV increased approximately three times for C₁₂RB and C₁₈RB. It should be noted that the increase in the chain length of the fatty acid from C₁₆ to C₂₂ resulted in a decrease of k_i by a factor of approximately five in the case of tetraphenylborate and by a factor of 10 in the case of dipicrylamine (23). An equally directed change in the transmembrane translocation rate for hydrophobic anions and cations with an increase in the length of the lipid fatty acid tails apparently indicates either a change in the hydrophobic interaction of ions, with the lipid membrane having different thickness, or a change in the electrostatic Born energy. In the case of a noticeable change in the dipole potential of the membrane with an increase in the membrane thickness (from C₁₆ to C₂₂), we would observe opposite changes for penetrating anions and cations. It was shown previously that the magnitude of the Born barrier increases with the increase in the bilayer thickness (27).

Figure 3.

Dependence of characteristic time τ of the C_nRB-mediated current relaxation after application of a voltage jump from 0 to 100 mV on n. Membranes were made from DPhytanylPC (closed circles), DOPC (open squares), DPhPC (closed triangles), and DEPC (open circles).

It should be noted that the current relaxation curves for C_nRB were fitted well by monoexponentials for n ≥ 8 and for all lipids studied. However, for n ≤ 4 and membranes made from DOPC, DPhPC, and DEPC, the relaxation curves exhibited a deviation from monoexponential kinetics at long times, suggesting the involvement of some additional slow process. In these cases, the value of τ in Fig. 3 was estimated using initial relaxation on short times after the voltage jump. It can be assumed that the slow component was related to the diffusion process in the unstirred layers (28). Fig. 4 shows the current relaxation mediated by 0.5 μM C₄RB upon the voltage jump from 0 to 100 mV for DPhPC membrane. The initial monoexponential process related to the translocation of bound C₄RB molecules from one side of the membrane to another (fitting red curve τ = 7.1 s) was followed by a slow relaxation process lasting tens of seconds. A stirring of the solution bathing the membrane was turned on at t = 100 s for 7 s, which led to a substantial increase in the current (up to 12% of I₀, Fig. 4). The slow relaxation resumed after turning off the stirring (Fig. 4). It can be concluded that the slow component in the case of hydrophilic compounds was accounted for by the concentration polarization in the unstirred layers, which was observed previously for tetraphenylborate and other anions (28, 29).

Figure 4.

Time course of electrical current through planar bilayer lipid membrane after application of a voltage jump from 0 to 100 mV (at t = 0) in the presence of 0.5 μM C₄RB. BLMs were formed from DPhPC. A stirring of the solution bathing the membrane was turned on at t = 100 s for 7 s. The solution was 10 mM MES, 10 mM TRIS, 100 mM KCl (pH 7.4). The red curve is a monoexponential decay fitting the initial current relaxation. To see this figure in color, go online.

It was shown previously that the characteristic time of tetraphenylborate (and other anions) relaxation increased at high concentrations of the ions because of changes in the membrane surface potential induced by the ions bound to the membrane (30). Similar changes in characteristic time were observed for C_nRB. To avoid an influence of penetrating cations to the properties of BLM and to decrease the value of concentration polarization, we used the lowest C_nRB concentrations possible for reliable determination of the current relaxation process (less than 0.5 μM).

Discussion

The main conclusion from our experiments was the increase in the translocation rate constants of C_nRB with the growth of the hydrocarbon chain length, especially at low n = 2–8. This result was in contrast to the dependence of pyrene-labeled lipid flip-flop (31) as well as to the data of (15) on the translocation of indocarbocyanine dyes (DiI). It was shown that DiI-C12 with two dodecyl chains had about a two times higher translocation rate compared to DiI-C18 with two octadecyl chains (15). It was suggested that this deceleration in transmembrane diffusion upon the increase in the alkyl chain length was related to the growth of the size of the molecules, which is known to decrease the rate of diffusion, especially in a polymer matrix (32). It was shown later that the major input to the membrane permeability gives the cross area of the molecule rather than their size in general (5, 33). The cross-sectional area of C_nRB should be determined mainly by the rhodamine moiety, especially at low n, and from this point of view, the transmembrane diffusion of C_nRB should be independent of n. Therefore, our experimental data cannot be accounted for by this model.

To explain our data, one can use the idea of transmembrane permeation as a process of jumps over several energy barriers with the predominating central barrier as it was utilized for the description of the translocation of penetrating anions (11, 12, 23). In the absence of surface potentials, the total free energy of transferring a hydrophobic ion from solution to a given position in the membrane can be approximated as ΔG⁰ = ΔG⁰_Born + ΔG_Image + ΔG⁰_Dipole + ΔG⁰_Hydro (34), where ΔG⁰_Born is the Born energy of the ion, which is positive for all ions (negative or positive) and is inversely proportional to the ionic radius; ΔG_Image is the image energy contribution; ΔG⁰_Hydro is the hydrophobic energy of attraction; and ΔG⁰_Dipole, having opposite signs for positive and negative ions, emerges from the interaction of the ion charge with the membrane dipole potential and determines the much higher free-energy profile near the center of the membrane for cations rather than for anions of similar structure. The membrane dipole potential is a manifestation of a nonrandom orientation of the electric dipoles in lipid headgroups, fatty acid carbonyl groups, and ordered water molecules (19, 22). The value of the central barrier between narrow minima near the two interfaces should be independent of the hydrophobicity of homologs penetrating ions, as it is determined by the molecular structure of the penetrating ion (34). Thus, the increase in the hydrophobicity of C_nRB with an increase in n should lead to a decrease in the energy of binding to the membrane, leaving the translocation barrier unaltered, and should increase water-membrane partitioning of positively charged ions. This should result in increased water-to-water permeability for penetrating cations with longer acyl chains, in agreement with the general solubility-diffusion model. In the case of planar lipid bilayers, this phenomenon was well documented for monocarboxylic acids (35, 36) and other small nonelectrolytes (1, 37).

However, the binding of C_nRB to DPhytanylPC membranes was independent of n for n = 4–12 and decreased substantially at n = 18 (Q₀ values in Table 1). The characteristic times of the relaxation were close for C₈RB, C₁₂RB, and C₁₈RB and increased for C₄RB and C₂RB. One reason for the acyl-chain dependence of the k_i may be different locations of C_nRB at the membrane-water interface. One may assume that long-chain compounds are likely to sit deeper in the interface and a smaller fraction of the interfacial dipole potential difference contributes to a potential barrier. Conversely, more hydrophilic derivatives are poorly inserted into the hydrocarbon region, and the dipole potential could slow down their translocation more pronouncedly. However, according to Table 1, the values of the parameter β, reflecting the position of C_nRB at the membrane-water interface, were almost the same for C₄RB and C₁₈RB. On the other hand, the value of β was higher for C₂RB compared to the other rhodamines (p < 0.01 for C₄RB, C₈RB, and C₁₂RB and p = 0.125 for C₁₈RB), and therefore the changing of the contribution of the dipole potential to the magnitude of the barrier for different n cannot be completely excluded. Of note, it has been shown that the ionized carboxylic group of the free fatty acids (C₁₆–C₂₀) occupied a fixed position in the lipid bilayer independent of the chain length (38).

Considering other reasons for the dependence of the k_i on n, one could address the role of orientational and conformational motions in the translocation process in addition to the monodimensional movement along the normal to the lipid bilayer (39, 40, 41, 42, 43). Indeed, the flip-flop is by nature a rototranslational process that involves both reorientations and displacements across the bilayer. It was found in a theoretical study (42) that a small permeant ethanol strongly oriented on the surface of a lipid membrane so that the hydroxyl moiety interacted with the head groups, whereas the ethyl moiety was in contact with the lipid tails. The conspicuous orientational anisotropy of ethanol at the aqueous interface is suggestive of a complete rotation of the permeant as it crosses the hydrophobic interior of the membrane. Calculated values of side-chain rotational correlation times of three aromatic dipeptides demonstrated extremely large slowing with peptide membrane insertion relative to solution (43). It was shown recently that the intrinsic activity of dithioureidodecalin anion carriers is remarkably sensitive to alkyl substitution, passing through a maximum as the chain length increased (44). Molecular dynamics simulations showed that the addition of two hydrocarbon chains tended to turn transporter/complex, pulling the polar region away from the interface. In this case, the interactions with the interface are easier to break so that the transport is facilitated. However, chains that are too long can also slow transport by inhibiting movement, and especially reorientation, within the phospholipid bilayer. The flip-flop of the fluorescent compound having two long alkyl chains was shown to be decelerated in the case of the increase in the alkyl length (15, 31).

We assume that the insertion of the almost planar xanthene moiety of C_nRB is facilitated in the case of its orientation normal to the membrane plane and requires an additional free-energy penalty in the case of parallel orientation (a scheme in Fig. 5). Assuming that the orientation of the xanthene moiety of C_nRB is dependent on n, one can rationalize the increase in the rate of the flip-flop on the alkyl chain length. In the case of short chain length (hydrophilic C_nRB), the xanthene plane is located parallel to the plane of the membrane and requires rotational diffusion for the permeation through the membrane. It was shown recently that the xanthene moiety of rhodamine 6G lies flat at the decane-water interface (45). Longer alkyl chains could promote the immersion of the plane of the xanthene moiety and tilting the plane with respect to the interface’s normal (Fig. 5, right side), thus facilitating the penetration of the cation through the membrane interface and a subsequent flip-flop. Molecular dynamics simulations of rhodamine-labeled phospholipids in a dipalmitoylphosphatidylcholine bilayer (46) showed that the sulforhodamine moiety located at a polar membrane interface formed the tilt angle of 44 ± 8° to the bilayer normal. The dependence of the tilt angle on the hydrophobicity of adsorbed fluorescent dye was demonstrated also for di-4-ANEPPS and di-8-ANEPPS by measuring polarization with multiphoton microscopy. It was shown that more hydrophobic dye di-8-ANEPPS has a lower steady-state tilt in phospholipid membrane than di-4-ANEPPS (47).

Figure 5.

Schematic of the orientation of cationic rhodamine B derivatives having almost planar xanthene moiety and n-alkyl chain of different length (C_nRB) at the membrane-water interface. The membrane is shown in yellow, the xanthene moiety is shown as a flat rectangular box, and the alkyl chain is shown as a zig-zag line of different length. In the case of short chain length (hydrophilic C_nRB), the xanthene plane is located parallel to the plane of the membrane and requires rotational diffusion for the permeation through the membrane (left side). Longer alkyl chains could promote the immersion of the plane of the xanthene moiety and tilting the plane with respect to the interface’s normal (right side), thus facilitating the penetration of the cation through the membrane interface and a subsequent flip-flop. To see this figure in color, go online.

Deceleration of the permeant’s transmembrane diffusion due to the involvement of additional rotational diffusion must depend on the shape of the permeant. In this work, we observed a severalfold increase in the flip-flop rate for a series of C_nRB compounds having an almost planar xanthene group (Table 1). It has been observed previously in our group that the rate of flip-flop of anionic monocarborane having a spherical shape is equal to that of a conjugate with chlorine e6, BACE (48). These results are in line with the idea that in the case of permeation of a spherical molecule, the input of the rotational diffusion should be negligible. We have studied also the flip-flop of halogenated cobalt bis(dicarbollide), COSAN (49). The compound has two rigid hydrophobic semicages and a hydrophilic center (θ shape) and demonstrates the following row of the rate of the flip-flop decrease: COSAN-F₂ < COSAN < COSAN-Cl₂ < COSAN-Br₂ < COSAN-I₂. Because halogenation of tetraphenylborate substantially increased the translocation rate of the fluorinated and chlorinated analogs owing to the increase in the effective radius of molecules (13), the slowing of the translocation rate of COSAN-F₂ compared to COSAN was surprising. It was found that the adsorption on the lipid membrane molecules of COSAN-F₂ having a dipole moment affected the dipole potential of the lipid membranes (49). It was concluded that the process of COSAN-F₂ translocation exhibited an additional rotational barrier slowing the flip-flop rate of the transmembrane translocation.

We think that conclusions from our investigation of membrane translocation of the alkylated penetrating cations may be extended to another series of structurally homologous amphiphiles (50) and help to resolve the problem concerning the rate-limiting step for free fatty acids translocation (7, 8).

Author Contributions

T.I.R. and Y.N.A. designed research. T.I.R. performed research. G.A.K. synthesized new compounds. T.I.R., G.A.K., and Y.N.A. analyzed data. T.I.R. and Y.N.A. wrote the article.

Editor: Timothy Cross.