Regulation of connexin36 gap junction channels by n-alkanols and arachidonic acid

Abstract

We examined junctional conductance (g_j) and its dependence on transjunctional voltage in gap junction (GJ) channels formed of wild-type connexin36 (Cx36) or its fusion form with green fluorescent protein (Cx36-EGFP) transfected in HeLa cells or endogenously expressed in primary culture of pancreatic β-cells. Only a very small fraction (∼0.8%) of Cx36-EGFP channels assembled into junctional plaques of GJs were open under control conditions. We found that short carbon chain n-alkanols (SCCAs) increased g_j, while long carbon chain n-alkanols resulted in full uncoupling; cutoff is between heptanol and octanol. The fraction of functional channels and g_j increased several fold under an exposure to SCCAs, or during reduction of endogenous levels of arachidonic acid (AA) by exposure to fatty acid-free BSA or cytosolic phospholipase A2 inhibitors. Moreover, uncoupling caused by exogenously applied AA can be rescued by BSA, which binds AA and other polyunsaturated fatty acids (PUFAs), but not by BSA modified with 1,2-cyclohexanedione, which does not bind AA and other PUFAs. We propose that under control conditions, Cx36 GJ channels in HeLa transfectants and β-cells are inhibited by endogenous AA, which stabilizes a closed conformational state of the channel that leads to extremely low fraction of functional channels. In addition, SCCAs increase g_j by interfering with endogenous AA-dependent inhibition, increasing open probability and the fraction of functional channels.

Key points

• Under control conditions only one of ∼128 connexin (Cx)36 gap junction channels assembled in junctional plaques are open at a given time. This ratio was increased several times by short carbon chain n-alkanols and BSA, and reduced by long carbon chain n-alkanols.

• Bovine serum albumin (BSA) increased junctional conductance (g_j) by removal of polyunsaturated fatty acids (PUFAs) including arachidonic acid (AA) from the plasma membrane of HeLaCx36-EGFP cells. BSA modified with 1,2-cyclohexanedione, which does not bind PUFAs, did not affect g_j.

• A primary culture of pancreatic β-cells, expressing solely Cx36, shows similar properties as HeLa transfectants, i.e. g_j increases under an exposure to BSA and hexanol, while decanol and nonanol caused full uncoupling.

• Methyl arachidonyl fluorophosphonate (MAFP) and thapsigargin, inhibitor and activator of AA synthesis, increased and reduced g_j, respectively.

• The g_j-enhancing effect of hexanol did not change during combined application with MAFP and BSA, whereas AA and thapsigargin reduced the potentiating effect of hexanol.

Introduction

Connexins (Cxs) are members of a large family of membrane proteins that oligomerize into hexamers called connexons or hemichannels (HCs). Docking of two HCs from cells in contact results in the formation of a gap junction (GJ) channel with a pore diameter of ∼1.3 nm connecting cytosols of neighbouring cells for direct electrical and metabolic cell–cell communication (Harris, 2007). GJs formed of different Cx isoforms exhibit differences in single GJ channel conductance (γ), perm-selectivity and dependence of junctional conductance (g_j) on transjunctional voltage (V_j; Gonzalez et al. 2007). Although sensitivity and kinetics of V_j-gating depend on the Cx type, homotypic GJs show a symmetric g_j–V_j dependence, which has been ascribed to the presence of fast and slow V_j-sensitive gates in each apposed/docked HC (aHC; Bukauskas & Verselis, 2004; Paulauskas et al. 2012). Typically, the fast gate exhibits ∼1 ms or faster gating transitions between the open state with conductance, γ_open, to the residual state, γ_res. The slow gate is characterized by slower kinetics (∼10 ms) between open and fully closed states. In addition to V_j sensitivity, GJ channels are sensitive to intracellular H⁺, cationic divalents, posttranslational modifications and a variety of chemical reagents (Harris, 2001). The mechanisms by which these factors exert their effect on GJs remain unclear, although H⁺ and Ca²⁺ ions may both act through the slow V_j gate (Peracchia, 2004). It was proposed that chemical uncoupling is performed by the gating element of the slow gate, which is triggered by different sensorial elements/Cx-domains specific for voltage as well as chemical reagents (Bukauskas & Verselis, 2004).

Here, we examined modulation of functional efficiency, defined as the fraction of channels that are functional, g_j and V_j-gating of Cx36 GJ channels by n-alkanols with different carbon chain (C-chain) lengths from pentanol to decanol, arachidonic acid (AA) and compounds that increase or reduce its endogenous production. We demonstrated that only a very small fraction of Cx36 channels (∼0.8%) are open under control conditions. This fraction increased several fold under application of pentanol, hexanol or heptanol, representatives of short C-chain n-alkanols (SCCAs), while it decreased under application of octanol, nonanol or decanol, representatives of long C-chain n-alkanols (LCCAs). Cell–cell coupling substantially increased during exposure to fatty acid-free BSA, which efficiently extracts polyunsaturated fatty acids (PUFAs), including AA, from the plasma membrane, but not by BSA modified with 1,2-cyclohexanedione (CHD), which has greatly reduced binding of PUFAs. Moreover, an increase of endogenous AA production by thapsigargin reduced g_j, while reduction of its production by inhibition of cytosolic phospholipase A2 (cPLA2) with methyl arachidonyl fluorophosphonate (MAFP) increased g_j. We propose that under control conditions, levels of endogenous PUFAs, especially AA, hold Cx36 GJ channels preferentially in a closed state. Thus, reducing AA concentration in the plasma membrane with BSA or MAFP increases cell–cell coupling, while increasing AA concentration with thapsigargin decreases coupling.

Methods

Cell lines and culture conditions

Experiments were performed on HeLa (human cervix carcinoma) cells stably transfected with wild-type Cx36 (Cx36) or its fusion form with green fluorescent protein tagged to the C-terminus (Cx36-EGFP). Stable cell lines of HeLaCx36 were obtained as described earlier (Teubner et al. 2000). HeLaCx36-EGFP transfectants were obtained by using pCx36EGFP-P plasmid, and stable cell lines were selected based on EGFP fluorescence visible as junctional plaques (JPs). Cells were maintained in DMEM medium containing 10% fetal bovine serum (FBS), penicillin/streptomycin mix (100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin; Gibco Laboratories) and puromycin 1 μg ml⁻¹. HeLa cell lines carrying wild-type mCx30.2, Cx36, Cx45 and Cx47, and a Novikoff cell line endogenously expressing Cx43 (Meyer et al. 1992) were kindly provided by Dr. K. Willecke and Dr. R. Johnson.

Pancreatic β-cell isolation and culture

Ethical approval in all animal experiments was provided by the Institutional Animal Care and Use Committee at Albert Einstein College of Medicine, and conforms to the Guide for the Care and Use of Laboratory Animals (National Institute of Health). Dispersed β-cell cultures were obtained by isolating CD1 mouse islets of Langerhans by collagenase, as previously described (Jarchum et al. 2008), and subsequent dissociation with trypsin. Briefly, mice were killed by asphyxiation with CO₂. Then the pancreas was perfused through the common bile duct with ∼2 ml of cold Hank's balanced salt solution (HBSS) containing 0.6 mg ml⁻¹ collagenase P (Roche, Indianapolis, NJ, USA), and after removal was placed in HBSS containing collagenase, and incubated for 15 min at 37°C. Then, the digested pancreas was mechanically dispersed by gentle pipetting with HBSS enriched with 2% FBS. After centrifugation, the supernatant was removed and 10 ml of new HBSS with 2% FBS was added and gently mixed with the pellet. Washed islets were resuspended in HBSS with 10 μg ml⁻¹ DNAse I (Worthington Biochemical, Lakewood, NJ, USA) to avoid aggregation. Islets free of exocrine tissue were handpicked using a glass pipette and placed in a dish with RPMI-1640 medium. Each pancreas yielded approximately, 100–200 islets, which were then trypsinized for 10 min (0.25%; Invitrogen). Dissociated β-cells were seeded onto glass coverslips placed in culture dishes containing RPMI-1640 medium supplemented with 10% FBS, 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin (GIBCO Laboratories). Cells remained in a 5% CO₂ incubator for at least 8 h before electrophysiological and imaging recordings.

Electrophysiological recordings

For simultaneous electrophysiological and fluorescence recordings, cells grown on coverslips were transferred to an experimental chamber mounted on the stage of an inverted Olympus IX-70 microscope equipped with an imaging system. Experiments were performed in a modified Krebs–Ringer (MKR) solution containing (in mm): NaCl, 140; KCl, 4; CaCl₂, 2; MgCl₂, 1; glucose, 5; pyruvate, 2; Hepes, 5 (pH 7.4). Patch pipettes were filled with a solution containing (in mm): KCl, 140; NaAsp, 10; MgATP, 2; MgCl₂, 1; CaCl₂, 0.2; EGTA, 2; Hepes, 5 (pH 7.2, free [Ca²⁺]_i= 5 × 10⁻⁸ m). Cells were perfused with MKR solution at room temperature. g_j was measured in selected cell pairs using dual whole-cell patch-clamp. V_j was induced by stepping the voltage in cell-1 (ΔV₁) and keeping the voltage in cell-2 constant, V_j=ΔV₁. Junctional current (I_j) was measured as the change in current in the unstepped cell-2, I_j=ΔI₂. Thus, g_j was obtained from the ratio, −I_j/V_j, where the negative sign indicates that I_j measured in cell-2 is oppositely oriented to the one measured in cell-1. To minimize effects of series resistance on g_j measurements (Wilders & Jongsma, 1992), we maintained the pipette resistances below 3 MΩ. Signals were acquired and analysed using custom-made software (Trexler et al. 1999) and an A/D converter (National Instruments, Austin, TX, USA).

Fluorescence imaging

Fluorescence signals were acquired using an ORCA digital camera (Hamamatsu Corp., Bridgewater, NJ, USA) with UltraVIEW software for image acquisition and analysis (Perkin Elmer Life Sciences, Boston, MA, USA). Fluorescence imaging was used to select HeLaCx36-EGFP cell pairs exhibiting at least one JP. Typically, cell pairs without JPs showed no coupling, while g_j was higher in cell pairs with larger JPs. Furthermore, fluorescence imaging was used to evaluate fluorescence of a single GJ channel and consequently the total number of channels assembled in JPs. Selected filters for EGFP excitation and emission were housed in Sutter filter wheels (LAMBDA 10-2).

Preparation and application of different compounds used in this study

AA, BSA, n-alkanols (1-pentanol, 1-hexanol, 1-heptanol, 1-octanol, 1-nonanol and 1-decanol), MAFP and CHD were purchased from Sigma-Aldrich (Seelze, Germany). Thapsigargin was obtained from EMD Millipore (Billerica, USA). Stock solutions of both AA and MAFP were prepared in DMSO, while thapsigargin was prepared in 100% ethanol. Stock solutions were stored at −80°C until use. Solubilities of pentanol, hexanol, heptanol and octanol in water (g/100 g H₂O) at 20°C are as follows: 2.7, 0.6, 0.093, 0.054, respectively (Ebbing & Gammon, 2009). Therefore, n-alkanols from pentanol to octanol were diluted directly in MKR solution (final concentrations were below the solubility limit). Nonanol and decanol exhibit very low solubility in H₂O, hence stock solutions (2 m) were prepared in DMSO and further diluted to working concentrations in MKR. The final concentration of DMSO was not higher than 0.1%, which shows no effect on g_j or toxicity on cells (Picoli et al. 2012). Furthermore, all solutions containing n-alkanols were vortexed for at least 10 min. All drugs were applied after establishing the patch recording and g_j had reached a constant value.

It was reported that to eliminate or significantly reduce binding of AA and other PUFAs to BSA, positively charged guanidine groups of arginine residues should be modified using CHD (Toi et al. 1967; Beck et al. 1998). To follow this protocol, we dissolved 825 mg of BSA in 25 ml of 0.2 m NaOH solution. Then, 330 mg of CHD was added and the mixture was incubated for 3 h at room temperature. Afterwards, the solution was titrated to pH 7.0 and dialysed with MKR solution at 4°C for 24 h. The resulting solution was titrated to pH 7.4 and SDS–PAGE was used to confirm that BSA was modified as previously shown (Michel et al. 1985; Katz & La Marche, 1995; Beck et al. 1998); the level of modification was >90%. SDS–PAGE showed a very small quantity of dimers and trimers of BSA and no oligomers of its modified form (BSA–CHD), which is consistent with previous results (Katz & La Marche, 1995). SDS–PAGE data show that BSA–CHD migrates with a broader spread than BSA, indicating that capping of the positively charged guanidine groups of the arginine residues was successful. The final concentration (mg ml⁻¹) of the BSA–CHD stock solution was evaluated using a standard Bradford protein assay (Sigma Aldrich, USA). Molar concentrations of BSA and BSA–CHD solutions used in this study were calculated assuming that the molecular weight of BSA is equal to ∼66 kDa (Hirayama et al. 1990; Beck et al. 1998; Ninomiya & Kayama, 1998).

Data analysis and statistics

Dose–response curves obtained with different concentrations of tested compounds were fitted to a three-parameter logistic sigmoid equation, and a concentration of the compound required to produce 50% of maximal effect (EC₅₀) was derived. To evaluate changes of voltage-gating parameters, the experimental g_j–V_j dependences were fitted using a stochastic four-state model (S4SM; Paulauskas et al. 2009). Data are reported as means ± SEM. Unpaired Student's t test was used for statistical evaluation. P < 0.05 was considered significant.

Results

Functional efficiency of Cx36 GJ channels

HeLa cells stably expressing Cx36-EGFP allowed us to select cell pairs exhibiting JPs in apposition between neighbouring cells. Experiments were performed on the second and third days after plating cells on glass coverslips. The fraction of GJ channels that is open at any given time over the total number of physical channels in the JP (functional and non-functional) was initially measured for Cx43-EGFP and denominated as the functional efficiency (K) of GJ channels (Bukauskas et al. 2000). K was defined as the ratio of the number of open channels at V_j= 0 (N_o) to the total number of channels in the JP (N_T), K = N_o/N_T. These measurements were achieved by combining a dual whole-cell patch-clamp for estimating the number of open channels, where N_o=g_j/γ, and fluorescence imaging of EGFP tagged to Cxs to estimate N_T. It has been shown that only ∼1/10 of Cx43, ∼1/50 of Cx45 and ∼1/100 of Cx57 GJ channels are open at any time at V_j= 0 (Bukauskas et al. 2000; Palacios-Prado et al. 2009, 2010). In this study, we used the same approach. Briefly, we selected cell pairs with JPs that were oriented parallel to the focal plane and could be viewed en face (Fig. 1A). To estimate N_T it was necessary to assess the fluorescence produced by a single GJ channel (F_γ). We selected large JPs with uniform fluorescence in order to measure fluorescence per unit area in a central region of the JP and avoid edge effects (Fig. 1A, inset). We assessed fluorescence per unit area (F_JP) in this region in arbitrary fluorescent units (a.u.) and subtracted background fluorescence outside the JP. We assumed that each Cx36-EGFP channel occupied 100 nm² (corresponding to 10 nm centre-to-centre spacing in a square array and 10⁴ channels per μm²) as an approximation of values seen in atomic force microscopy (Lal et al. 1995; Muller et al. 2002). Thus, we assessed F_γ from the ratio, F_JP/10,000; on average F_γ= 0.140 ± 0.004 a.u. (n= 29). Light intensity and recording conditions were kept constant in all experiments to maintain a.u. consistent.

Figure 1. Functional efficiency of Cx36-EGFP GJ channels.

A, fluorescence image of a cell pair with a JP oriented parallel to the focal plane; the inset shows an enlarged region of JP containing the region of interest (ROI) where fluorescence intensity was measured to estimate the fluorescence of a single GJ channel. B, fluorescence image of a cell pair with a JP oriented perpendicular to the focal plane. The total intensity of JP used to determine the total number of GJ channels was measured in the ROI encircled with a dashed line. C, dependence of the number of open GJ channels on the total number of channels in JPs under control conditions (filled circles) and exposure to 5 mm hexanol (open circles). Linear regression shows that under control conditions (continuous line) the ratio N_o/N_T was 0.0078 ± 0.0005 (n= 21; P < 0.001). Hexanol increased this ratio (dashed line) to 0.03 ± 0.004 (n= 11; P < 0.05).

In combined imaging and electrophysiological studies, we measured g_j and the total fluorescence intensity (F_T) of JPs, independently of their spatial orientation. F_T was estimated by measuring the total fluorescence in the region of interest (ROI) enclosing a JP (dashed ellipse in Fig. 1B). To collect all light including that from out-of-focus regions, the ROI was made several times larger than the size of the JPs (Bukauskas et al. 2000). The total number of GJ channels present in each JP was determined from the ratio, N_T = F_T/F_γ. JPs from ∼2 to 9 μm in diameter contained from ∼5.6 × 10³ to 19 × 10⁴ GJ channels (n= 27), respectively. In estimating N_o, we used the latest report on the single channel conductance of Cx36, which was estimated in pancreatic β-cells (Moreno et al. 2005) expressing solely Cx36; γ= 6 pS. In summary, we found that under control conditions K=N_o/N_T= 0.0078 ± 0.0005 (n= 21), i.e. only ∼1/128 of Cx36-EGFP channels were open at any given time (Fig. 1C, filled circles).

Regulation of g_j and V_j-gating by n-alkanols

Macroscopic gating properties of Cx36 homotypic GJs were reported earlier (Srinivas et al. 1999; Teubner et al. 2000; Moreno et al. 2005), and it was shown that Cx36 is among the least V_j-sensitive Cxs. To measure g_j and V_j-gating of Cx36 GJ channels, we evoked I_j by repeated V_j ramps from 0 to –105 mV (assuming that g_j–V_j is symmetric around V_j= 0 mV) of 35 s in duration and consecutive 20 mV steps (0.5 s) in between them (Fig. 2A, upper trace).

Figure 2. Effect of n-alkanols on g_j and V_j-gating of Cx36-EGFP GJ channels.

A, I_j record from HeLaCx36-EGFP cell pair in response to repeated V_j ramps (35 s) and steps (0.5 s) under application of pentanol (5 mm) and during washout. B, dependence of normalized g_j on the concentration of pentanol (triangles) and hexanol (circles). The values of EC₅₀ for pentanol and hexanol were 4.6 ± 0.5 mm (n= 5) and 2.8 ± 0.4 mm (n= 6), respectively. C, normalized g_j–V_j plots (in black) from one representative experiment were measured under control and after application of 3 or 5 mm hexanol. Experimental data were obtained only at negative values of V_j and during fitting using a S4SM; it was assumed that g_j–V_j plots are symmetric. Fitted curves of g_j–V_j plots are shown in grey. D, dependence of open probability of aHCs (P_o,H) on V_j obtained by fitting experimental g_j –V_j plots shown in C with the S4SM; P_o,H–V_j plots of left and right HCs are shown in grey and black, respectively. E, decanol (0.5 mm) completely uncoupled HeLaCx36-EGFP cell pair.

Lipophilic agents, such as n-alkanols, are broadly used for cell–cell uncoupling (Bruzzone et al. 1987; Burt & Spray, 1988; Weingart & Bukauskas, 1998; Rozental et al. 2001). In addition, it was reported that some n-alkanols had no or little effect on cell–cell coupling depending on the cell type and Cx isoform (Chanson et al. 1989; Rup et al. 1993; Ehrlich et al. 2000); however, there are no reports showing an increase in g_j under an exposure to n-alkanols. Thus, we were surprised when application of pentanol (5 mm) strongly and reversibly increased I_j in Cx36-expressing cells (Fig. 2A, lower trace). To study the effect of n-alkanols on g_j and the V_j-gating in more detail, we used a homologous series of n-alkanols from pentanol to decanol. Hexanol, as well as pentanol, induced a concentration-dependent increase in g_j of Cx36-EGFP GJ channels (Fig. 2B). The value of EC₅₀ for g_j increase under an exposure to pentanol or hexanol was 4.6 ± 0.5 mm (n= 5) or 2.8 ± 0.4 mm (n= 6), respectively. These data are in good agreement with measurements described in Fig. 1 where K was increased ∼four times under an exposure to hexanol (up to 0.030 ± 0.003; n= 11; Fig. 1C). The size of JPs ranged from ∼5.4 × 10³ to 8.6 × 10⁴ GJ channels, and neither the size nor the total fluorescence intensity of JPs changed during application of n-alkanols (not shown), indicating no changes in N_T.

Figure 2C shows typical g_j–V_j plots measured in response to V_j ramps from 0 to –100 mV under control and ∼7 min after application of 3 and 5 mm hexanol, which caused an increase in I_j. Similar g_j–V_j plots were obtained in three other experiments. All experimental g_j–V_j plots (n= 4) were fitted using a S4SM of GJ channels combined with a global optimization algorithm (Paulauskas et al. 2009; see at http://connexons.aecom.yu.edu performance of the S4SM and Movies 1–3). During the fitting process, we assumed that only slow gates function in Cx36. This assumption is based on low sensitivity to V_j-gating and slow kinetics of g_j decay during application of V_j steps that resembles properties of Cx43-EGFP in which tagged EGFP eliminates function of the fast gate (Bukauskas et al. 2001). Furthermore, there are no reliable data demonstrating gating of Cx36 to the residual state, which is the main property of the fast gate. The S4SM allowed us to estimate gating parameters characterizing sensitivity to V_j for each aHC/gate (A_H and V_o,H), open probability of aHC (P_o,H) and the number of functional channels (N_F). A_H characterizes the steepness of changes in P_o,H as a function of voltage across the aHC (V_H), while V_o,H is V_H at which P_o,H= 0.5. Values of A_H and V_o,H estimated during the fitting process allowed calculation of N_F and P_o,H as a function of V_j. Data fitting (Table 1) shows that hexanol (5 mm) reduced the sensitivity of V_j-gating by enhancing V_o,H from ∼42 to 75 mV. Figure 2D shows the dependence of P_o,H on V_j obtained using gating parameters from fitting experimental g_j–V_j plots shown in Fig. 2C. The averaged data from four P_o,H–V_j plots show that under control conditions at V_j= 0, approximately 61% (0.78 × 0.78) of functional GJ channels are open at any given time. An increase in V_o,H resulted in an increase in P_o,H at V_j= 0 mV by ∼1.1-fold at 3 and 5 mm (Table 1), which in part explains the observed g_j increase. In addition, g_j increase was due to an increase in N_F, on average from 377 channels at control to 891 and 1292 channels at 3 and 5 mm hexanol, respectively. This is also reflected in an increase of a steepness of the N_o–N_T relationship under an effect of 5 mm hexanol (Fig. 1C) obtained in a separate series of experiments.

Table 1.

Parameters of voltage gating obtained using a S4SM from four experiments before and after application of 3 and 5 mm of hexanol and 15 μm BSA

Concentration		AH (mV−1)	Vo,H (mV)	γres,H (pS)	γo,H (pS)	NF	Po,H
Control	0	0.029 ± 0.001	42 ± 1.3	0	12	377 ± 66	78 ± 1.0
Hexanol (mm)	3	0.029 ± 0.006	69 ± 4.9**	0	12	891 ± 114**	86 ± 3.8*
	5	0.025 ± 0.004	75 ± 4.3**	0	12	1292 ± 141**	85 ± 3.7*
BSA (μm)	15	0.032 ± 0.004	57 ± 4.5*	0	12	582 ± 187	86 ± 3.1**

N_F is measured at V_j= 0 mV. *P < 0.05; **P < 0.01, compared with control.

Further studies revealed that heptanol (2 mm) increased g_j by 3.0 ± 0.4-fold (n= 7), while octanol (2.5 mm; n= 4), nonanol (0.5 mm; n= 11) and decanol (0.5 mm; n= 5) caused full uncoupling; an example of the uncoupling effect of decanol is shown in Fig. 2E. We observed similar g_j changes depending on the C-chain length of n-alkanols in HeLa cells expressing wild-type Cx36; 2 mm heptanol increased g_j 3.6 ± 0.3-fold, n= 18. Furthermore, hexanol and pentanol increased g_j, while octanol and decanol uncoupled HeLaCx36 cells (not shown). In summary, Cx36 and Cx36-EGFP GJs exhibit g_j increase under application of SCCAs and g_j decrease under LCCAs with the transition between heptanol and octanol.

Regulation of g_j and V_j-gating by BSA

Typically, application of AA leads to full uncoupling of GJs formed from a variety of Cxs and innexins (Spray & Burt, 1990; Schmilinsky-Fluri et al. 1997; Weingart & Bukauskas, 1998; Teubner et al. 2000; Rozental et al. 2001; Contreras et al. 2002), with very slow g_j recovery during washout. It is well established that fatty acid-free BSA accelerates coupling recovery (Fluri et al. 1990; Schmilinsky-Fluri et al. 1997) by directly binding AA and other PUFAs on three potential sites (Reed, 1986; Hamilton et al. 1991; Bojesen & Bojesen, 1994). In addition, BSA decreases plasma membrane fluidity by reducing the concentration of PUFAs (Beck et al. 1998) and by binding divalent metal ions (Sommer-Knudsen & Bacic, 1997; Bal et al. 1998).

We found that in HeLa cell pairs expressing Cx36-EGFP, BSA (15 μm) increased g_j on average ∼1.6 ± 0.1 (n= 8) times. The maximal effect of BSA was reached ∼10 min after its application and g_j returned to the initial level ∼11 min after washout; Fig. 3A shows an example of g_j changes. Figure 3B shows the averaged percentage change in g_j of Cx36 GJs over control depending upon BSA concentration; EC₅₀= 2.3 μm. In subsequent experiments, we used BSA at a concentration of 15 μm, which produced a maximal effect on g_j. To eliminate the possibility that BSA-dependent changes in g_j are due to alterations in free [Ca²⁺]_i (Firek & Weingart, 1995), we substituted 2 mm EGTA in the pipette solution for 3 mm BAPTA ([Ca²⁺]_i= 5 × 10⁻⁸ m). This change in buffer had no effect on the BSA-dependent increase in g_j (1.6 ± 0.3; n= 4), suggesting that the BSA-dependent effect on g_j was not due to changes in [Ca²⁺]_i. We observed no changes in EGFP fluorescence or size of JPs during application and washout of BSA, indicating that observed g_j changes were not due to variations in N_T caused by de novo formation or internalization of channels in JPs (not shown). Furthermore, using the same protocol as for Fig. 2C, we examined changes of V_j-gating under application of 15 μm BSA. The averaged data from four experiments (Table 1) indicate that BSA reduced the sensitivity to V_j by enhancing V_o,H from 42 to 57 mV, which increased P_o,H at V_j= 0 by 1.1-fold. In addition, N_F increased from ∼377 to 582.

Figure 3. The effect of BSA on Cx36-EGFP GJ channels.

A, reversible g_j changes in response to BSA (15 μm) application. B, dose–response curve of the BSA effect on g_j; EC₅₀= 2.3 μm. C, application of AA (2 μm) leads to partial uncoupling, while BSA (15 μm) applied during washout accelerated g_j recovery, and the final g_j was increased to 1.27 ± 0.22-fold of control. The dashed line shows the predicted g_j recovery in the absence of BSA. D, a representative experiment showing that BSA modified with 1,2-cyclohexanedione (BSA–CHD) had no effect under control conditions or after preinhibition of g_j with AA (10 μm). E, comparable data showing that under control conditions BSA increased g_j by 1.62 ± 0.10-fold while BSA–CHD had no effect on g_j. *P < 0.005.

Figure 3C shows an accelerated g_j recovery by BSA from partial uncoupling caused by 2 μm AA; g_j was normalized to its initial value. Spontaneous g_j recovery during washout from AA was close to linear with a steepness of ∼3% per min (n= 8), which is shown by a dashed line. On average, 2 μm AA reduced normalized g_j to 0.31 ± 0.04 (n= 11), while application of BSA (15 μm) increased it to 1.27 ± 0.22 (n= 6). These experiments suggest that BSA extracts PUFAs (including AA) from the membrane; however, due to their constant production, it takes ∼11 min to reestablish the initial g_j that presumably corresponds to the time necessary for reestablishing the initial AA level in the plasma membrane (Fig. 3A). Higher concentrations of AA (≥10 μm) typically caused full uncoupling (n= 20), and BSA promoted g_j recovery to the control level (n= 10).

The binding of AA and other PUFAs to BSA depends upon positively charged arginine residues (Beck et al. 1998). To eliminate or significantly reduce the binding of AA to BSA, positively charged guanidine groups of the arginine residues were neutralized using CHD (for details, see Methods). Previous studies have shown that this modification significantly decreases the potency of BSA to bind PUFAs (Beck et al. 1998). Figure 3D shows that BSA–CHD caused neither g_j increase under control conditions nor g_j recovery after application of AA. Figure 3E shows that on average BSA–CHD did not change normalized g_j (1.03 ± 0.05; n= 8), while BSA increased it to 1.6 ± 0.1 (n= 8). Even application of a threefold higher concentration (45 μm) of BSA–CHD did not evoke an increase in g_j under control conditions or during g_j recovery after washout from AA (n= 3, not shown). These data suggest that endogenous PUFAs could be one of the key factors in promoting a stable closed state of Cx36 GJ channels, consequently resulting in very low functional efficiency (Fig. 1C).

Modulation of endogenous AA concentration

It has been reported that among various phospholipase A2 isoforms (sPLA2: secretory PLA2; cPLA2: cytosolic PLA2; PlsEtn-PLA2: plasmalogen-selective PLA2; iPLA2: calcium-independent PLA2), cPLA2 and PlsEtn-PLA2 have greater specificity for AA production (Gijon & Leslie, 1999; Cummings et al. 2000; Leslie, 2004; Farooqui et al. 2006). In order to reduce endogenous AA production, we used MAFP to irreversibly inhibit cPLA2 and iPLA2. It has been shown that MAFP reduces endogenous AA production in HeLa cells (Huang et al. 2002). If the levels of endogenous AA, and other PUFAs derived from AA, exert an inhibitory effect on Cx36 GJ channels, then the inhibition of cPLA2 by MAFP should enhance g_j by reducing the endogenous concentration of AA. We used MAFP at concentrations in the range of 0.01–0.03 μm, which, as reported in Cummings et al. (2000) and Farooqui et al. (2006), should block cPLA2. Figure 4A shows that MAFP at a concentration of 0.01 μm increased g_j by ∼28% over initial g_j, and this increase was further potentiated by BSA (15 μm). Figure 4C shows that MAFP at 0.01 and 0.03 μm increased g_j on average up to ∼27% over initial g_j.

Figure 4. Pharmacological manipulation of PLA2 activity modulates g_j of Cx36-EGFP GJ channels.

A, a representative experiment demonstrating g_j increase during inhibition of cPLA2 and iPLA2 with methyl arachidonyl fluorophosphonate (MAFP; 0.01 μm), and the combined application of MAFP and BSA (15 μm). B, a representative experiment showing a decrease in g_j during activation of cPLA2 with thapsigargin (0.05 μm), while combined application of thapsigargin and BSA enhanced g_j approximately to initial levels. C, summarized data showing g_j increase after application of MAFP (0.01 and 0.03 μm) and a dose-dependent g_j decrease with thapsigargin (0.05 and 0.1 μm). D, averaged data showing that BSA (15 μm) applied in the presence of thapsigargin (0.05 and 0.1 μm) recovered g_j approximately to the control level. For comparison, the g_j-enhancing effect of BSA (15 μm) under control conditions is shown with a dotted line.

Elevation of endogenous AA in the membrane could be achieved by activation of PLA2 (Balsinde et al. 1999). In order to increase endogenous AA production, we activated cPLA2 using the well-known inhibitor of the Ca²⁺-ATPase, thapsigargin, which causes [Ca²⁺]_i to increase by mediating Ca²⁺ release from the endoplasmic reticulum (Croxtall et al. 1997). Thapsigargin has been used as an effective treatment to activate cPLA2 and increase AA concentration in many cell types, such as mouse RAW 264.7 macrophages, rat liver (C-9) cells, human colon carcinoma cells, estrogen receptor-negative A549 cells, HEL30 keratinocytes, etc. (Croxtall et al. 1997; Levine & Rabouille, 2005). Figure 4B shows that application of 0.05 μm thapsigargin reduced g_j, presumably by increasing endogenous AA production. Figure 4B also shows that BSA restored g_j during thapsigargin application, presumably by reducing endogenous PUFAs and AA concentration. On average, 0.05 and 0.1 μm of thapsigargin decreased normalized g_j to 0.57 ± 0.03 (n= 20) and 0.34 ± 0.03 (n= 19), respectively (Fig. 4C). Figure 4D shows that BSA (15 μm) applied in the presence of thapsigargin increased g_j to a value slightly above control (∼1.1-fold in both cases), i.e. the g_j-enhancing effect caused by BSA alone (1.6-fold, dotted grey line) was reduced. This reduction could be explained by increased levels of AA in the membrane. These results suggest that a BSA-dependent increase in g_j of Cx36 GJs could be related to a removal of endogenous AA from the plasma membrane.

Modulation of AA-dependent inhibition of Cx36 GJs by hexanol

We demonstrated that the effect of n-alkanols on g_j of Cx36 GJs depends on the length of their C-chain (Fig. 2B). Here, we examined whether the g_j-enhancing effect of hexanol is affected by factors that modulate the concentration of AA. Figure 5A shows that application of exogenous AA (10 μm) causes complete uncoupling, and hexanol (5 mm) applied during washout from AA increased the rate of recovery of g_j; the dashed line indicates a predicted trace for spontaneous g_j recovery from washout of AA (∼3% per min). The g_j-enhancing effect of hexanol after pre-inhibition by 2 μm AA was reduced (1.4 ± 0.2; Fig. 5B and E; P < 0.001) as compared with an effect of hexanol alone (3.4 ± 0.3; dashed line in Fig. 5E). Presumably, at 10 μm AA, the blocking effect is too strong for hexanol to compete (Fig. 5E). After pre-inhibition of g_j by 0.1 or 0.05 μm thapsigargin, hexanol increased g_j 1.8 ± 0.2 (n= 8) and 2.4 ± 0.2 (n= 10) times, respectively, i.e. the increase in g_j was smaller as compared with hexanol application alone (Fig. 5C and E). Thus, the g_j-enhancing effect of hexanol was reduced during its combined application with thapsigargin (0.1 and 0.05 μm) or AA (2 μm). Figures 5D and 6C show that application of BSA does not reduce significantly the g_j-enhancing effect of hexanol. Possible interaction of hexanol with BSA during their combined application is excluded because it was shown that n-alkanols (including hexanol) do not interact with BSA (Tan & Siebert, 2008). Similarly, we did not observe a statistically significant difference in the g_j-enhancing effect of hexanol in the presence of MAFP. All these data indicate that an increase in AA levels reduces the g_j-enhancing effect of hexanol, and support the hypothesis that AA and hexanol have opposing effects upon the same gating mechanism.

Figure 5. Modulation of g_j by hexanol during its combined application with arachidonic acid (AA), BSA, thapsigargin or methyl arachidonyl fluorophosphonate (MAFP).

A–D, representative experiments showing the g_j-enhancing effect of hexanol after 10 μm AA (A, the dashed line shows a predicted trace for spontaneous g_j recovery) or during its combined application with 2 μm AA (B, dashed line shows predicted rundown of g_j with AA alone), 0.1 μm thapsigargin (C) and 15 μm BSA (D). E, summary of hexanol g_j-enhancing effect in the presence of different reagents: AA (10 and 2 μm); thapsigargin (0.1 and 0.05 μm); BSA (15 μm); and MAFP (0.03 μm). All data are normalized to initial g_j values before application of indicated compounds. The dashed grey line shows the g_j-enhancing effect of hexanol (5 mm) under control conditions.

Figure 6. Effect of hexanol and BSA on freshly isolated pairs of pancreatic β-cells.

A, image of a freshly isolated pair of β-cells during dual whole-cell voltage-clamp. Cells were cultured for 12 h after isolation. B, a representative experiment showing reversible g_j changes in response to hexanol (5 mm) followed by an increase in g_j with BSA (15 μm). C, a representative experiment showing reversible g_j changes in response to BSA (15 μm) followed by hexanol (5 mm). D, decanol (0.5 mm) completely and reversibly uncouples pairs of pancreatic β-cells.

Effects of BSA and n-alkanols on pancreatic β-cells

It is well established that mouse pancreatic β-cells solely express Cx36 (Serre-Beinier et al. 2000; Moreno et al. 2005). To test whether n-alkanols and BSA cause similar effects in β-cells as in HeLa transfectants, we performed experiments using freshly isolated pairs of β-cells (Fig. 6A; see Methods). Typically, g_j under control perfusion conditions was relatively low and varied from 0.1 and 1.2 nS; on average g_j= 0.32 ± 0.09 nS (n= 16). Figure 6B shows that hexanol (5 mm) caused an increase in g_j, and following BSA (15 μm) application caused an approximately similar increase in g_j but with slower kinetics (fivefold). Figure 6C shows that BSA caused an increase in g_j, and that hexanol can cause a further increase. In summary, we found that BSA and hexanol increased g_j on average 1.8 ± 0.2-fold (n= 8) and 2.2 ± 0.6-fold (n= 7), respectively. Hexanol, when applied straight after BSA, increased g_j on average 1.3 ± 0.1-fold (n= 6; normalized to BSA steady-state value; Fig. 6C). Furthermore, nonanol (0.5 mm) and decanol (0.5 mm) caused full uncoupling; an example of the uncoupling effect of decanol is shown in Fig. 6D. Thus, pancreatic β-cells expressing endogenous Cx36 show very similar modulation of g_j by AA and SCCAs as shown in HeLa transfectants.

Do Cx isoforms differ in sensitivity to BSA and n-alkanols?

To answer this question, we tested the effect of BSA and hexanol on g_j of GJs formed by mCx30.2, Cx36, Cx45 and Cx47 expressed in HeLa cells, and Cx43 endogenously expressed in Novikoff cells (rat hepatoma cells). All these Cxs are expressed in the nervous system: Cxs m30.2, 36 and 45 in neurons; Cx43 in astrocytes; and Cx47 in oligodendrocytes (Eugenin et al. 2012). BSA (15 μm) was applied 10 min after reaching a steady-state after establishing a dual whole-cell recording.

The summarized data presented in Fig. 7 show that BSA (15 μm) promoted coupling only in Cx36- and Cx47-expressing cells, while it had no effect in cells expressing mCx30.2, Cx43 and Cx45. These results indicate that endogenous PUFAs inhibit Cx36 and Cx47 GJ channels at resting conditions. If the mechanism by which PUFAs and SCCAs exert opposite effects on g_j is through competition for a conserved lipophilic binding domain, then hexanol should increase g_j in Cx47 GJs – as shown for Cx36 – by competing with endogenous PUFAs. However, we found that hexanol (5 mm) reduces g_j in Cx47 GJs to 0.07 ± 0.02 (n= 6; Fig. 7), suggesting that there are different binding domains or sensorial elements for PUFAs and SCCAs in Cx47. GJs formed by other Cxs were blocked by hexanol as well (Fig. 7).

Figure 7. Summary of BSA and hexanol effects on cell–cell coupling in HeLa cells expressing different Cx isoforms.

Normalized g_j (to initial g_j values) changes after 10 min of BSA (15 μm) application (grey bars) in Cx30.2, Cx36, Cx43, Cx45 and Cx47 GJs were 1.0 ± 0.11 (n= 5), 1.62 ± 0.10 (n= 8), 0.99 ± 0.10 (n= 5), 0.99 ± 0.02 (n= 6) and 1.48 ± 0.14 (n= 9), respectively; normalized g_j changes after 5–7 min of hexanol (5 mm) application (white bars) on the same Cxs (Cx30.2, Cx36, Cx43, Cx45 and Cx47) were 0.03 ± 0.01 (n= 5), 3.4 ± 0.3 (n= 9), 0.05 ± 0.01 (n= 11), 0.04 ± 0.02 (n= 13) and 0.07 ± 0.02 (n= 6), respectively.

Discussion

Combined fluorescence imaging and electrophysiological studies showed that the ratio between the number of open Cx36 GJ channels at V_j= 0 and the total number of channels assembled in JPs is extremely low, K= 0.008 (Fig. 1). For example, in a cell pair exhibiting g_j= 1 nS, only ∼167 GJ channels with γ= 6 pS are open, while N_T is ∼21,410 GJ channels (167/0.0078). Furthermore, because we estimated an open probability of functional channels equal to ∼0.61 (see Table 1), then ∼274 (167/0.61 ≍ 274) are functional. Of these functional channels, only 167 are open at any given time on average. We assume that most channels in JPs are permanently closed, and presumably channels that were inhibited by AA could be transferred to a functional state by SCCAs, BSA or lowering the activity of different PLA2 isoforms. Such a small fraction of functional channels is not unique to Cx36, and it has been demonstrated that other Cxs also show low functional efficiency (K_Cx43=∼0.1; K_Cx45=∼0.04; K_Cx57=∼0.01; Bukauskas et al. 2000; Palacios-Prado et al. 2009, 2010). Recently, it was shown that in electrical synapses between neurons in the mesencephalic trigeminal nucleus only ∼0.1% of Cx36 GJs are functional (Curti et al. 2012). Also, our unpublished data (F.F.B.) demonstrate the value of K in the range of 0.005–0.05 for Cx46 and Cx50. Accumulating data suggest that GJs formed by all Cxs will have only a small fraction of functional channels under control conditions. Different factors may influence channel transitions between functional and non-functional populations, such as time needed for maturation after docking of two HCs, interaction with other proteins and components of cytoskeleton, phosphorylation, etc. However, we found no changes in the fluorescence of JPs or their size during hexanol, BSA or AA application. The kinetics of changes in g_j during an exposure to hexanol, BSA or AA, and their washout, are too fast to be explained by de novo formation or internalization of GJ channels. This property of GJs may allow having a high number of GJ channels ready to be recruited to function on a short time scale.

Numerous studies have shown that n-alkanols, including pentanol and hexanol, caused full uncoupling in cells expressing Cx26, Cx30.2, Cx32, Cx37, Cx40, Cx43, Cx45, Cx46 and Cx57 GJs (Burt et al. 1991; Moreno et al. 1991; Veenstra et al. 1994; Harris, 2001; Spray et al. 2002; Skeberdis et al. 2011). Other studies have shown that SCCAs (including hexanol) have no effect on electrical coupling between pancreatic acinar cells, cultured human fibroblasts or only slightly inhibit cell coupling in cumulus cells (Chanson et al. 1989; Fagbohun & Downs, 1991), which do not express Cx36. However, we found that SCCAs produce a marked g_j increase in HeLaCx36 cells (Fig. 2) and pancreatic β-cells (Fig. 6). Fitting of g_j–V_j plots with a S4SM (Fig. 2C and D; Table 1) revealed that hexanol increased V_o,H, which in turn enhanced open probability of gates and could explain, at least in part, an increase in g_j. Therefore, to explain a ∼fourfold increase of functional efficiency of Cx36 GJs (Fig. 2B), it should be assumed that in parallel with an increase in the open probability there was also an increase in N_F. According to Ebihara & Steiner (1993) and Oh et al. (1999), the gates could be in two or more closed states. Our unpublished data support these notions, and we assume that gates could be in the initial closed (c1) state from which gates can operate between closed and open states, and that there is one or more deep closed (c2) states from which gates can transit only to the initial closed state but not directly to the open state. We speculate that, for example, hexanol facilitates c2→c1 transitions.

Besides AA, many other PUFAs and monounsaturated fatty acids are known to block GJ channels when externally applied. For example, oleic (18:1), linoleic (18:2), linolenic (18:3), palmitoleic (16:1) and myristoleic (14:1) acids and their derivatives oleamide and anandamide have all been shown to reduce cell–cell coupling (Giaume et al. 1989; Malewicz et al. 1990; Burt et al. 1991; Hirschi et al. 1993; Hii et al. 1995; Lavado et al. 1997; Boger et al. 1998; Harris, 2001). Here, we show that BSA at a concentration of 15 μm increased g_j∼1.6-fold in Cx36- and ∼1.5-fold in Cx47-expressing cells, but had no effect on Cx30.2-, Cx43 and Cx45-expressing cells (Figs 3 and 7). Presumably, this difference could be influenced by AA-binding affinity, i.e. the stronger the binding of AA to Cx, the lower the ability of BSA to affect coupling. There could be many other reasons for Cx-type dependent differences in g_j changes under an effect of BSA and n-alkanols. We assume that the g_j-enhancing effect of BSA is mainly achieved by extracting PUFAs (including AA) from the plasma membrane in HeLa cells (Figs 3 and 7) and pancreatic β-cells (Fig. 6). This assumption is strongly supported by an absence of g_j-enhancing effect of BSA–CHD (Fig. 3D and E). Changes of gating properties under application of BSA were similar to those obtained under application of hexanol (Table 1), which in both cases resulted in a substantial increase in g_j. The simplest explanation for the observed changes in A_H and V_o, which revealed an increase of both P_o,H and N_F, is that the dominant effect of hexanol and BSA on the channels is to stabilize an open state of the gate relative to the closed state, as reflected in the large shifts in V_o,H and relatively small changes in A_H.

The observed changes in g_j during reduction or elevation of endogenous PUFA levels with MAFP or thapsigargin (Fig. 4), respectively, support the notion that Cx36 GJs are normally inhibited by endogenous PUFAs and the g_j-enhancing effect of BSA depends on reduction of endogenous PUFA levels in the plasma membrane (Fig. 4). It has been shown that the effect of thapsigargin on [Ca²⁺]_i is transient, and normal [Ca²⁺]_i is reestablished after ∼10 min of thapsigargin exposure in HeLa cells (Peppiatt et al. 2004). However, the observed effect of thapsigargin on g_j was stable and persistent. Thus, it is unlikely that direct Ca²⁺-dependent uncoupling caused g_j reduction.

The content of AA varies depending on cell type: ∼1 μm in human leukocytes (Chilton et al. 1996), ∼5 μm in human platelets (Neufeld & Majerus, 1983), ∼13 μm in human normal skin tissue (Hammarstrom et al. 1975) and ∼15 μm in isolated islets of Langerhans from rat (Ramanadham et al. 1992). We presume that normal levels of AA in β-cells (15 μm) keep Cx36 GJs inhibited (Serre-Beinier et al. 2009), which correlates with low intrinsic coupling between isolated pairs of β-cells found in this study; g_j varied between 0.1 and 1.2 nS. Other groups also reported low cell–cell coupling between pancreatic β-cells from rodents; g_j varied in the range of ∼0.17–0.95 nS (Moreno et al. 2005; Zhang et al. 2008; Benninger et al. 2011). Cx36 knockout mice exhibit coupling deficiency and asynchronous Ca²⁺ transients between pancreatic β-cells, leading to irregular dynamics and insufficient insulin secretion (Ravier et al. 2005; Head et al. 2012). Furthermore, specific deletion of Cx36 in β-cells leads to major alterations in glucose-induced insulin secretion (Wellershaus et al. 2008). Thus, Cx36-mediated coupling between β-cells plays an important role in insulin secretion and the development of diabetes. Our data show that in β-cells, g_j changes in response to BSA and n-alkanols (Fig. 6) are very similar to g_j changes induced in HeLaCx36 cells. There are some differences in the magnitude and kinetics of g_j changes between HeLa and β-cells that can be attributed to differences in levels of expression of Cx36, composition of lipids of the plasma membrane, etc. Nonetheless, collected data support the notion that in β-cells under control conditions most of the functional Cx36 GJ channels are affected by endogenous levels of PUFA derivatives (including AA). PUFA-dependent inhibition of coupling between β-cells and its modulation by relatively small lipophilic molecules, such as SCCAs, may play an important role in insulin secretion and highlights new avenues for regulation of β-cell function. The g_j-enhancing effects of SCCAs and BSA are not a unique property found in HeLa cells expressing Cx36, but are also found in pancreatic β-cells and possibly neurons expressing endogenous Cx36.

In summary, our data show that the g_j-enhancing effect of SCCAs is related to their ability to interfere with AA-dependent inhibition of Cx36. Furthermore, elevation of AA concentration by thapsigargin or exogenous application of AA (2 μm) reduced the potentiating effect of hexanol on g_j. Conversely, MAFP and BSA, which tend to reduce the levels of PUFAs (including AA) in the plasma membrane, did not reduce the g_j-potentiating effect of hexanol (Fig. 5D and E) or did they enhance it, which may suggest that at relatively low levels of AA the ability of hexanol to lessen an effect of AA is enhanced. Collected data allow us to assume that Cx36 contains a unique structural domain, which results in g_j increase under an exposure to SCCAs, while all other Cxs isoforms examined in this study and reported by others show uncoupling (Burt et al. 1991; Moreno et al. 1991; Veenstra et al. 1994; Harris, 2001; Spray et al. 2002; Skeberdis et al. 2011).

Glossary

γ

single GJ channel conductance

AA

arachidonic acid

aHC

apposed hemichannel

BSA–CHD

BSA modified with 1,2-cyclohexanedione

C-chain

carbon chain

cPLA2

cytosolic phospholipase A2

Cx

connexin

FBS

fetal bovine serum

GJ

gap junction

g_j

junctional conductance

HBSS

Hank's balanced salt solution

HC

hemichannel

I_j

junctional current

JP

junctional plaque

LCCA

long carbon chain n-alkanols

MAFP

methyl arachidonyl fluorophosphonate

MKR

modified Krebs–Ringer solution

N_F

number of functional channels

N_o

number of open channels

N_T

total number of channels

PUFA

polyunsaturated fatty acid

ROI

region of interest

S4SM

stochastic four-state model

SCCA

short carbon chain n-alkanols

V_j

transjunctional voltage

Author contributions

A.M., N.P-P. and F.F.B.: conception, design of experiments, collection, analysis and interpretation of data, drafting of manuscript; L.R. and V.A.S.: collection and analysis of data, critically revising the manuscript. All authors approved the final version for publication.