pH-dependent modulation of connexin-based gap junctional uncouplers

Non-technical summary

Gap junction channels formed from connexin (Cx) proteins are expressed in a variety of tissues providing exchange of metabolites and electrical communication between cells. We demonstrate that the uncoupling effects of gap junction channel blockers such as long carbon chain alkanols (LCCAs), volatile anaesthetics and the antimalarial drug mefloquine can be modulated by intracellular pH in a Cx-type-dependent manner. In addition, we provide data to support the idea that the uncoupling effect of LCCAs and other uncouplers could be related to a modification of hydrogen bonding between histidine residues and uncouplers, and/or inside the Cx structure.

Abstract

Abstract

Gap junction (GJ) channels formed from connexin (Cx) proteins provide a direct pathway for electrical and metabolic cell–cell communication exhibiting high sensitivity to intracellular pH (pH_i). We examined pH_i-dependent modulation of junctional conductance (g_j) of GJs formed of Cx26, mCx30.2, Cx36, Cx40, Cx43, Cx45, Cx46, Cx47 and Cx50 by reagents representing several distinct groups of uncouplers, such as long carbon chain alkanols (LCCAs), arachidonic acid, carbenoxolone, isoflurane, flufenamic acid and mefloquine. We demonstrate that alkalization by NH₄Cl to pH ∼8 increased g_j in cells expressing mCx30.2 and Cx45, yet did not affect g_j of Cx26, Cx40, Cx46, Cx47 and Cx50 and decreased it in Cx43 and Cx36 GJs. Unexpectedly, cells expressing Cx45, but not other Cxs, exhibited full coupling recovery after alkalization with NH₄Cl under the continuous presence of LCCAs, isoflurane and mefloquine. There was no coupling recovery by alkalization in the presence of arachidonic acid, carbenoxolone and flufenamic acid. In cells expressing Cx45, IC₅₀ for octanol was 0.1, 0.25 and 2.68 mm at pH_i values of 6.9, 7.2 and 8.1, respectively. Histidine modification of Cx45 protein by N-bromosuccinimide reduced the coupling-promoting effect of NH₄Cl as well as the uncoupling effect of octanol. This suggests that LCCAs and some other uncouplers may act through the formation of hydrogen bonds with the as-of-yet unidentified histidine/s of the Cx45 GJ channel protein.

Introduction

Gap junction (GJ) channels are composed of two apposed hemichannels in contiguous cells and provide a direct pathway for electrical and metabolic cell-to-cell communication (Paul, 1986; Bennett, 1994; Elfgang et al. 1995; Goodenough et al. 1996; Rackauskas et al. 2010). Six connexin (Cx) subunits oligomerize into a connexon, which after insertion into the surface plasma membrane is called a hemichannel. The family of connexin genes consists of 20 genes in the mouse and 21 genes in the human genome. Connexins are expressed in all tissues except differentiated skeletal muscle, erythrocytes and mature sperm cells. Various tissues express more than one type of connexin, and therefore homotypic, heterotypic and heteromeric GJ channels may form between cells. Gating and permeability properties of GJ channels are regulated largely by transjunctional voltage (V_j), intracellular calcium, pH and phosphorylation (Goodenough et al. 1996; Lampe & Lau, 2004; Rackauskas et al. 2010). For studies of functional properties of GJs and for the development of new therapeutic approaches involving regulation of gap junctional coupling, high affinity uncouplers, especially those which affect channels in a connexin-type-specific manner, are required. Current nomenclature of channel uncoupling agents includes glycyrrhetinic acid and its derivatives, polyamines, antimalarial drugs, fenamates, 2-aminophenoxyborate, volatile anaesthetics, long carbon chain alkanols (LCCAs), fatty acid amides, cyclodextrins, arachidonic acid, and peptides targeting extracellular loops of connexins (reviewed in Rozental et al. 2001; Srinivas, 2009). Even though some of these agents inhibit channels in a Cx-type-specific manner (Spray et al. 2002), the mechanisms of their action remain elusive. Moreover, the potency and efficacy of uncouplers may depend on the composition of the extracellular or intracellular environment. For instance, arylaminobenzoates (Srinivas & Spray, 2003), local anaesthetics and antimalarial drugs (Srinivas et al. 2001) at physiological pH exist in both charged and uncharged forms. Uncharged drugs are more lipid-soluble which allows them to cross the membrane and after protonation in the aqueous environment of the cytoplasm to interact with the receptor (Hille, 2001). Relatively little is known about the impact of intracellular pH (pH_i) on the blocking capacity of GJ uncoupling agents and how this effect depends on the Cx isoform. This is important in understanding the mechanisms of modulation of junctional communication by uncouplers under ischaemic or other pathological conditions leading to pH_i changes.

In the present study, we examined the blocking capacity of octanol and other GJ inhibitors as a function of pH_i in cells expressing Cx26, mCx30.2 (mouse orthologue of human Cx31.9), Cx36, Cx40, Cx43, Cx45, Cx46, Cx47 and Cx50. We demonstrate that: (1) the uncoupling potency of long carbon chain alkanols and other uncouplers on Cx45 GJ channels is pH_i dependent; (2) pH_i-dependent modulation of uncoupling by long carbon chain alkanols is Cx-type specific; (3) octanol-induced uncoupling of Cx45 GJ channels may be mediated by formation of hydrogen bonds with histidines of a Cx protein.

Methods

Cells and culture conditions

Experiments were performed using HeLa cells (human cervix carcinoma cells, ATCC CCL2) transfected with wild-type mouse Cx26, Cx40, Cx45, Cx47, Cx50 and mCx30.2, Cx36, Cx46 fused with enhanced green fluorescent protein (EGFP) (mCx30.2-EGFP, Cx36-EGFP and Cx46-EGFP, respectively). In addition, we examined Novikoff cells endogenously expressing Cx43 (Meyer et al. 1992). Cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. The cells were passaged weekly, diluted 1:10 and maintained in a CO₂ incubator in a moist atmosphere at 37°C. All media and culture reagents were obtained from Life Technologies (GIBCO-BRL).

Electrophysiological measurements

For simultaneous electrophysiological and fluorescence recording, cells grown on glass coverslips were transferred to an experimental chamber with a constant flow-through perfusion mounted on the stage of an inverted microscope Olympus IX70 (Olympus America, Melville, NY, USA). Junctional conductance, g_j, was measured in selected cell pairs by using a dual whole-cell patch clamp. Cell-1 and cell-2 of a cell pair were voltage clamped independently with separate patch clamp amplifiers EPC8 (HEKA Elektronik, Lambrecht, Germany) at the same holding potential, V₁ = V₂. Voltages and currents were acquired and analysed using a MIO-163 A/D converter (National Instruments, Austin, TX, USA) and custom-made software (Trexler et al. 1999). By stepping the voltage in cell-1 (ΔV₁) and keeping the other constant, junctional current 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/ΔV₁, where ΔV₁ is equal to transjunctional voltage (V_j) and the negative sign indicates that I_j measured in cell-2 is oppositely oriented to the one measured in cell-1. To minimize the effect of series resistance on measurements of g_j (Wilders & Jongsma, 1992), we maintained pipette resistances below 3 MΩ. Patch pipettes were pulled from glass capillary tubes with filaments. Experiments were performed at room temperature in modified Krebs–Ringer solution (in mm): NaCl, 140; KCl, 4; CaCl₂, 2; MgCl₂, 1; glucose, 5; pyruvate, 2; Hepes, 5 (pH 7.4). Patch pipettes were filled with saline containing (in mm): KCl, 130; sodium aspartate, 10; MgATP, 3; MgCl₂, 1; CaCl₂, 0.2; EGTA, 2; Hepes, 5 (pH 7.3). Ammonium chloride (NH₄Cl) and sodium acetate (NaCH₃CO₂) were used to alkalize or acidify, respectively, the intracellular milieu with no change in pH_o. Hexanol, octanol, nanonol, carbenoxolone, forane, mefloquine, flufenamic acid and arachidonic acid were used to inhibit GJ channels. All chemical reagents were purchased from Sigma-Aldrich Corp.

Fluorescence imaging studies

Fluorescence signals were acquired using a Hamamatsu cooled digital camera and UltraVIEW software (PerkinElmer Life Sciences, Boston, MA, USA). For pH_i measurement, cells were loaded with 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein (BCECF; 4 μm) introduced into the cells through the patch pipettes in whole-cell voltage clamp mode. Dye was alternately excited with 436 nm and 500 nm light and the emitted light was filtered at 540 nm and recorded. Ratios of emitted light collected at excitation light of 436 and 500 nm (background subtracted) were converted to pH_i values based on a calibration curve. The latter was obtained using ionophore nigericin at 20 μm concentration in the presence of 140 mm potassium to equilibrate the intracellular pH with extracellular medium of different pH (Thomas et al. 1979). To prevent dye bleaching, imaging was performed in time-lapse mode by exposing every 15 s to a low-intensity excitation light for 500 ms, as described in more detail earlier (Rackauskas et al. 2007; Palacios-Prado et al. 2009, 2010).

Statistical analysis

Cumulative dose–response curves of octanol were obtained by testing 4 or 5 successively increasing concentrations of the compound in control, alkaline or acidic conditions. For each individual experiment, the results of g_j changes over concentration of uncouplers were fitted to the 3-parameter logistic sigmoid equation, and IC₅₀, a concentration of the compound required to produce 50% of inhibition, was derived. To evaluate the g_j–V_j dependence of Cx45 GJ channels, the experimental g_j–V_j curves obtained during slow voltage ramps were fitted using a stochastic four-state model (S4SM) (Paulauskas et al. 2009). Data are reported as means ± SEM. Student's t test was used for statistical evaluation. P < 0.05 was considered significant.

Results

Uncoupling of Cx45 GJs by long carbon chain alkanols depends on pH_i

Experiments were performed on HeLa cell pairs (Fig. 1A) transfected with Cx45, which is broadly expressed in many different organs including the conductive system of the heart, smooth muscles of blood vessels and neurons (Kreuzberg et al. 2006; Rackauskas et al. 2010). Lately, it was reported that g_j of Cx45 GJs is regulated by pH_i through modulation of the sensitivity to V_j of the fast gate and open probability of the slow gate (Palacios-Prado et al. 2010). To study g_j dependence on pH_i, we used NH₄Cl (Swietach & Vaughan-Jones, 2005) to increase pH_i measured by loading cells with BCECF (4 μm) through the patch pipettes. pH_i was measured in the regions of interest, ROI-1 and ROI-2, positioned on cell-1 and cell-2 (Fig. 1A). To measure g_j and V_j-gating, we used a V_j protocol composed of repeated 30 s-long V_j ramps changing from 0 to −100 mV and five short steps of +10 mV (upper trace in Fig. 1B). g_j was evaluated during the 5th V_j step followed by V_j ramp allowing maximal recovery of g_j after V_j-gating caused by long V_j ramps. Homotypic GJs typically demonstrate symmetric g_j–V_j dependence; therefore, we measured I_j (lower trace in Fig. 1B) only in response to one polarity of V_j. As it is demonstrated in Fig. 1C, application of octanol (1 mm) blocked g_j fully (n = 8) with roughly no effect on pH_i (ΔpH_i≈–0.1). Then, addition of NH₄Cl (15 mm) to the external solution in the continuous presence of octanol increased pH_i to 8.2 ± 0.1 (n = 4) and unexpectedly increased g_j to a level that exceeded the control level ∼2-fold (195 ± 2.4%; n = 4). Furthermore, g_j remained unchanged after washout of octanol and reached the control level during washout. Thus, alkalization in the presence of octanol not only fully abolished the uncoupling effect of octanol, but g_j significantly exceeded its control value resulting in the g_j observed with application of NH₄Cl alone (Palacios-Prado et al. 2010). Octanol applied to cells already exposed to NH₄Cl did not cause any detectable effect on g_j until washout of NH₄Cl (Fig. 1D).

Figure 1. Intracellular alkalization attenuates the blocking effect of octanol on Cx45 GJ channels.

A, fluorescence image of Cx45-expressing HeLa cell pair loaded with pH-sensitive fluorescent dye BCECF (4 μm). B, V_j-gating and g_j was measured in response to repeated V_j ramps changing from 0 to −100 mV. V_j steps of +10 mV were used to measure g_j between V_j ramps. Intracellular pH in cell-1 and cell-2 (open and filled circles, respectively, in C and D) were measured from the regions of interest (ROI; dashed circles) concomitantly with g_j (open squares in C and D) under control conditions and during application of octanol (1 mm) and/or NH₄Cl (15 mm). C, octanol causes full uncoupling without essential changes in pH_i, while NH₄Cl quickly increases g_j exceeding its control value almost twice. D, g_j increase caused by application of NH₄Cl is not affected by octanol.

Further, we measured the dependence of g_j of Cx45 GJ channels on octanol concentration under control, acidic and alkaline conditions. To measure g_j and V_j-gating, we used a similar protocol to that described above (Fig. 1B). As shown in Fig. 2A, octanol produced dose-dependent inhibition of g_j. A cumulative dose–response curve for the effect of octanol on g_j is presented in Fig. 2B. The data are normalized to the control g_j and IC₅₀ was derived from the fit of the experimental points to the logistic sigmoid equation. Thus, under control conditions (pH_i = 7.2), octanol (0.01–1 mm) inhibited g_j with an IC₅₀ value of 0.25 mm (n = 5; Fig. 2B, circles). Elevating pH_i (8.2 ± 0.1) by exposing cells to NH₄Cl (15 mm; n = 5; triangles), resulted in an increase of IC₅₀ to 2.68 mm. At reduced pH_i (6.9 ± 0.11), when cells were exposed to NaCH₃CO₂ (10 mm; n = 5; squares), IC₅₀ decreased to 0.1 mm.

Figure 2. The inhibition of Cx45 GJ channels by octanol is pH_i dependent.

A, g_j measured in response to V_j steps applied in between voltage ramps. During indicated periods, the cells were exposed to four concentrations of octanol. B, concentration–response curves of the inhibitory effect of octanol on g_j in control, alkaline and acidic conditions. The points show the mean and SEM. Continuous lines were derived from a non-linear least-mean-squares regression of the means to the logistic sigmoid equation. IC₅₀ of octanol in control (pH_i = 7.2; circles), alkaline (pH_i = 8; triangles) and acidic (pH_i = 6.9; squares) conditions was 0.25 mm (n = 5), 2.68 mm (n = 5) and 0.1 mm (n = 5), respectively. C, g_j–V_j dependencies recorded at the times indicated by corresponding numbers in A in response to voltage ramps from 0 to −100 mV. Experimental g_j–V_j plots were fitted to S4SM assuming that V_j-gating was symmetric around V_j = 0.

Figure 2C shows g_j–V_j plots measured at times indicated by numbers (1–5) on g_j trace in Fig. 2A in response to voltage ramps of negative polarity from 0 to −100 mV (scatter plots). To estimate changes of gating parameters of GJ channels, g_j–V_j plots were fitted to a stochastic 4-state model (S4SM) (Paulauskas et al. 2009), which in the process of simulation follows the principle of ‘contingent gating’ (Harris et al. 1981) and accounts for the voltage across each apposed hemichannel (aHC) of the GJ channel. The latter depends on the state of the aHC in series and unitary conductances of open (γ_open,H) and residual (γ_res,H) states of the aHC, which in the model correspond to the gate. During simulation, we assumed that g_j–V_j plots of Cx45 homotypic GJs are symmetric around V_j = 0 resulting in symmetric fitting curves (continuous lines in Fig. 2C). It was reported that a single Cx45 open unapposed hemichannel (uHC) conductance rectifies depending on transmembrane voltage, and in our evaluations γ_open,H = 62 pS (Valiunas, 2002) that is ∼2-fold higher than the conductance of a single Cx45 GJ channel (γ_open,GJ = 32 pS; Bukauskas et al. 2002). Our reported (Bukauskas et al. 2002; Palacios-Prado et al. 2010) and unpublished studies (F. F. Bukauskas) of HeLaCx45 cells revealed that γ_open,GJ remains the same when measured in cells expressing only a few channels or during uncoupling by transient applications of alkanols and CO₂. Therefore, we assumed that γ_open,H was not modified by octanol.

Experimental evaluation of a unitary conductance of the residual state of uHC is more complicated; therefore, in our evaluations, γ_res,H was left as an independent parameter and its value was estimated during the fitting process. Another reason for doing so was because of the limitations of using S4SM, which accounts for only one gate per hemichannel, although indeed there are fast and slow gates per hemichannel (Bukauskas & Verselis, 2004). Formally, for the slow gate the γ_res,H = 0, while for the fast gate γ_res,H > 0, therefore S4SM accounts for some averaged value of γ_res,H which depends on the ratio of gated fast and slow gates. The S4SM allowed us to define parameters characterizing V_j-gating, namely V_o,H (voltage across an aHC at which its open probability, P_o,H = 0.5) and A_H (coefficient characterizing the steepness of P_o,H changes as a function of voltage across the aHC). In addition, the S4SM allows us to define the number of Cx45 GJ channels (N_F) that are functional at any given time thereby providing the actual g_j at V_j = 0. Thus, during the fitting process γ_res,H, V_o,H, A_H and N_F were left as free parameters. A summary of data from five experiments similar to that shown in Fig. 2C are shown in Table 1.

Table 1.

Gating parameters of Cx45 GJs derived during fitting of experimental g_j–V_j plots, similar to those shown in Fig. 2C, using S4SM (n = 5)

No. of gj–Vj plot in Fig. 2C	γopen,H (pS)	γres,H (pS) mean ± SEM	Vo,H (mV) mean ± SEM	AH (mV−1) mean ± SEM	NF mean ± SEM
1	62	8.2 ± 0.9	37 ± 1	0.14 ± 0.01	838 ± 162
2	62	8.0 ± 0.6	36 ± 2	0.14 ± 0.02	780 ± 155
3	62	8.2 ± 0.7	33 ± 2	0.14 ± 0.01	596 ± 117
4	62	7.9 ± 0.7	25 ± 2	0.14 ± 0.01	315 ± 61
5	62	3.9 ± 0.6	14 ± 2	0.13 ± 0.01	139 ± 41

γ_open,H is conductance of fully open aHC; γ_res,H is residual conductance of the closed aHC; V_o,H is a voltage across an aHC at which its open probability, P_o,H, is equal to 0.5; A_H is a coefficient characterizing the steepness of P_o,H changes as a function of voltage across the aHC; N_F is the number of Cx45 GJ channels that are functional at any given time.

Figure 3 demonstrates that the uncoupling effect of other alkanols, such as hexanol and nonanol with shorter and longer carbon chains, respectively, also depends on pH_i. Hexanol and nonanol alone did not change pH_i (not shown). Figure 3A and B shows that uncoupling caused by hexanol (5 mm) and nonanol (0.2 mm) was fully reversed, even enhanced, by exposing cells to NH₄Cl in the continuous presence of alkanols, suggesting that pH_i modulates the blocking potency of LCCAs independently of the length of their carbon chain. Interestingly, recovery of coupling by wash-in of NH₄Cl in the presence of hexanol, octanol or nonanol completes in 10–15 s (see Fig. 1C, the inset in Fig. 3A and the g_j changes in Fig. 3B indicated by dashed arrow), which is comparable with the time constant of g_j increase during application of NH₄Cl alone (Fig. 1D). Time constants of g_j changes during application or washout of LCCAs are several times longer (∼100 s; Figs 1C and 3A and B). This may be explained by a higher mobility of H⁺ ions than LCCAs.

Figure 3. pH_i modulates the blocking potency of LCCAs independent of the length of their carbon chain in Cx45-expressing HeLa cells.

A, hexanol (5 mm) reduced g_j to 23.2 ± 5.6% (n = 4) while combined application of hexanol and NH₄Cl (15 mm) increased g_j to 158 ± 6.9% (n = 4) of the control value. B, nonanol (0.2 mm) reduced g_j to 1.8 ± 1.6% (n = 4), while combined application of nonanol and NH₄Cl (15 mm) increased g_j to 137 ± 4.7% (n = 4) of the control value. g_j increase during application of NH₄Cl was much faster (see the inset in A and dashed arrow in B) than g_j changes during application or washout of alkanols alone.

pH_i-dependent modulation of g_j by octanol varies among Cx isoforms

Numerous reported data show that GJ channels and hemichannels composed of different Cx isoforms can be fully blocked by LCCAs and at lower concentrations by alkanols with longer carbon chains (Srinivas, 2009). To examine whether the uncoupling potency of octanol can be also modulated by pH_i in GJs formed of other Cxs rather than Cx45, we examined HeLa cells stably transfected with Cx26, mCx30.2, Cx36, Cx40, Cx46, Cx47 and Cx50 and Novikoff cells expressing Cx43 endogenously by using the same protocol as shown in Figs 1 and 3. These Cxs include all ‘cardiac’ Cxs (30.2, 40, 43 and 45), all ‘neuronal’ Cxs (30.2, 36 and 45), major Cxs expressed in the lens (43, 46 and 50) and other tissues (reviewed in Harris, 2001; Rackauskas et al. 2010). Figure 4A–H shows that octanol induced uncoupling in all cell pairs expressing the Cxs mentioned above and that g_j did not recover during combined application of octanol and NH₄Cl. Furthermore, NH₄Cl promoted full uncoupling of Cx36 GJs, which is associated with the uncoupling effect of NH₄Cl on Cx36 GJs under control conditions (see Fig. 4I). Different kinetics of g_j decay among Cx isoforms may reflect differences in sensitivity to alkanols, e.g. only partial uncoupling of Cx36 GJs (Fig. 4C) by 1.5 mm octanol; full uncoupling was reached only at ∼2.5 mm octanol (not shown). There were also differences in the kinetics of g_j recovery during washout. In this respect, Cx46 was most distinct by not showing g_j recovery at least during 7 min of washout in all six experiments. Thus, modulation of octanol-induced uncoupling by NH₄Cl is Cx-type specific.

Figure 4. Modulation of g_j and octanol-induced uncoupling by NH₄Cl is connexin-type specific.

In HeLa cells expressing Cx26 (A), mCx30.2 (B), Cx36 (C), Cx40 (D) Cx46 (F), Cx47 (G) and Cx50 (H) as well as in Novikoff cells endogenously expressing Cx43 (E), NH₄Cl did not remove the inhibitory effect of octanol on g_j (n≥ 4 in each set of experiments). I, g_j sensitivity to intracellular alkalinization is Cx-type specific. After application of NH₄Cl (10 mm) alone, g_j values measured in cells expressing Cx26, mCx30.2, Cx36, Cx40, Cx43, Cx45, Cx46, Cx47 and Cx50 were 1.06 ± 0.08 (n = 5), 1.35 ± 0.04 (n = 5), 0.41 ± 0.02 (n = 7), 0.92 ± 0.03 (n = 4), 0.15 ± 0.09 (n = 4), 2.12 ± 0.16 (n = 5), 0.99 ± 0.05 (n = 6), 1.0 ± 0.04 (n = 4) and 1.15 ± 0.02 (n = 5), respectively, relative to their control values.

Figure 4I shows that intracellular alkalization with NH₄Cl (10 mm; pH_i = 8.1 ± 0.03; n = 6) potentiated g_j of mCx30.2 and Cx45, had no significant effect on Cx26, Cx40, Cx46, Cx47 and Cx50, and inhibited g_j of Cx36 and Cx43 GJs. An inhibitory effect of alkalization on Cx36 GJs is in concert with the report of González-Nieto et al. (2008). Thus, Cxs differ highly in their reaction to alkalization with a distinctly high increase in g_j of Cx45 GJs.

pH_i-dependent modulation of Cx45 GJs by different uncouplers

We examined whether pH_i modulates Cx45 mediated cell–cell coupling by other well-established GJ uncouplers such as: (1) forane (isoflurane), a general anaesthetic that exhibits uncoupling effects at concentrations used for inhalation anaesthesia (Burt & Spray, 1989); (2) flufenamic acid, an anti-inflammatory drug (Harks et al. 2001; Srinivas & Spray, 2003); (3) mefloquine, antimalarial drug, which inhibits GJs in connexin-type specific manner (Srinivas et al. 2001; Cruikshank et al. 2004); (4) carbenoxolone, a derivative of glycyrrhetinic acid used to treat oesophageal ulceration and inflammation (Davidson & Baumgarten, 1988); and (5) arachidonic acid, a precursor of prostaglandins and leukotrienes, one of the essential fatty acids, freed from a phospholipid molecule by the enzyme phospholipase A2 (Rozental et al. 2001). We have tested all these compounds for their effect on g_j when applied alone or in combination with NH₄Cl by performing at least four experiments with each of them.

During application of forane (3 mm) for ∼500 s, g_j gradually approached zero (Fig. 5A), while pH_i remained unchanged. NH₄Cl (15 mm) applied in a presence of forane resulted in full g_j recovery. Thus, the uncoupling effect of forane is pH_i dependent, similar to the effect of LCCAs (Figs 1 and 3).

Figure 5. pH_i-dependent modulation of Cx45 GJs by different uncouplers.

pH_i in the cell pair was measured concomitantly with g_j under control conditions and during application of GJ blockers alone and together with NH₄Cl. A, g_j in response to forane (3 mm) decreased, approaching zero. Combined application of forane and NH₄Cl (15 mm) increased g_j, exceeding its control value by 87 ± 9% (n = 5). Forane alone had no effect on pH_i. B, mefloquine slightly and transiently increased pH_i which was followed by a transient increase in g_j; however, it later caused the complete block of g_j which fully recovered under application of NH₄Cl. g_j in response to mefloquine (10 μm) and mefloquine together with NH₄Cl (15 mm) was 12.1 ± 2.1% (n = 4) and 150 ± 13.9% (n = 4) of control conditions, respectively. C, flufenamic acid alone slightly reduced pH_i (∼0.2 units) and caused a virtually complete block of g_j that was not recovered by NH₄Cl; the same effect was observed in all four experiments. D and E, uncoupling effects of carbenoxolone (D) and arachidonic acid (E) were not affected by NH₄Cl.

It was reported that mefloquine inhibits function of Cx26, Cx32, Cx43 and Cx46 GJs with a 10–20-fold higher potency than it inhibits the function of Cx35 (fish orthologue of Cx36) and Cx50 GJs (Srinivas et al. 2001). Mefloquine (10 μm) caused full uncoupling of Cx45 GJs (Fig. 5B). Since mefloquine is a weak base, a small initial increase in g_j was probably caused by a small transient increase in pH_i. NH₄Cl (15 mm) applied in the presence of mefloquine induced a rapid rise in pH_i and the complete removal of the mefloquine uncoupling effect. Washout of NH₄Cl returned the pH_i and g_j to their control values.

The mechanism of the uncoupling effect of flufenamic acid is not known. Flufenamic acid (100 μm) completely blocked g_j (Fig. 5C). Since fenamates are weak acids, their channel blocking capability may manifest due to a reduction of intracellular pH. Indeed, they caused pH_i decay from ∼7.2 to 7.0 (not shown) which could explain the g_j decrease by a maximum of ∼30% (Palacios-Prado et al. 2010), while we recorded full uncoupling. Application of NH₄Cl (15 mm) did not reverse the uncoupling effect of flufenamic acid.

The mechanism of uncoupling by carbenoxolone is largely unexplored. There are no reports indicating detectable differences in its uncoupling efficacy among different Cx isoforms. Carbenoxolone (1 mm), like flufenamic acid, caused a decrease in pH_i by ∼0.2 units, which was too small to explain the observed uncoupling (Fig. 5D). A subsequent increase in pH_i by NH₄Cl (15 mm) did not cause the recovery of g_j. Similarly, arachidonic acid (10 μm) completely blocked g_j and NH₄Cl (15 mm) failed to re-establish the cell–cell coupling (Fig. 5E). Thus, the pH_i-dependent modulation of the blocking capacity of GJ channel inhibitors is apparently inhibitor-type specific.

To test the hypothesis that inhibition of g_j by octanol is mediated by forming hydrogen bonds with histidine residues, which are probably involved in pH-dependent effects (Spray & Burt, 1990), we compared the influence of 0.3 mm octanol in a control condition and after pre-incubation with N-bromosuccinimide (NBS), which reduces histidine's ability to form hydrogen bonds by modifying its imidazole ring. Such an experimental protocol was recently used to demonstrate a Zn²⁺ interaction with histidines of Cxs in the horizontal cells of retina (Sun et al. 2009). After pre-incubation of HeLa cells expressing Cx45 in the solution containing 1 mm NBS for 40 min, a degree of uncoupling by octanol (0.3 mm) was reduced from 0.66 ± 0.05 (n = 5) to 0.24 ± 0.03 (n = 6) (P < 0.001). In addition, the coupling-promoting effect of NH₄Cl (10 mm) was reduced from 203 ± 18% (n = 5) (control) to 157 ± 5% (n = 6) (P < 0.05) (Fig. 6). The proportional diminution of both effects indicates that the action of H⁺ and octanol may take place at the same histidine residue/s. NBS (1 mm) did not affect g_j in HeLaCx45 cells exposed to 1 mm NBS during electrophysiological experiments lasting ∼30 min.

Figure 6. Modulation of g_j by pH_i and LCCAs depends on histidine's ability to form hydrogen bonds.

A, g_j trace measured in HeLaCx45 cell pair, which was pre-incubated for 40 min with N-bromosuccinimide (NBS, 1 mm), demonstrates g_j increase and decrease in response to applications of NH₄Cl and octanol, respectively. g_j changes were significantly smaller than those without pre-incubation with NBS (see Figs 1D and 2C). NH₄Cl, when applied in conjunction with octanol, resulted in full g_j recovery. B, summarized data demonstrating the decay in coupling-promoting effect of NH₄Cl (10 mm) and uncoupling effect of octanol (0.3 mm) relative to the control.

Discussion

This study was stimulated by the highly unexpected observation that in cells expressing Cx45, the uncoupling effect of octanol and other LCCAs can be fully eliminated by intracellular alkalization in the continuous presence of uncouplers. Interestingly, LCCAs as well as carbenoxolone and mefloquine and maybe some other not yet examined uncouplers, have the potential to reduce neurological tremors (Bushara et al. 2004; Martin & Handforth, 2006). More-detailed studies revealed that IC₅₀ for an uncoupling effect of octanol increased from 0.1 to 0.25 and 2.68 mm with an increase in pH_i from 6.9 to 7.2 and 8.1, respectively. Analysis of g_j–V_j plots (Fig. 2C) using a stochastic 4-state model (Paulauskas et al. 2009) revealed that g_j decay under application of octanol was caused by a decrease of V_o,H∼2.6-fold and N_F∼6-fold (Table 1). Interestingly, Cx45 GJs exhibited similar changes under acidification (Palacios-Prado et al. 2010), which reduced V_o,H and N_F without substantial changes in A_H. This suggests that pH_i- and octanol-induced uncoupling of Cx45 GJs share similar mechanisms. The blocking effect of chemical uncouplers is realized mainly through the slow gate that correlates with full but not partial uncoupling. This may explain data showing a reduction of γ_res,H supporting the notion that, at the final state of uncoupling, most channels gate by the slow gate between the open and fully closed states instead of by the fast gate between the open and residual/substate states.

Cells expressing other examined Cxs (26, m30.2, 36, 40, 43, 46, 47 and 50), however, did not exhibit coupling recovery by exposing them to NH₄Cl under a continuous presence of octanol. In cells expressing Cx45, alkalization eliminated the uncoupling effect of isoflurane and mefloquine but not of arachidonic acid, carbenoxolone and flufenamic acid. Due to ionization or tautomerization processes, many drug-like molecules can be transformed to different protonation states. The solubility and membrane permeability of a drug can vary significantly depending on the protonation state, which can be affected by pH and the local surrounding environment. The predicted conformations of a drug-like molecule, as well as the binding mode and binding affinity of ligands with proteins, are all expected to be influenced by the protonation state of a ligand (Shelley et al. 2007). Hence, the pH_i-dependent inhibition of g_j may be explained in several ways.

Reported data demonstrate that the effects of uncouplers, other than LCCAs, are also sensitive to pH_i. It has been shown that the blocking potency of flufenamic acid (Srinivas & Spray, 2003), mefloquine (Srinivas et al. 2001) and halothane (Tamalu et al. 2001) depends on the extracellular pH suggesting that these agents in their uncharged form may cross the membrane more easily. In all our experiments, however, extracellular pH was maintained at 7.4 and GJ uncoupling agents had no or negligible effect on pH_o. Moreover, an elevation of pH_o causes an additional increase in pH_i and g_j of Cx45-expressing cells (Palacios-Prado et al. 2010). The protonated state of uncouplers can also be affected by the intracellular environment when they cross the membrane or dwell in the inner site of the plasma membrane. However, LCCAs cannot be protonated due to their nucleophilicity.

The second possibility is that protons may mediate the interaction of some uncouplers with the receptor. It is broadly assumed that acidification-induced uncoupling is caused by protonation of histidines of Cxs (Spray & Burt, 1990). Hydrogen bonding is one of the factors modulating the 3-dimensional structure of Cxs and therefore their probability to open. Factors that can change this structure include heat, pH, alkanols, etc. Octanol, due to its OH functional group, is very likely to form hydrogen bonds with both proton acceptors and donors (three bonds in total: one with partly positive hydrogen, the other two with two lone pairs of oxygen). Octanol can replace existing hydrogen bonds in connexins by forming the new ones with its OH group, which can lead to transformation of the protein structure. This bonding can be markedly influenced by pH. The only amino acid, which is affected by pH in the physiological range is histidine. Its imidazole ring at pH < 6 is protonated, while at pH > 7 it is neutral, and the nitrogen that does not carry a hydrogen makes a double bond instead. Therefore, at higher pH_i, the ability of octanol to form hydrogen bonds with histidine is reduced and successively the uncoupling properties of octanol in Cx45 are lost. Hydrogen bonds can vary in strength from very weak (1–2 kJ mol⁻¹) to extremely strong (>155 kJ mol⁻¹). Fluorine is the most electronegative of all the elements and the bond between it and hydrogen is ∼6-fold stronger than between hydrogen and oxygen. Three out of five blockers used in our experiments, mefloquine, forane and flufenamic acid, have two or more fluorine atoms in their structure and are very likely to interact with Cxs through hydrogen bonds. However, this mechanism does not fit to flufenamic acid since its effect was not reversed by NH₄Cl.

The third possibility is that protonation of histidines may result in a favourable Cx conformation for interaction with channel blockers. Such an interaction resembles the allosterism with positive cooperativity when histidine serves as an allosteric centre of Cx protein for H⁺, and protonation of histidine causes the increased affinity of GJ channel blocker to the receptor. The receptor site may be highly conserved among the connexins since the blocking effect of octanol is not connexin specific.

There is something unique in the Cx45 structure compared with other Cxs that demonstrates its distinctive ability to reverse octanol-induced uncoupling by alkalization. Modification of histidine residues by N-bromosuccinimide reduced the coupling-promoting effect of NH₄Cl and the uncoupling effect of octanol in Cx45 GJs. However, a role of histidines in the regulation of GJ conductance may differ depending on Cx type. For instance, histidine modification by diethyl pyrocarbonate evoked different effects on Cx46 and Cx50 hemichannels (Beahm & Hall, 2002). However, neither of them exhibited coupling recovery during combined application of octanol and NH₄Cl. This suggests that LCCAs and some other uncouplers act on Cx45 through the formation of hydrogen bonds with histidines distinctly positioned in the Cx45 protein. NBS indeed was not able to completely abolish the actions of both octanol and NH₄Cl in Cx45 GJs even though we occasionally observed nearly complete blocking. These results are consistent with other reports. For instance, Sun et al. (2009) have shown that NBS was not able to completely abolish the effect of Zn²⁺ presumably acting through histidines. A possible explanation may be that Zn²⁺, as well as in our case H⁺, could also act via other residues, which cannot be modified by NBS.

It can be expected that coupling recovery by alkalization somehow relates to an exceptionally high g_j increase in Cx45 GJs under application of NH₄Cl. We can speculate that the potency of alkalization on Cx45 conformational changes, e.g. due to effects on hydrogen bonding, prevail over those caused by LCCAs. It also supports the notion that they both affect the same or closely positioned domains/histidines in Cx45. The latter is supported by the proportional diminution of g_j changes induced by NH₄Cl and LCCA in the presence of NBS, which indicates that the action of H⁺ and octanol may take place at the same histidine residue/s.

It is assumed that all uncoupling agents execute their action through the slow gating mechanism, which closes GJs fully (Bukauskas & Verselis, 2004). A slow (or ‘loop’) gate is located most likely in the gap region of the GJ channel that includes both extracellular loops (Bukauskas & Verselis, 2004; Verselis et al. 2009). Since LCCAs and some other GJ blockers are lipophilic compounds, which probably can affect preferentially extracellular or transmembrane domains of Cxs, then three histidines present in the extracellular loops of Cx45 are likely candidates for such specific interaction as there are no histidines in transmembrane domains. At this point, we can only speculate that histidine-204, which is present in the second extracellular loop of Cx45 but not in other examined Cxs, may be important for the observed alkalization-dependent coupling recovery.

Therefore, the presented data demonstrate for the first time that the uncoupling potency of long carbon chain alkanols, forane and mefloquine on GJ channels is pH_i- and Cx-type dependent. Presumably, this type of modulation is mediated by the formation of hydrogen bonds between uncouplers and histidines of Cx protein.

Glossary

Abbreviations

aHC

apposed hemichannel

Cx

connexin

GJ

gap junction

HeLa cells

human cervix carcinoma cells

LCCA

long carbon chain alkanol

NBS

N-bromosuccinimide

ROI

region of interest

S4SM

stochastic four-state model

uHC

unapposed hemichannel

Author contributions

V.A.S., L.R. and F.F.B. designed the study, performed, analysed and interpreted the experiments, and drafted the manuscript. A.S. designed and performed the experiments relating to NBS involvement in pH- and LCCA-dependent regulation of junctional conductance, and N.P. analysed the experimental data relating to the voltage gating of gap junctions using a S4SM. All authors read and commented on the manuscript. The work of V.A.S., L.R., N.P. and F.F.B. was performed at the Institute of Cardiology at the Lithuanian University of Health Sciences (LUHS) and in the Dominick P. Purpura Department of Neuroscience at the Albert Einstein College of Medicine. The experiments of A.S. were performed at the Institute of Cardiology at LUHS.