2.1. Theoretical Background
The transport of charge through artificial lipid membranes has been studied for a great variety of uncoupling agents.19 Among these we differentiate the heterocyclic aromatic compounds which act as protonophores: benzimidazoles.19–24 It can be observed that the chemical structure of indole is similar to the chemical structure of benzimidazoles. In consequence, we expect that indole and its derivative 4 F-indole will have a charge transport mechanism similar to benzimidazoles. The equilibrium theory described here25, 26 has been previously developed and successfully applied14, 20–23, 26 for molecules, such as 4,5,6,7-tetrachloro-2-trifluoromethyl benzimidazole (TTFB) and 5,6-dichloro-2-trifluoromethyl benzimidazole (DTFB). Therefore, we apply the same theoretical model to indoles.
Experimentally a small ionic current is found to pass directly through the lipid membrane in the absence of the uncoupler molecules. Consequently the measured membrane conductance (Gm) is the sum of the uncoupler-induced membrane conductance (G) plus the leakage membrane conductance (G0).
For simplicity, the dependence of uncoupler-induced membrane conductance as a function of uncoupler concentration, expressed in Equation (1), can be rewritten as Equation (2):
As shown by Equation (1), the model predicts that the membrane conductance increases with the square of the uncoupler concentration. This is a consequence of the cooperative action between two uncoupler molecules. If a monomolecular transport model were to be employed, then a linear increase in the membrane conductance with the uncoupler concentration would be expected.14 In addition, Equation (1) also predicts a dependence of the conductance on the pH of the solution. The conductance is expected to increase when the pH is increased, to reach a maximum when the pH is equal to the pKa of the uncoupler, and to decrease as the pH is further increased.
Therefore, the model predicts that the uncoupler-induced membrane conductance is inversely proportional to the hydrogen concentration present in solution when the pH is much smaller than the pKa of the uncoupler [Eq. (3)].
At constant uncoupler concentration Equation (3) can be rewritten to Equation (4):
In addition, the model predicts that even though the only charge permeating through the membrane is the  complex, the membrane behaves as if it were selectively permeable to [H+] ions. The Nernstian behaviour for the trans-membrane potential (V), required to stop the ionic current that passes through the lipid membrane when a pH difference exists between the CIS and TRANS chamber of the bilayer, is described by Equation (5):25, 26
Equation (5) predicts that at 21 °C the Nernstian trans-membrane potential needed in order to stop the current flow due to a pH gradient of one unit is 58.4 mV. However, if the leakage membrane conductance is comparable to the membrane conductance induced by the uncoupler molecules, then the measured trans-membrane voltage (Vm) is smaller than the Nernstian value.30, 31 In this case, the leakage conductance creates a voltage divider (see the Supporting Information Figure S3) which reduces the measured trans-membrane potential Vm by a factor depending on Gm and G0 as shown in the literature [Eq. (6)]:30, 31
Here, the terms of the equation are the same as previously defined in Equations (1)–(3).
2.2. The Conductance of the Lipid Membrane in the Presence of Indole and 4 F-Indole
First, we analyse our electrophysiology measurements which demonstrate that indole and 4 F-indole transport charge across artificial lipid membranes. A black lipid membrane (BLM) is reconstituted in a classic electrophysiology setup16, 17 and the membrane conductance is measured upon the addition of indoles. In Figure 1 a, b single representative experiments are presented. It can be observed that the specific conductance of the lipid membranes, calculated as described in the Experimental Section (Gmspecific−G0specific), increases with the increase in concentration (Ctot), of indole or 4 F-indole. The dependence of the indole-induced specific conductance (Gmspecific−G0specific) on the indole or 4 F-indole concentration can be described with a quadratic function as predicted by Equation (1) (Figure 1 a, b). This supports the hypothesis that indole forms dimers when uncoupling the membrane. To complement Figure 1 a and b with a better representation of the data, the logarithm of the indole-induced specific conductance, Gmspecific−G0specific, is plotted as a function of the logarithm of the substrate concentration Ctot (Figure 1 c). Here, an average of five different experiments is shown for both indole and 4 F-indole. The data can be described with Equation (2). At concentrations over 3 mM the experimental data increase faster than the proposed fit (Figure 1 c). This is probably due to the additional leakage current passing through the lipid bilayer when the indole or 4 F-indole concentration is increased. In support of this hypothesis we observed that the stability of the lipid bilayer decreases when the indole concentration exceeds 5 mM.
Figure 1. The specific conductance of the lipid membrane (aqueous solution, pH 7.5) as a function of the indole concentration (circles) and 4 F-indole concentration (squares). Specific membrane conductance plotted as a function of the a) indole concentration and b) 4 F-indole concentration. The curves through the points are fitted according to Equation (1). c) Logarithm of the bilayer conductance plotted against the logarithm of the concentration of indole or 4 F-indole (1 mM to 5 mM only). The curves through the points are drawn according to Equation (2). In panels (a) and (b) single traces are shown, while in panel (c) the means and standard deviations are calculated from five independent repeats.
Download figure to PowerPoint
The dependence of membrane conductance on the concentration of indole and 4 F-indole can be explained qualitatively within the framework of the theoretical model. The quadratic dependence of membrane conductance on the indole or 4 F-indole concentration is predicted by Equation (1), and is a consequence of the cooperative action between the uncoupler molecules. The increased conductance in the presence of 4 F-indole is due to the electronegative nature of the F atom, which delocalizes the charge on the indole dimer [I2H−], and thus it reduces the energy required for the deprotonated 4 F-indole to pass through the hydrophobic core of the lipid membrane.32 As the current-versus-voltage characteristics measured for both indole and 4 F-indole are sub-linear (Figure S4), it is suggested33 that the rate-limiting step for the conduction process is the passage of the charged complex through the lipid membrane. Consequently, 4 F-indole is a better ionophore increasing Gmspecific−G0specific by a factor of ∼21 compared to indole.
2.3. The Selectivity of Charge Transport in the Presence of Indole and 4 F-Indole
Furthermore, we tested the selectivity of charge transport for indole and 4 F-indole by investigating the response of the lipid bilayer to a pH gradient. When a pH gradient is formed across a lipid membrane in the presence of indole or 4 F-indole, an ionic current flows through the membrane, even if there is no voltage applied (Figure S4). If the pH is lower in the CIS chamber (Figure S1), an equivalent proton current is flowing from the CIS to the TRANS chamber. For both indole and 4 F-indole, the membrane potential (Vt) required to stop the current generated by the pH gradient has a linear dependence on the magnitude of the pH difference between the TRANS and the CIS chambers of the lipid bilayer apparatus. In Figure 2 we show the dependence of Vt on the pH difference for two concentrations of indole (2.5 mM and 5 mM) and 4 F-indole (0.5 mM and 2.5 mM). A linear fit is applied to the data to determine the measured trans-membrane potential required to stop the current generated by a pH gradient of one pH unit (Vm). For 4 F-indole we found Vm to be 40±2 mV per pH unit at 0.5 mM (Figure 2 a). Increasing the concentration to 2.5 mM leads to an increase in Vm to 55±1 mV per pH unit (Figure 2 b). For indole we find a similar trend, with 39±1 mV per pH unit obtained at 2.5 mM (Figure 2 c) and rising to 49±2 mV per pH unit at 5 mM (Figure 2 d). The trans-membrane potential measured increases towards the Nernstian prediction of 58.6 mV per pH unit, when the concentration of the indoles is increased. This effect is explained by the presence of the leakage conductance (G0 is around ∼2 pS) and was discussed earlier in Equation (6). Therefore, in the next paragraph we present the theoretical predictions obtained with Equation (6) by plugging in the experimental values measured for the leakage and conductance in the presence of indole.
Figure 2. Selectivity of charge transport for indole and 4 F-indole. The dependence of the membrane potential (Vt) required to stop the current generated by a pH difference between the TRANS and the CIS chambers of the lipid bilayer in the presence of a) 0.5 mM 4 F-indole, b) 2.5 mM 4 F-indole, c) 2.5 mM indole, and d) 5 mM indole. The measurements were started at pH 7 in 15 mM PB buffer and 100 mM KCl. The pH was modified in the CIS chamber by the addition of HCl or KOH in aqueous solution. For a pH lower than 7 in the CIS chamber the pH difference is denoted as positive and for a pH higher than 7 as negative. For each set of data at least four different bilayers were investigated and the raw data are presented here. A straight line was fitted through the data and the trans-membrane voltages per pH unit were found from the slope of the linear fit.
Download figure to PowerPoint
The total membrane conductance Gm measured in the presence of 2.5 mM indole is 9.3±2.7 pS and in the presence of 0.5 mM 4 F-indole is 10.9±3.8 pS. Using Equation (6) we predict the measured trans-membrane voltage, Vm, to be 45.8±3.6 mV per pH unit for 2.5 mM indole and 47.3±3.7 mV per pH unit for 0.5 mM 4 F-indole. These values are close to our experimental data presented in Figure 2. We suggest that the slight difference between the measured values and the theoretical predictions comes from an underestimation of the leakage conductance in the presence of indole. As mentioned earlier, a significant increase in leakage conductance at high indole concentrations is suggested by the data in Figure 1 c. At concentrations above 3 mM the increase in conductance with concentration is slightly more than the quadratic dependence. Increasing the concentration to more than 5 mM makes the lipid bilayer prone to rupture. In the case of 4 F-indole the deviation from the quadratic increase is much less than that of indole, and indeed it reflects a trans-membrane voltage closer to the Nernst potential (55±1 mV per pH unit). Hence, our data indicate that indole and 4 F-indole show characteristics typical of protonophores.
Moreover, the pH dependence of the specific membrane conductance minus the specific leakage conductance, Gmspecific−G0specific, in the presence of either 2.5 mM indole or 4 F-indole was measured. Increasing the pH of the solution in the presence of 2.5 mM indole or 2.5 mM 4 F-indole led to a change of ionic conductance through the lipid membrane (Figure 3). Nevertheless, the increase is slower than predicted by the theoretical model [Eq. (4)]. Consequently, we fitted the experimental data to the empirical Equation (7), which is similar to Equation (4):
Figure 3. Dependence of the logarithm of the specific membrane conductance on the pH of the bathing solution in the presence of 2.5 mM indole (circles) or 2.5 mM 4 F-indole (squares). The aqueous solution contained 0.1 M KCl and 15 mM PB buffer, 15 mM potasium citrate and 15 mM tris. The pH was modified by the addition of KOH or HCl. The curves through the points is drawn according to the empirical Equation (7).
Download figure to PowerPoint
where α and β are empirical fitting parameters.
The theoretical model predicts β=1 [Eq. (4)]. Experimentally we obtain β=0.127±0.014 for 2.5 mM indole and β=0.130±0.019 for 2.5 mM 4 F-indole. Thus, the increase in conductance with the increase in pH is less than predicted by Equation (4). This means that the membrane conductance does not strictly follow the increasing formation of ionized dimer species [A2H−] in solution. Therefore, the model fails to explain the pH dependence. In the literature a similar behaviour is observed for other ionophore molecules, for example, decylamine and picric acid.34, 35 To the best of our knowledge a satisfactory explanation for this behaviour has not yet been formulated and is beyond the scope of this paper.
2.4. The Effect of Indole and 4 F-Indole on the Mitochondrial Oxygen Consumption
After discussing the effects of indole and 4 F-indole we now turn to the biological relevance of these protonophores and their effect on eukaryotic mitochondria. The generation of ATP by mitochondria in eukaryotic cells depends on the establishment of a proton gradient across the inner mitochondrial membrane. This proton pumping is associated with the multi-step transfer of electrons from NADH or FADH2 to oxygen through the electron transport chain (ETC). If the proton gradient is reduced, for example, by an ionophore, this then allows the protons to pass through the membrane, operation of the ETC is stimulated, and oxygen consumption is increased. As we showed herein, indole and 4 F-indole are proton ionophores and therefore should stimulate mitochondrial oxygen consumption.
The effect of indole and 4 F-indole on the rate of oxygen consumption by isolated rat liver mitochondria was measured in the presence of succinate as substrate, but in the absence of ADP in order to preclude any proton movement through the ATP-synthase (Figure 4). Below 0.4 mM, indole does not affect the mitochondrial respiration rate, but when the concentration is raised from 0.4 mM to 0.8 mM there is a 60 % increase. In contrast, 4 F-indole affects the mitochondrial oxygen consumption rate at a concentration as low as 0.125 mM. An increase of the 4 F-indole concentration to 0.375 mM raises the rate of mitochondrial oxygen consumption by more than 65 %. Thus, in agreement with the electrophysiology measurements, charge transport across the inner membrane of mitochondria is modified in the presence of indole or 4 F-indole. Both molecules uncouple mitochondrial oxidative phosphorylation, with 4 F-indole being the more effective uncoupler, as expected from our in vitro characterization.
Figure 4. Stimulation of oxygen consumption in rat liver mitochondria by indole (circles) and 4 F-indole (squares). The mean and the standard deviation are calculated from four independent repeats.
Download figure to PowerPoint
Our measurements on diphytanoyl phosphatidylcholine (DPhPC) lipids together with our experiments on mitochondrial oxygen consumption indicate that indole not only affects bacteria, but has the capacity to affect nearby cells that may not themselves produce indole. By its uncoupling action the bacterial indole production could regulate the energy production of surrounding cells. This could give the indole producer an advantage when struggling to survive (e.g. slowing down the energy production in macrophages which are actively fighting against bacteria). Therefore, our data provide a biophysical explanation for how indole may link to the metabolism of bacterial and eukaryotic cells and thus act as an inter-kingdom signal.