The Amino Terminal Domain and Modulation of Connexin36 Gap Junction Channels by Intracellular Magnesium Ions

Abstract

Electrical synapses between neurons in the mammalian CNS are predominantly formed of the connexin36 (Cx36) gap junction (GJ) channel protein. Unique among GJs formed of a number of other members of the Cx gene family, Cx36 GJs possess a high sensitivity to intracellular Mg²⁺ that can robustly act to modulate the strength of electrical synaptic transmission. Although a putative Mg²⁺ binding site was previously identified to reside in the aqueous pore in the first extracellular (E1) loop domain, the involvement of the N-terminal (NT) domain in the atypical response of Cx36 GJs to pH was shown to depend on intracellular levels of Mg²⁺. In this study, we examined the impact of amino acid substitutions in the NT domain on Mg²⁺ modulation of Cx36 GJs, focusing on positions predicted to line the pore funnel, which constitutes the cytoplasmic entrance of the channel pore. We find that charge substitutions at the 8th, 13th, and 18th positions had pronounced effects on Mg²⁺ sensitivity, particularly at position 13 at which an A13K substitution completely abolished sensitivity to Mg²⁺. To assess potential mechanisms of Mg²⁺ action, we constructed and tested a series of mathematical models that took into account gating of the component hemichannels in a Cx36 GJ channel as well as Mg²⁺ binding to each hemichannel in open and/or closed states. Simultaneous model fitting of measurements of junctional conductance, g_j, and transjunctional Mg²⁺ fluxes using a fluorescent Mg²⁺ indicator suggested that the most viable mechanism for Cx36 regulation by Mg²⁺ entails the binding of Mg²⁺ to and subsequent stabilization of the closed state in each hemichannel. Reduced permeability to Mg²⁺ was also evident, particularly for the A13K substitution, but homology modeling of all charge-substituted NT variants showed only a moderate correlation between a reduction in the negative electrostatic potential and a reduction in the permeability to Mg²⁺ ions. Given the reported role of the E1 domain in Mg²⁺ binding together with the impact of NT substitutions on gating and the apparent state-dependence of Mg²⁺ binding, this study suggests that the NT domain can be an integral part of Mg²⁺ modulation of Cx36 GJs likely through the coupling of conformational changes between NT and E1 domains.

Keywords: connexin; electrophysiology; gap junction; intracellular Mg2+; ion permeability.

FIGURE 1

Homology model of Cx36 GJ channel structure. (A) Cx36 GJ channel based on crystal structure of Cx46 (6MHQ) following energy minimization and equilibration. (Inset) The segment of GJ channel corresponding to NT domain (dark gray). The color-coded residues of Cx36 were replaced by the residues presented in panel (B). (B) Sequence alignment of Cx43, Cx36 and Cx46 NT domains.

FIGURE 2

Differences in sensitivity to [Mg²⁺]_p between Cx36 WT and NT variants. Experiments were performed in pairs of RIN cells expressing Cx36 WT and NT variants, tagged with EGFP. (A) Initial junctional conductance (g_j,init) was registered immediately after the patch was opened. (B) Average g_j value of Cx36 and mutants at the steady-state level, normalized to initial conductance, g_j,ss/g_j,init, recorded using pipette solutions with 0.01, 1, or 5 mM [Mg²⁺]_p. Statistically significant (p < 0.05) differences of each mutant compared to WT Cx36 are indicated by asterisks. Data were obtained by applying repeated small voltage ramps from -10 to 10 mV. Error bars correspond to standard errors, and total numbers of experiments are indicated within or above the columns.

FIGURE 3

Averaged data from simultaneous electrophysiological and fluorescence recordings in RIN cells expressing WT or mutated Cx36 GJ channels. (A) Bar graphs of initial g_j values (g_j,init), measured upon establishing a whole-cell patch recording in the second cell of a pair (cell-2) with a pipette containing high intracellular Mg²⁺. Thus, g_j,init is a measure of g_j prior to any appreciable changes in intracellular Mg²⁺. (B) Mean g_j (g_j,_final) assessed at the end of a 5 minute time interval normalized to initial value (g_j,_init) assessed at the time a whole cell recording was established in cell-2. (C) Measurement of the change in Mag-Fluor-4 fluorescence in the same recipient cells (cell-1; FI₁) at the end of the 5 min time interval. FI₁ was normalized to the initial value approximating the resting Mg²⁺ level. All experiments were performed with pipette-2 containing 5 mM (first column) or 10 mM (second column) Mg²⁺. Values represent the mean and standard error; n = 6 for 5 mM Mg²⁺ and n = 5 for 10 mM Mg²⁺.

FIGURE 4

Model fitting to data from simultaneous electrophysiological and fluorescence recordings. (A,B) Solid lines are fits to the electrophysiological (A) and fluorescence imaging (B) data presented in Figure 3 using Model 5 with Hill equation. (C) Reductions in g_j recorded under symmetric Mg²⁺ conditions (both pipettes contained 5 mM Mg²⁺. Error bars denote standard deviations (n = 7). Data in panel (C) was not used in the fits to obtain parameters, but only for validation of the model.

FIGURE 5

Application of mathematical model to simulate Mg²⁺-induced closure of hemichannels, formed of WT and mutant Cx36. (A) Simulated g_j decrease of Cx36 WT GJ channels (g_j) and hemichannels (g₁ and g₂) at different levels of [Mg²⁺]_i. (B) Kinetics of open (i.e., both hemichannels are open) and closed (i.e., both hemichannels reside in closed, Mg²⁺-bound state) state probabilities of GJ channels formed of different Cx36 variants, simulated using [Mg²⁺]₁ = [Mg²⁺]₂ = 5 mM protocol. (C,D) Experimentally measured (upper panels) and simulated (lower panels) g_j decrease and recovery after application of V_j ramps (middle panels), in A13K and H18K channels. Electrophysiological recordings and simulations were performed using [Mg²⁺]₁ = [Mg²⁺]₂ = 5 mM protocol.

FIGURE 6

Electrostatic potentials of Cx36 and NT variants. (A) Slices along the pore of the Cx36 showing the electrostatic potentials. Black lines indicate locations of mutated residues. (B) Cross-sections through the electrostatic potential surfaces of Cx36 WT and mutants in the vicinities of the 3rd, 8th, 13th, and 18th residues viewed from the extracellular side. In both panels (A,B), potentials were scaled between –10 and 10 kTe^–1 (see color scale below). (C) Sums of electrostatic potentials, calculated in the vicinities of the substituted residues. Calculations were performed by summing the average values of electrostatic potentials in 1x1x1 angstrom cubes, located alongside the pore in three layers, closest to the cross-sections presented in panel (B). (D) The relationship between the differences in sums of the electrostatic potentials from panel (C) and the ratios of permeability coefficients P, estimated from Model 5 with Hill equation, for each variant. The black solid line shows a fitted linear regression curve.