Charge at the 46th residue of connexin 50 is crucial for the gap-junctional unitary conductance and transjunctional voltage-dependent gating

Abstract

Gap-junction (GJ) channels are twice the length of most membrane channels, yet they often have large unitary channel conductance (γj). What factors make this possibly the longest channel so efficient in passing ions are not fully clear. Here we studied the lens connexin (Cx) 50 GJs, which display one of the largest γj and the most sensitive transjunctional voltage-dependent gating (Vj gating) among all GJ channels. Introduction of charged residues into a putative pore-lining domain (the first transmembrane and the first extracellular loop border) drastically altered the apparent γj. Specifically, G46D and G46E increased the Cx50 γj from 201 to 256 and 293 pS, respectively and the G46K channel showed an apparent γj of only 20 pS. G46K also drastically altered Vj gating properties in homotypic G46K and heterotypic Cx50/G46K channels, causing an apparent loss of fast Vj-dependent gating transitions and leaving only loop gating transitions at the single channel current records. Both macroscopic and single channel currents of heterotypic Cx50/G46K channels showed a prominent rectification. Our homology structural models indicate that the pore surface electrostatic potentials are a dictating factor in determining the γj. Our data demonstrate, at the whole GJ channel level, a crucial role of the surface charge properties in the first transmembrane/first extracellular border domain in determining the efficiency of ion permeation and the Vj gating of Cx50 and possibly other GJ channels.

Figure 1. Macroscopic V_j gating properties of Cx50 G46D gap junction channels

A, V_j pulses from ±20 mV to ±100 mV in a 20 mV increment were applied to one cell of the cell pair expressing Cx50 or G46D and macroscopic transjunctional currents (I_j) recorded from the other cell are presented. B, normalized G_j,ss of Cx50 (filled circles) and G46D (open circles) were plotted against different V_j. Smooth dash and continuous lines represent best fitting curves of averaged data from Cx50 (n = 6) and G46D (n = 6) channels, respectively, to a two-state Boltzmann function. Cx50, connexin 50.

Figure 2. G46D alters single channel properties

A, representative single channel current records of Cx50 channel (Aa) and G46D channel (Ab) are illustrated in response to different V_j as indicated. Single channel currents of G46D displayed a shortened dwell time of most open events at different V_j and long-lived fully closed state (indicated by arrows in Ab) in response to ±60 mV and +80 mV pulses. Only one brief transition to closed state was observed in Cx50 channel (arrow in Aa). Dotted lines indicate the fully closed current level. B, average single channel slope conductance (γ_j) of G46D channel (n = 8) was much higher than that of Cx50 (n = 8, P < 0.001). C, P_o, P_s and P_c represent the probabilities of the channel in open, subconductance and fully closed state, respectively. Bar graph illustrates the average data from four different cell pairs. Number of asterisks above the bar indicates the level of statistical difference (*P < 0.05, **P < 0.01). G46D channel demonstrated a markedly increased P_c at these V_j. D, open dwell time of G46D channel is shorter than that of Cx50. Dotted lines are the Gaussian fit of a two-term exponential function to the histograms. Time constants τ₁ and τ₂ with their relative weight are shown. τ_mean is the mean open dwell time obtained from the sum of the product of each τ and its relative weight. τ₁, τ₂ and τ_mean are all reduced in G46D channel. Cx50, connexin 50.

Figure 3. Macroscopic and single channel properties of G46E GJs

A, macroscopic junctional currents (I_j) of homotypic G46E channels are shown in response to the same V_j as shown in Fig.1A. G_j,ss–V_j relationships of G46D were constructed (n = 6) and were fitted to Boltzmann functions. Fitting curves of Cx50 (grey dashed lines) are obtained from Fig.1B for comparison. B, linear regression of i_j–V_j plots showed an increased γ_j of G46E channel [n = 4, P < 0.001 vs. Cx50 (same as shown in Fig.2B)]. C, single channel current traces of G46E channel under the indicated V_j showed the existence of the fully open state, subconductance state and fully closed state. Arrows point to the long-lived fully closed state. Temporal expanded trace (inset) with multiple openings indicates that the open dwell times in these open events were short. Cx50, connexin 50.

Figure 4. G46K displays drastically altered V_j gating and single channel properties

A, a set of macroscopic I_j traces of G46K GJs in response to V_j of ±20 ∼ ±100 mV. B, Boltzmann fitting curves of G46K GJs (continuous lines) generated from G_j,ss–V_j plots (n = 6) exhibited lower V_j sensitivities than those of Cx50 GJs (grey dashed lines, same as in Fig.1B). C, single channel current (i_j) of a G46K channel at 80 mV V_j showed a very low unitary conductance (21 pS). Closing and opening current levels are indicated by dotted lines. Despite the frequent transfer to closed state, the dominant state of G46K channel is open state at this V_j. D, i_j of a G46K channel at 120 mV V_j showed slow transitions between open and closed states. A portion of the trace is expanded in temporal domain. Cx50, connexin 50; GJ, gap-junction.

Figure 5. Heterotypic Cx50/G46K channels show asymmetrical V_j gating and rectification

A, two sets of representative I_j of heterotypic Cx50/G46K GJs in response to the V_j protocol (±20 to ±100 mV) applied to the Cx50-expressing cell (top set) or to the G46K-expressing cell (bottom set). I_j inactivations were present only when the Cx50 cell with +V_j (or the G46K cell with –V_j). Initial amplitudes of I_j were also different between the corresponding +V_j and –V_j. B, G_j,ss–V_j plot of heterotypic Cx50/G46K GJs from six cell pairs. Smooth line on the +V_j is the Boltzmann fitting curve. At the –V_j, no V_j gating (I_j inactivation) was evident. Boltzmann fittings of Cx50 channels (grey dashed lines) are shown for comparison. C, initial conductance of +V_j [G_j,ini(+)] and –V_j [G_j,ini(–)] were calculated and the ratio is plotted with V_j. Cx50/G46K GJs showed a strong V_j-dependent rectification. D, heterotypic Cx50/G46K channel showed rectification. i_j was recorded from the G46K cell in response to ±80 mV and ±100 mV V_j pulses (on the Cx50-expressing cell). As indicated in the enlarged box below the current, the gating closure reached the fully closed state (indicated by arrows) and the gating transitions typically took tens of milliseconds. E, when the Cx50 cell with +V_j, the γ_j(+) was 54 pS (from the i_j portion indicated by an asterisk). γ_j(−) with –V_j was 22 pS. Cx50, connexin 50; GJ, gap-junction.

Figure 6. G46D channel show a similar ion preference as that of Cx50

A, with the substitutions of the major salt (either from CsCl to CsGlu or to TEACl) in the pipette solution, single channel recordings of Cx50 and G46D GJ displayed distinctive γ_j. i_j in response to V_j of −80 mV were shown for each of the ion substitutions. B, bar graph shows the average of the γ_j of Cx50 and G46D while using different pipette solutions and their ratios to the control γ_j (using CsCl-based pipette solution). All the γ_j values were obtained by linear regression of i_j–V_j plots. CsGlu, caesium glutamate; Cx50, connexin 50; TEACl, tetraethylammonium chloride.

Figure 7. Homology models of the Cx50 mutants

A, stick view in PyMOL of a portion of the mutant or wild-type Cx50 channels near the 46th residue (top view). Estimated diameters of G46D, G46E and G46K were predicted to decrease as indicated. B, side view of a cut open Cx50 channel shows the pore surface electrostatic potentials (calculated with an adaptive Poisson–Boltzman solver) using dielectric constants of 2 (protein) and 80 (solutions) (Baker et al. 2001). A portion of the Cx50 channel pore surface containing TM1/E1 domains are enlarged as indicated. Electrostatic potentials of the mutant channels at the same position are illustrated. Drastic differences in electrostatic potentials are observed near the mutant residue (dotted horizontal line). Displayed surface electrostatic potentials range from −40 (red) to +40 (blue) kTe^-1. C, when the G46K-expressing cell was held with different polarity of V_j, two different orientations of Lys46 could be observed and are superimposed in stick view in PyMOL. G46K channel with –V_j (or Cx50 side with +V_j in the heterotypic channel) showed a larger diameter than the G46K channel with +V_j, which could play a role in channel rectification. Cx50, connexin 50.