Gap junction (GJ) channels are oligomers of connexins forming channels linking neighboring cells. GJs formed by different connexins show distinct unitary channel conductance (γj), transjunctional voltage-dependent gating (Vj-gating) properties, and modulation by intracellular magnesium ([Mg2+]i). The underlying molecular determinants are not fully clear. Previous experimental evidence indicates that residues in the amino terminal (NT) and initial segment of the first extracellular (E1) domain influence the γj, Vj-gating, and/or [Mg2+]i modulation in several GJs. Increasing negatively charged residues in Cx50 (connexin50) E1 (G46D or G46E) increased γj, while increasing positively charged residue (G46K) reduced the γj. Sequence alignment of Cx50 and Cx37 in the NT and E1 domains revealed that in Cx50 G8 and V53, positions are negatively charged residues in Cx37 (E8 and E53, respectively). To evaluate these residues together, we generated a triple variant in Cx50, G8E, G46E, and V53E simultaneously to study its γj, Vj-gating properties, and modulation by [Mg2+]i. Our data indicate that the triple variant and individual variants G8E, G46E, and V53E significantly increased Cx50 GJ γj without a significant change in the Vj gating. In addition, elevated [Mg2+]i reduced γj in Cx50 and all the variant GJs. These results and our homology structural models suggest that these NT/E1 residues are likely to be pore-lining and the variants increased the negative electrostatic potentials along the GJ pore to facilitate the γj of this cation-preferring GJ channel. Our results indicate that electrostatic properties of the Cx50 GJ pore are important for the γj and the [Mg2+]i modulation.

Introduction

Gap junction (GJ) intercellular channels enable direct cell-to-cell communication between neighboring cells by facilitating exchange of ions, small metabolites, and other biological molecules under 1 kDa in size [1,2]. GJs are ubiquitously expressed throughout the body and are essential for maintaining metabolic and electrical homeostasis [3]. Each GJ channel consists of two hemichannels which are docked head-to-head at the extracellular space of two adjacent cells [4]. Hemichannels are hexamers of connexins. There are 21 genes in the human genome encoding different connexin isoforms. Each connexin has a distinct pattern of tissue distribution [5]. All connexins are assumed to have the same structural topology consisting of four transmembrane domains (M1–M4), two extracellular loops (E1 and E2), a cytoplasmic loop (CL), and an amino-terminus and a carboxyl-terminus (NT and CT, respectively) found in the cytosol. Depending on the component connexin isoform, GJs can display a wide variety of biophysical properties including unitary channel conductance (γj), transjunctional voltage-dependent deactivation (known as Vj gating), and modulation by intracellular divalent cations, such as calcium (Ca2+) and magnesium (Mg2+). Molecular mechanisms responsible for these channel properties are not fully resolved.

GJ single channel conductance (γj), defined as the rate of ion permeation through a single GJ channel, ranges between 9 pS (in mouse Cx30.2 GJ) and ∼300 pS (Cx37 GJ) [69]. Structural determinants to this wide range in γj are not fully understood. One hypothesis suggests that γj is partially determined by the channel pore diameter and length [10]. With twice the length of most membrane ion channels, GJ channel γjs can be as high as 100–300 pS, which is probably due to a much larger pore diameter [10,11]. Contradictorily, studies on GJ dye transfer using differently sized dyes have consistently demonstrated that Cx37 GJ was the least permeable to large dyes despite having the largest γj [7,1214], challenging the pore size as the main determinant for γj. Several functional studies on Cx46 and Cx50 (connexin50) GJs suggest that electrostatic properties of the residues in NT and E1 domains are an important determinant to γj [1521]. For example, Cx50 G46D and G46E mutants both increased the proportion of negatively charged residues at the E1 domain and were found to increase their GJ γj, while the G46K mutant added a positively charged residue in E1 and showed a much reduced γj [22]. Mutations in the Cx50 NT domain with an increased positive charge (N9R) or a decreased negative charge (D3N) were also found to reduce γj or the equivalent single hemichannel conductance [19,23]. Consistent with these results, substituted cysteine accessibility study on the M1/E1 residues of Cx46 with positively or negatively charged methanethiosulfonate reagents often resulted in reduced or no change (or elevated in D51C) single hemichannel conductance, respectively [15], indicating the importance of charges in the M1/E1 domain. To follow this idea, we looked into residue differences in the NT and E1 domains between Cx50 and Cx37 (with the largest γj of all characterized GJs) and found two key residue differences at the 8th and 53rd positions (Figure 1). At these positions, negatively charged glutamic acid residues (E8 and E53) were found in Cx37, whereas small nonpolar residues (G8 and V53) were in Cx50 (Figure 1). As both Cx50 and Cx37 GJs show cation-preference [14,22,24], we propose that these negatively charged residues in Cx37 GJ channels may be pore-lining residues that facilitate the rate of cation permeation, consequently imparting a larger γj. To further investigate the charge changes in the residues in NT and E1 domains, we generated a triple mutation G8EG46EV53E and individual single mutations in Cx50 to study their GJ γj changes.

Sequence alignment of the NT and the E1 domain of mouse Cx50 and Cx37.

Figure 1.
Sequence alignment of the NT and the E1 domain of mouse Cx50 and Cx37.

Mouse Cx50 residues 1–20 (NT) and 45–61 (part of E1) are aligned with those of Cx37. Three residues are targeted (with asterisks) to generate individual variants or a triple variant. G8E and V53E were designed due to that these residues in Cx37 (with a larger γj) are negatively charged glutamate (E). G46E was chosen due to a previous study showed that this variant can substantially increase unitary GJ conductance (γj) [22]. The NT and E1 domains of both Cx50 and Cx37 contain lots of negatively charged residues (red).

Figure 1.
Sequence alignment of the NT and the E1 domain of mouse Cx50 and Cx37.

Mouse Cx50 residues 1–20 (NT) and 45–61 (part of E1) are aligned with those of Cx37. Three residues are targeted (with asterisks) to generate individual variants or a triple variant. G8E and V53E were designed due to that these residues in Cx37 (with a larger γj) are negatively charged glutamate (E). G46E was chosen due to a previous study showed that this variant can substantially increase unitary GJ conductance (γj) [22]. The NT and E1 domains of both Cx50 and Cx37 contain lots of negatively charged residues (red).

It has been documented that macroscopic GJ conductance (Gj) of Cx36 GJ could be elevated/reduced by decreasing/increasing intracellular Mg2+ concentration ([Mg2+]i), respectively, with altered Vj-gating properties [25,26]. It is not clear if Cx50 GJ could be modulated by different [Mg2+]i, specifically in the context of Vj-gating properties and γj. Our experimental results with the triple and individual Cx50 variants and our homology structure models suggest that these residues are likely lining the pore to facilitate the rate of cation permeation and that [Mg2+]i can modulate the γj of Cx50 GJ without substantially changing its Vj-gating properties. We provide experimental evidence on the role of pore electrostatic properties regulating channel permeation and modulation by [Mg2+]i at the individual Cx50 or its variant GJ channels.

Methods

Construction of Cx50 variants

Untagged expression vector of mouse Cx50 cDNA was generated by polymerase chain reaction and inserted into pIRES2-EGFP vector [22]. The untagged Cx50 construct was used as a template to generate single-point variants, G8E, G46E, and V53E using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA, U.S.A.). The primers used to generate G8E, forward: 5′-TGG AGT TTC CTG GAA AAC ATC TTG GAA-3′, reverse: 5′-TTC CAA GAT GTT TTC CAG GAA ACT CCA-3′, and V53E, forward: 5′-G CAA TCT GAT TTT GAA TGC AAC ACC CAG-3′, reverse: 5′-CTG GGT GTT GCA TTC AAA ATC AGA TTG C-3′. Primers for G46E have been described previously [22]. The Cx50 triple variant, G8EG46EV53E (or G8G46V53), was generated using sequential mutagenesis.

Cell culture and transient transfection

GJ-deficient mouse neuroblastoma (N2A) cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA, U.S.A.) and cultured in Dulbecco's Modified Eagle Medium (DMEM) (Life Technologies Corporation, Grand Island, NY, U.S.A.) containing 10% fetal bovine serum, 1% penicillin, 1% streptomycin, and 1% GlutaMax. Cells were transferred onto 35 mm dishes at 50% confluency to be cultured overnight. Transfection was performed on the following day with 1 µg of Cx50 or variant vector and 2 µl of X-tremeGENE HP DNA transfection reagent (Roche Diagnostics GmbH, Indianapolis, IN, U.S.A.) for 5 h. Transfection reagent was then replaced with DMEM to culture overnight. On recording day, N2A cells were replated onto 1 cm glass coverslips and incubated for another 1–2 h prior to electrophysiological recordings; with a few exceptions, a much longer incubation time was needed to obtain a higher experimental yield of stable single channel and/or macroscopic recordings.

Homology structure modeling

Sequence alignment of mouse Cx50 and human Cx26 demonstrated an overall sequence identity of 49% [20,22]. The crystal structure of Cx26 (2ZW3) was used as a template to construct a structural model for Cx50 [27,28]. Abnormal inter-atomic contacts were adjusted manually in COOT and then revised by Crystallography and NMR System (CNS) energy refinement. Manual inspection of the model was done to assess structural validity [22]. To calculate electrostatic potentials of all atoms in Cx50 and variants, adaptive Poisson–Boltzmann solver [29] and PDB2PQR server (http://nbcr-222.ucsd.edu/pdb2pqr_1.8/) were used with parameters described previously [22,27]. PyMOL was used to estimate the GJ pore diameters and construct structural presentations of Cx50 as described earlier [22]. The electrostatic potentials in the center of the channel were calculated as described in our earlier study [20]. Homology models were aligned first with the x-axis using PoreWalker [30] and then rotated 90° to align with the z-axis. Electrostatic potentials were calculated using APBS v.1.5 [29] as implemented in PyMOL v.2.0.6 and using the default settings [31]. The output ‘dx’ file comprises a three-dimensional grid of electrostatic potentials. Values for the grid points along the z-axis (in the center of the channel) were extracted from the dx file with the program OpenDX 4.4.4 (http://opendx.org/index2.php) using ‘Potential.net’, which is a modified version of a sample program, ‘PlotLine.net’ that is included with the OpenDX software. Potential.net is provided in Supplementary Material.

Electrophysiological recordings

GJ channel properties of cell pairs expressing either Cx50 or one of its variants were measured using the dual whole-cell patch-clamp technique as described previously [32]. A coverslip containing transfected N2A cells was transferred to a recording chamber placed on an upright microscope (BX51WI, Olympus). N2A cells were bathed in an extracellular solution at room temperature containing (in mM): 135 NaCl, 2 CsCl, 2 CaCl2, 1 MgCl2, 10 HEPES, 5 KCl, 5 d-glucose, 2 Na pyruvate, and 1 BaCl2 at pH 7.4, and an osmolarity of 320 mOsm. Isolated cell pairs with green fluorescence (GFP-positive), indicating a successful transfection, were selected for dual patch-clamp recording. Glass patch micropipettes were filled with intracellular solution (ICS) containing (in mM): 130 CsCl, 10 EGTA, 0.5 CaCl2, and 10 HEPES at pH 7.2, and 295 mOsm (resistance 2–4 MΩ). Sensitivity of GJ channels to different [Mg2+]i was tested by using different ICSs without Mg2+ (Mg2+-free ICS), or the following concentrations of [MgCl2]i, 0.1, 1, or 3 mM. Whole-cell voltage clamp was performed on each cell in the recorded cell pair. One cell of the pair was clamped at 0 mV while the apposed cell was administered a series of voltage steps ranging from ±20 to ±100 mV in 20 mV increments for a duration of 7 s per voltage step [19]. Recorded GJ currents were amplified with MultiClamp 700A (Molecular Devices, Sunnyvale, CA, U.S.A.) and digitized using an ADDA converter at a sampling frequency of 10 kHz (Digidata 1322A, Molecular Devices, Sunnyvale, CA, U.S.A.). All unitary channel currents (ij) and macroscopic junctional currents (Ij) were acquired using pClamp9.2. Selection criteria for single channel analysis were cell pairs expressing one or two operational channels [19,22]. Representative ij recordings chosen for analysis and illustrations underwent digital low-pass filter (200 Hz) with Clampfit (pClamp9, Molecular Devices, Sunnyvale, CA, U.S.A.). All points current amplitude histograms fitted with two Gaussian functions measured the mean and variance of baseline and open channel current amplitude to determine ij [19]. The ijs of different cell pairs were averaged under the same transjunctional voltage (Vj), regardless of Vj polarity, to generate an ij − Vj plot. Linear regression of ij − Vj plot with three different Vjs was used to calculate slope unitary conductance (γj).

To analyze Vj-gating properties, steady-state Ijs, found near the end of a Vj pulse, were normalized to the peak Ijs, found at the beginning of the pulse for each tested Vjs to obtain normalized steady-state conductance (Gj,ss). Gj,ss was plotted at corresponding positive and negative Vjs to obtain Gj,ss − Vj plot, which was fitted with a two-state Boltzmann equation:

 
formula

V0 is the voltage at which conductance is reduced by half [(Gmax − Gmin)/2], Gmax is the maximum normalized conductance, Gmin is the normalized voltage-insensitive residual conductance, and A defines the slope of the curve, reflecting Vj-gating sensitivity [33]. To avoid voltage clamp errors, only cell pairs with a Gj lower than 5 nS were selected for Vj-gating analysis [34,35].

Data analysis

All data are expressed as mean ± SEM. Comparison between Cx50 and G8G46V53 GJ unitary conductance (γj) in [Mg2+]i-free was tested using Student's unpaired t-test. For multiple groups of data (γjs of each single and the double or triple mutant), we used one-way ANOVA followed by Tukey's post hoc test to compare statistical difference between each of the variant and wild-type Cx50. Similarly, one-way ANOVA followed by Tukey's post hoc test was used to compare Boltzmann fitting parameters of each variant with wild-type Cx50 (V0, Gmin, and A) or as specified.

Results

Single channel conductance (γj) of Cx50 G8EG46EV53E GJ

Representative unitary channel currents (ijs) of triple-variant G8EG46EV53E (or G8G46V53) and Cx50 GJ in response to Vj pulses of 40, 60, and 80 mV are illustrated in Figure 2A. All-point histograms were constructed for a representative segment of ijs containing main open state as indicated to measure ij amplitudes of G8G46V53 and wild-type Cx50 (Figure 2B). ijs of different G8G46V53 cell pairs were averaged under the same Vj and were plotted with Vj to obtain ij − Vj plot (Figure 2C). Linear regression of ij − Vj plot was used to determine the slope single channel conductance (γj = 329 ± 10 pS), which was significantly larger than the slope γj of Cx50 (219 ± 5 pS, P < 0.001, Figure 2C). At least two subconductance states (also known as substates or residual states, see Figure 2A) were observed for the G8G46V53 GJ at different Vjs (Figure 2, open arrows). The main subconductance state (at ±80 mV Vjs) had an average conductance of 65 ± 5 pS (n = 6) with relatively long dwell time and frequently observed throughout the entire duration of the Vj pulse. Another subconductance state with a higher averaged conductance (119 ± 18 pS, n = 5) was observed usually at the beginning of the Vj pulse with a shorter dwell time (Figure 2A, open arrow on 80 mV Vj). Fully closed state during this Vj pulse was also commonly observed (see black arrow in Figure 2A). For Cx50 GJ under the same recording conditions, a significantly lower subconductance state (44 ± 6 pS, n = 10, P < 0.05) was frequently observed with long dwell time throughout the entire Vj pulse (see open arrows in Figure 2A). It is very rare to observe high level of subconductance state and fully closed state in Cx50 GJ. Different subconductance states at other Vjs were not systematically analyzed due to low signal/noise ratio, rare appearance, and/or lots of variation.

The γj of Cx50 G8EG46EV53E GJ is drastically higher than that of Cx50.

Figure 2.
The γj of Cx50 G8EG46EV53E GJ is drastically higher than that of Cx50.

(A) Representative single channel currents (ijs) for Cx50 G8EG46EV53E (G8G46V53) and Cx50 GJ are illustrated in response to indicated Vj pulses. Boxed regions of the currents were used to generate the all-point histograms. (B) All-point histograms of ijs are shown under different Vjs as indicated. The current amplitude of main open state was determined by fitting the histogram with Gaussian functions. The conductance of main open state was calculated (shown on each of the histograms). (C) Linear regressions of ij − Vj plots (shown on the left panel) of Cx50 (gray dashed line) and G8G46V53 (black line) were used to obtain slope γj. Number of cell pairs included in average γj analysis is indicated in the bar graph (right panel). Student's unpaired t-test revealed that the γj of G8G46V53 GJ is significantly higher than that of Cx50 (***P < 0.001).

Figure 2.
The γj of Cx50 G8EG46EV53E GJ is drastically higher than that of Cx50.

(A) Representative single channel currents (ijs) for Cx50 G8EG46EV53E (G8G46V53) and Cx50 GJ are illustrated in response to indicated Vj pulses. Boxed regions of the currents were used to generate the all-point histograms. (B) All-point histograms of ijs are shown under different Vjs as indicated. The current amplitude of main open state was determined by fitting the histogram with Gaussian functions. The conductance of main open state was calculated (shown on each of the histograms). (C) Linear regressions of ij − Vj plots (shown on the left panel) of Cx50 (gray dashed line) and G8G46V53 (black line) were used to obtain slope γj. Number of cell pairs included in average γj analysis is indicated in the bar graph (right panel). Student's unpaired t-test revealed that the γj of G8G46V53 GJ is significantly higher than that of Cx50 (***P < 0.001).

γjs of G8E, G46E, and V53E GJs

Having learned that the triple variant increased the γj of Cx50 GJ channels, the extent of each residue's contribution to this increase was investigated. Representative ijs of G8E, G46E, and V53E GJs in response to Vj pulses of 40, 60, and 80 mV are illustrated in Figure 3A. The averaged ijs were plotted against corresponding Vjs for each variant to obtain ij − Vj plot (Figure 3B). Linear regression of ij − Vj plot was used to obtain each variant slope γj and compared with that of Cx50 (gray dashed lines in Figure 3B). The slope γjs of G8E (254 ± 2 pS, n = 7) and G46E (272 ± 9 pS, n = 7), but not V53E (230 ± 5 pS, n = 4), were significantly higher than that of Cx50 (Figure 3C, P < 0.001 for both G8E and G46E).

γj of G8E and G46E, but not V53E, GJ channels were higher than that of Cx50 GJ.

Figure 3.
γj of G8E and G46E, but not V53E, GJ channels were higher than that of Cx50 GJ.

(A) Representative ijs for G8E, G46E, and V53E are illustrated in response to indicated Vjs. (B) Linear regressions of ij − Vj plots for each single variant (black line in each panel) were used to obtain slope γjs. Cx50 regression line (dashed gray line) was plotted for comparison. (C) The average slope γjs of the GJs of G8G46V53 as well as three individual variants are plotted as a bar graph. The statistical differences of each variant in comparison with γj Cx50 GJ channel are shown (***P < 0.001).

Figure 3.
γj of G8E and G46E, but not V53E, GJ channels were higher than that of Cx50 GJ.

(A) Representative ijs for G8E, G46E, and V53E are illustrated in response to indicated Vjs. (B) Linear regressions of ij − Vj plots for each single variant (black line in each panel) were used to obtain slope γjs. Cx50 regression line (dashed gray line) was plotted for comparison. (C) The average slope γjs of the GJs of G8G46V53 as well as three individual variants are plotted as a bar graph. The statistical differences of each variant in comparison with γj Cx50 GJ channel are shown (***P < 0.001).

Vj-gating properties of GJs formed by G8G46V53 and individual variants

Increases in γj may alter the Vj distribution along the elongated GJ pore, which could influence the sensor or gate responsible for Vj gating. To test this, we investigated Vj-gating properties of these variant GJs. Representative macroscopic transjunctional currents (Ij) from cell pairs expressing Cx50, G8G46V53, G8E, G46E, and V53E GJ channels in response to corresponding Vjs are shown in Figure 4. At higher Vjs (40–100 mV), Ijs from all constructs showed a mirror symmetrical Vj-dependent deactivation as those observed from Cx50 GJ (Figure 4). Normalized steady-state junctional conductance (Gj,ss) was plotted against corresponding Vjs and the obtained plot was fitted with a two-state Boltzmann fitting curve for each polarity of Vj. Each variant was plotted with wild-type Cx50 GJ for comparison (dashed smooth gray lines in Figure 4). Statistical tests revealed no major alterations when comparing each of the variant's fitting parameters with those of Cx50 GJs (for Boltzmann fitting parameters of each variant GJs, see Table 1).

The GJs of Cx50 variants showed little change in Vj gating.

Figure 4.
The GJs of Cx50 variants showed little change in Vj gating.

Representative superimposed macroscopic junctional currents (Ijs) for Cx50, G8G46V53, G8E, G46E, and V53E GJ channels in response to the Vj pulses are shown. Normalized steady state to peak junctional conductance (Gj,ss) are plotted as a function of Vj and fitted with two-state Boltzmann equation on each Vj polarity. Boltzmann fitting curves of each variant (solid black lines) are plotted with Cx50 (dashed gray lines) for comparison.

Figure 4.
The GJs of Cx50 variants showed little change in Vj gating.

Representative superimposed macroscopic junctional currents (Ijs) for Cx50, G8G46V53, G8E, G46E, and V53E GJ channels in response to the Vj pulses are shown. Normalized steady state to peak junctional conductance (Gj,ss) are plotted as a function of Vj and fitted with two-state Boltzmann equation on each Vj polarity. Boltzmann fitting curves of each variant (solid black lines) are plotted with Cx50 (dashed gray lines) for comparison.

Table 1
Boltzmann fitting parameters for the Vj gating of Cx50 and its variant GJs at different [Mg2+]i

Data are presented as means ± SEM and V0 are absolute values. One-way ANOVA followed by Tukey's post hoc test was used to compare each Boltzmann fitting parameter of the variant GJ to wild-type Cx50. No statistical differences were observed, except G8E at +Vj (P < 0.05). Student's unpaired t-test was used to compare each of the Boltzmann fitting parameter of each variant or Cx50 at 0 (in the plain text) and 3 mM (in the bold text) [Mg2+]i. No statistical differences were observed, except the V0 of Cx50 at +Vj (*P < 0.05) and the A of G8E + Vj (**P < 0.01).

Cx50 variant [Mg2+]i (mM) Vj polarity Gmin V0 (mV) A 
Cx50 0 (n = 6) 0.10 ± 0.02 34.0 ± 1.4 0.14 ± 0.02 
− 0.10 ± 0.03 34.4 ± 2.4 0.14 ± 0.04 
3 (n = 6) + 0.14 ± 0.3 41.6 ± 2.0* 0.11 ± 0.03 
− 0.12 ± 0.05 35.0 ± 3.5 0.11 ± 0.04 
G8G46V53 0 (n = 4) 0.13 ± 0.03 28.1 ± 2.5 0.17 ± 0.04 
− 0.09 ± 0.03 30.3 ± 2.1 0.15 ± 0.03 
3 (n = 4) + 0.09 ± 0.02 28.9 ± 1.6 0.20 ± 0.03 
− 0.12 ± 0.04 29.4 ± 3.1 0.13 ± 0.04 
G8E 0 (n = 5) 0.13 ± 0.01 28.6 ± 0.8 0.19 ± 0.01 
− 0.14 ± 0.01 33.0 ± 1.6 0.26 ± 0.05 
3 (n = 3) + 0.12 ± 0.01 25.6 ± 2.0 0.28 ± 0.1** 
− 0.12 ± 0.02 28.2 ± 1.4 0.16 ± 0.02 
G46E 0 (n = 4) 0.14 ± 0.02 32.9 ± 2.2 0.09 ± 0.02 
− 0.12 ± 0.03 32.4 ± 2.1 0.14 ± 0.03 
3 (n = 4) + 0.10 ± 0.07 34.1 ± 2.8 0.09 ± 0.02 
− 0.10 ± 0.03 37.8 ± 2.5 0.09 ± 0.02 
V53E 0 (n = 4) 0.09 ± 0.02 34.1 ± 1.3 0.17 ± 0.03 
− 0.14 ± 0.03 30.2 ± 2.5 0.16 ± 0.04 
3 (n = 3) + 0.13 ± 0.05 35.0 ± 4.1 0.17 ± 0.10 
− 0.09 ± 0.05 31.7 ± 4.9 0.09 ± 0.03 
Cx50 variant [Mg2+]i (mM) Vj polarity Gmin V0 (mV) A 
Cx50 0 (n = 6) 0.10 ± 0.02 34.0 ± 1.4 0.14 ± 0.02 
− 0.10 ± 0.03 34.4 ± 2.4 0.14 ± 0.04 
3 (n = 6) + 0.14 ± 0.3 41.6 ± 2.0* 0.11 ± 0.03 
− 0.12 ± 0.05 35.0 ± 3.5 0.11 ± 0.04 
G8G46V53 0 (n = 4) 0.13 ± 0.03 28.1 ± 2.5 0.17 ± 0.04 
− 0.09 ± 0.03 30.3 ± 2.1 0.15 ± 0.03 
3 (n = 4) + 0.09 ± 0.02 28.9 ± 1.6 0.20 ± 0.03 
− 0.12 ± 0.04 29.4 ± 3.1 0.13 ± 0.04 
G8E 0 (n = 5) 0.13 ± 0.01 28.6 ± 0.8 0.19 ± 0.01 
− 0.14 ± 0.01 33.0 ± 1.6 0.26 ± 0.05 
3 (n = 3) + 0.12 ± 0.01 25.6 ± 2.0 0.28 ± 0.1** 
− 0.12 ± 0.02 28.2 ± 1.4 0.16 ± 0.02 
G46E 0 (n = 4) 0.14 ± 0.02 32.9 ± 2.2 0.09 ± 0.02 
− 0.12 ± 0.03 32.4 ± 2.1 0.14 ± 0.03 
3 (n = 4) + 0.10 ± 0.07 34.1 ± 2.8 0.09 ± 0.02 
− 0.10 ± 0.03 37.8 ± 2.5 0.09 ± 0.02 
V53E 0 (n = 4) 0.09 ± 0.02 34.1 ± 1.3 0.17 ± 0.03 
− 0.14 ± 0.03 30.2 ± 2.5 0.16 ± 0.04 
3 (n = 3) + 0.13 ± 0.05 35.0 ± 4.1 0.17 ± 0.10 
− 0.09 ± 0.05 31.7 ± 4.9 0.09 ± 0.03 

[Mg2+]i modulated the γjs of G8G46V53 and Cx50

To investigate G8G46V53 and Cx50 GJ channel's sensitivity to [Mg2+]i, three intracellular concentrations of Mg2+ (0.1, 1, or 3 mM) were tested in independent cell pairs. No added Mg2+ ICS ([Mg2+]i = 0 mM) used in the previous experiments was used as a baseline control. Representative ijs of G8G46V53 and Cx50 GJs at different [Mg2+]i are illustrated in Figure 5A. The same procedure as described in Figure 2 was used to obtain average slope γjs. The γjs were plotted as bar graphs to compare the dose-dependent modulation by [Mg2+]i for G8G46V53 and Cx50 GJs (Figure 5B). With increasing [Mg2+]i, the γjs decreased significantly for both GJs. The γjs of G8G46V53 GJ showed a larger relative reduction at each dose of [Mg2+]i (21%, 31%, and 39% reduction for 0.1, 1, and 3 mM, respectively), while the reduction in the γjs of Cx50 GJ was moderate (5%, 12%, and 20%, respectively, see Figure 5B, right panel). The main subconductance state was measured for G8G46V53 (44 ± 7 pS, n = 7) and Cx50 (24 ± 3 pS, n = 7) at 3 mM Mg2+ and Vjs of ±80 mV, and both were lower than their respective subconductance in Mg2+-free ICS.

G8G46V53 GJ channels show a greater decrease in γj with increasing [Mg2+]i than Cx50 GJ.

Figure 5.
G8G46V53 GJ channels show a greater decrease in γj with increasing [Mg2+]i than Cx50 GJ.

(A) Representative ijs for G8G46V53 and Cx50 at 0.1, 1, and 3 mM [Mg2+]i are illustrated in response to indicated Vjs. (B) Average slope γjs of G8G46V53 GJ channels (left panel) and Cx50 GJ channels (right panel) at all tested [Mg2+]i are shown. The number of cell pairs (numbers on each bar) and statistical differences of the slope γj are indicated (***P < 0.001).

Figure 5.
G8G46V53 GJ channels show a greater decrease in γj with increasing [Mg2+]i than Cx50 GJ.

(A) Representative ijs for G8G46V53 and Cx50 at 0.1, 1, and 3 mM [Mg2+]i are illustrated in response to indicated Vjs. (B) Average slope γjs of G8G46V53 GJ channels (left panel) and Cx50 GJ channels (right panel) at all tested [Mg2+]i are shown. The number of cell pairs (numbers on each bar) and statistical differences of the slope γj are indicated (***P < 0.001).

[Mg2+]i modulated γjs of G8E, G46E, and V53E GJs

To examine which residue in the triple variant was the most sensitive to elevated [Mg2+]i, 3 mM Mg2+-ICS was used to study the GJs of single variants, G8E, G46E, or V53E. Representative ijs in response to corresponding Vjs are shown in Figure 6A. The averaged ijs were plotted against corresponding Vjs for each single mutation at 3 mM Mg2+-ICS to obtain ij − Vj plot. Linear regression of ij − Vj plot was used to obtain each variant slope γj at 3 mM Mg2+-ICS compared with that of their corresponding slopes at 0 mM Mg2+-ICS (Figure 6B). The average γjs of G8E, G46E, and V53E GJs were significantly reduced by 23%, 31%, and 21%, respectively, in 3 mM [Mg2+]i (Figure 6B,C, P < 0.001 in each variant).

The γjs of individual variant GJs reduced in 3 mM [Mg2+]i.

Figure 6.
The γjs of individual variant GJs reduced in 3 mM [Mg2+]i.

(A) Representative ijs for G8E, G46E, and V53E GJs at 3 mM [Mg2+]i are illustrated in response to indicated Vjs. (B) Linear regressions of ij − Vj plots of G8E, G46E, and V53E GJs to obtain slope γjs in 3 mM [Mg2+]i. The linear regression lines of the GJs in 0 mM [Mg2+]i are also shown for comparison, G8E (dashed black line), G46E (solid gray line), and V53E (dotted black line). (C) Representative ijs for G8EG46E GJs at 0 and 3 mM [Mg2+]i are illustrated in response to indicated Vjs. Linear regressions of ij − Vj plots of G8EG46E GJs to obtain slope γjs in 0 and 3 mM [Mg2+]i. (D) Comparing the slope γjs of Cx50, the triple-, double-, and single-variant GJ channels under 0 and 3 mM [Mg2+]i. Number of cell pairs are indicated. A two-way ANOVA followed by a Bonferroni post hoc test was used to compare average slope γj of each variant under 0 and 3 mM [Mg2+]i. Statistical differences are shown (***P < 0.001).

Figure 6.
The γjs of individual variant GJs reduced in 3 mM [Mg2+]i.

(A) Representative ijs for G8E, G46E, and V53E GJs at 3 mM [Mg2+]i are illustrated in response to indicated Vjs. (B) Linear regressions of ij − Vj plots of G8E, G46E, and V53E GJs to obtain slope γjs in 3 mM [Mg2+]i. The linear regression lines of the GJs in 0 mM [Mg2+]i are also shown for comparison, G8E (dashed black line), G46E (solid gray line), and V53E (dotted black line). (C) Representative ijs for G8EG46E GJs at 0 and 3 mM [Mg2+]i are illustrated in response to indicated Vjs. Linear regressions of ij − Vj plots of G8EG46E GJs to obtain slope γjs in 0 and 3 mM [Mg2+]i. (D) Comparing the slope γjs of Cx50, the triple-, double-, and single-variant GJ channels under 0 and 3 mM [Mg2+]i. Number of cell pairs are indicated. A two-way ANOVA followed by a Bonferroni post hoc test was used to compare average slope γj of each variant under 0 and 3 mM [Mg2+]i. Statistical differences are shown (***P < 0.001).

It is noted that the GJ of V53E showed a similar level of γj and percentage reduction in the γj in 3 mM Mg2+ comparing with those of Cx50 GJ. This observation raised the question of whether the residue change at this position (V53E) plays any role in the triple mutant (G8G46V53). To explore this further, we generated a double-variant, G8EG46E, without any change at the V53 position in Cx50. Expression of Cx50 G8EG46E in N2A cells resulted in formation of functional GJ channels with a slope γj of 295 ± 5 pS using a Mg2+-free ICS and a slope γj of 215 using an ICS containing 3 mM Mg2+ (Figure 6C,D), indicating that V53E does play a role in facilitating ion flow through the triple-variant GJ channel resulting in a significantly higher γj of the triple variant than that of the double-variant (G8EG46E) GJ.

Vj-gating properties of Cx50 and each variant GJs under different [Mg2+]i

To determine whether elevated [Mg2+]i influences Vj-gating properties of Cx50 and the variant GJ channels, we studied Vj gating using pipette solution containing 3 mM [Mg2+]i. Representative Ijs in response to Vjs at [Mg2+]i = 3 mM for each variant were recorded (Figure 7). Gj,ss − Vj plots were constructed for each variant GJ under 3 mM [Mg2+]i (filled circles) and the data were well fitted by Boltzmann equations with little difference from those obtained without added Mg2+ (open circles and dashed lines in Figure 7). Individual Boltzmann parameters (Gmin, V0, and A) of the Vj gating for each variant showed no significant change comparing with those obtained from no added Mg2+ (Table 1).

[Mg2+]i showed little influence on the Vj-gating properties of Cx50 and individual variant GJs.

Figure 7.
[Mg2+]i showed little influence on the Vj-gating properties of Cx50 and individual variant GJs.

Representative Ijs for Cx50 or variant GJs at 3 mM [Mg2+]i in response to the Vjs (top). Normalized Gj,ss was plotted against Vj to generate Gj,ss − Vj plots and fitted with Boltzmann equations. Cx50 and these variant GJs under 3 mM [Mg2+]i showed a similar Boltzmann fitting curve (solid smooth lines) as those obtained without intracellular Mg2+ (dashed smooth lines).

Figure 7.
[Mg2+]i showed little influence on the Vj-gating properties of Cx50 and individual variant GJs.

Representative Ijs for Cx50 or variant GJs at 3 mM [Mg2+]i in response to the Vjs (top). Normalized Gj,ss was plotted against Vj to generate Gj,ss − Vj plots and fitted with Boltzmann equations. Cx50 and these variant GJs under 3 mM [Mg2+]i showed a similar Boltzmann fitting curve (solid smooth lines) as those obtained without intracellular Mg2+ (dashed smooth lines).

Homology models and pore electrostatic potentials of Cx50, G8E, G46E, V53E, and G8G46V53

To characterize the electrostatic changes in the channel due to the residue substitutions, we developed a homology model of Cx50 (Figure 8A) based on Cx26 GJ structure [22]. A sequence identity of 49% between Cx50 and Cx26 argues that Cx50 is likely to have a similar structure as Cx26 GJ. The homology structural model of Cx50 predicted that the glutamate substitutions at positions G8, G46, and V53 in Cx50 could reduce the local pore diameter at the respective positions (Figure 8A). In addition, we also generated pore surface/center electrostatic potentials (Figure 8) for each model. In comparison with Cx50 GJ, the glutamate substitution of each residue (G8E, G46E, or V53E) predicted an increase in local negative surface/center potentials near the residue location (Figure 8B,C). The triple-variant model showed elevated negative surface/center potentials in three residue sites, and therefore had the highest negative surface/center potential among the Cx50 variant GJs (Figure 8B,C). Since Cx50 GJ is a cation-preferring channel, an increase in negative electrostatic surface potential by these individual and combined variants could facilitate ion permeation by increasing local cation concentrations [22,24]. In descending ranking order, these variant GJs according to their slope γj, G8G46V53 > G46E > G8E > V53E = Cx50, suggest that negative electrostatic potential near the G46 and G8 positions (the narrowest portion and near the pore entrance, respectively) showed a much stronger effect on ion permeation than that of V53 position (middle of the GJ pore).

Homology models and pore electrostatic potentials in G8E, G46E, V53E, and G8G46V53 in comparison with Cx50.

Figure 8.
Homology models and pore electrostatic potentials in G8E, G46E, V53E, and G8G46V53 in comparison with Cx50.

(A) A side view of Cx50 homology structural model (cartoon view in green) superimposed with a cut open electrostatic surface potential model (left panel top) and calculated pore center electrostatic potential (left panel bottom). Enlarged portion of pore-lining domains in Cx50 and the triple variant are shown (right panels). Side chains of variant residues with estimated pore diameters are represented as spheres before (Cx50) and after mutation (G8G46V53). (B) Side view of cut open Cx50 and variant GJ channels shows pore surface electrostatic potentials (calculated with an adaptive Poisson–Boltzmann solver) using dielectric constants of two (protein) and 80 (solutions) [29]. Surface electrostatic potentials range from −40 (red) to +40 (blue). (C) Pore center potentials of each variant are superimposed onto that of Cx50 GJ (gray lines). Clearly, variants with the highest increase in γj showed a much higher negative potential in the pore center.

Figure 8.
Homology models and pore electrostatic potentials in G8E, G46E, V53E, and G8G46V53 in comparison with Cx50.

(A) A side view of Cx50 homology structural model (cartoon view in green) superimposed with a cut open electrostatic surface potential model (left panel top) and calculated pore center electrostatic potential (left panel bottom). Enlarged portion of pore-lining domains in Cx50 and the triple variant are shown (right panels). Side chains of variant residues with estimated pore diameters are represented as spheres before (Cx50) and after mutation (G8G46V53). (B) Side view of cut open Cx50 and variant GJ channels shows pore surface electrostatic potentials (calculated with an adaptive Poisson–Boltzmann solver) using dielectric constants of two (protein) and 80 (solutions) [29]. Surface electrostatic potentials range from −40 (red) to +40 (blue). (C) Pore center potentials of each variant are superimposed onto that of Cx50 GJ (gray lines). Clearly, variants with the highest increase in γj showed a much higher negative potential in the pore center.

Discussion

In the present study, we investigated the GJ channel properties of a variant with triple glutamate substitutions in the 8th, 46th, and 53rd positions in Cx50 (G8EG46EV53E). The triple-variant GJ channels displayed drastically increased single channel conductance (γj = 329 pS), ∼50% higher than that of Cx50, and reached a level similar to the largest known Cx37 GJ γj [7,8,36]. Our homology structural models predict that the substituted residues in the triple variant are likely located near the inner surface of the pore and could alter the pore diameters and electrostatic properties at different positions. Specifically, replacing small, nonpolar, and hydrophobic residues (glycine or valine) with large, hydrophilic, negatively charged glutamates would be expected to reduce the pore size and increase negative electrostatic potentials of the pore in their respective positions. The fact that γj was substantially elevated in the triple-variant GJ channel suggests that increased negativity of electrostatic potentials are likely to play a dominant role in facilitating ion permeation of the GJ channel. Investigating individual variants revealed that G8E and G46E, but not V53E, GJs were also able to increase γj, suggesting that these two positions in Cx50 are in critical positions to affect Cx50 GJ γj. However, minimum changes in Vj gating were observed in the GJs of triple and individual variants, suggesting that a substitution into glutamate at these residues does not likely alter the Vj sensor and Vj gate in Cx50 GJ. Additionally, we have provided experimental evidence that elevated [Mg2+]i decreases the γjs of Cx50 and the triple variant. Results from GJs formed by individual variants suggest that position G46 in Cx50 may play an important role in the γj changes in different [Mg2+]i. Our results from residues in NT and E1 domains of Cx50 GJ are consistent with a model where these residues either directly line or indirectly affect the pore to alter the rate of ion permeation as well as the modulation by [Mg2+]i. Our study combined single channel recording and molecular modeling approaches to reveal that pore electrostatic properties and [Mg2+]i are important factors in modulating ion permeation at individual Cx50 GJ channels.

Structural determinants of γj in Cx50

Previous studies showed several characteristic properties in Cx50 GJ channels, including prominent deactivation in macroscopic junctional currents (Ij) with increasing Vjs (strong Vj gating), a higher permeability to cations than anions, and the second highest γj (200–220 pS) of all characterized homotypic GJs [19,22,24,37,38]. Molecular mechanisms underlying these characteristics, particularly γj and Vj gating, are not fully understood. Previous studies using exchanged domains of Cx50 with Cx36 and site-directed mutagenesis have demonstrated that the NT and M1–E1 border domains and residues in these domains in Cx50 influence Vj gating and γj [1719,22,39]. According to the crystal structure of Cx26 and the homology model of Cx50 GJ, these domains are likely pore-lining and thus may be crucial in the involvement in the ion permeation along the channel lumen [22,27]. Single-point variants in these domains are predicted to alter pore diameter and electrostatic properties along the pore in the Cx50 GJ and thereby alter the rate of ion permeation [18,20,22]. The combination of experimental evidence and homology structure models suggests that electrostatic properties of pore-lining residues are likely to be the most critical parameter in determining Cx50 GJ γj [20,22].

The effect of mutating any single residue in one connexin is amplified six times in a hemichannel and 12 times in a GJ channel. Therefore, a triple variant like G8EG46EV53E would create three additional negatively charged glutamate rings (each ring with six glutamate residues) in one hemichannel and six rings in a GJ channel. According to our homology model, these variants are likely pore-lining residues, which substantially increases the negative pore surface and center electrostatic potentials (Figure 8). Our experimental data indicate that the triple variant GJ increases Cx50 GJ γj by nearly 50%, similarly or slightly higher to the highest γj attributed to Cx37 GJs [7,8,36]. This poses the question of whether there is an upper limit to ion permeation to any GJ channel. Since both Cx50 and Cx37 form cation-preferring GJ channels [22,24], increasing negative charges in the pore may facilitate a higher concentration of local permeating cations, thereby increasing γj. Our data and the above interpretation parallel studies in high-conducting, Ca2+ and voltage-activated K+ (BK) channels as well as nicotinic acetylecholine receptor (AChR) channels as both channels have been shown that anionic rings of glutamic acids in pore-lining domains are likely to facilitate a higher rate of ion permeation [4043]. We postulate that the differences in γj between Cx50 and Cx37 GJs, and potentially other cation-preferring GJs, are likely due to the differences in electrostatic profile in the channel lumen. Though the Cx50 triple-variant and Cx37 GJs with similar γjs, they are different from each other, especially at the 46th position, where is a glutamate in the triple variant and a glycine in Cx37.

An apparent minor difference was observed for the γj of G8E in the present study and our previous study [18]. The most likely reason is the difference in the ICS in our previous study, which contained 3 mM Mg2+ in the form of MgATP (together with 2 mM Na2ATP), while our current study we used either Mg2+-free ICS or ICS containing 3 mM MgCl2 without any added ATP. We deliberately removed ATP in patch pipette solution in our current study as ATP has been reported to be capable of chelating Mg2+, making actual free concentrations of Mg2+ hard to predict [44,45]. In addition, ATP is not very stable at room temperature and can be hydrolyzed into ADP/AMP. The fact that our measured γjs of Cx50 (192 ± 2 pS) and G8E (218 ± 4 pS) in 1 mM MgCl2 were much closer to those reported in our previous study (Cx50 205 ± 5 pS, G8E 222 ± 4 pS) [18] also supports this prediction. Of course, that we cannot rule out other small differences in patch pipette solution pH and osmolarity, patch pipette seal, whole-cell access resistance, room temperature, and different handling procedures may also partially contribute this difference.

The role of Cx50 variant charge location and their contingent interactions on γj

The Cx50 variants we studied are in different parts of the GJ pore. The homology model predicted that G8E was located at the beginning part of the pore entrance/exit positions (Figure 8) with relatively wide pore diameter and therefore produced only a moderate negative shift in the center electrostatic potentials and its GJ showed an increased γj. G46E is also predicted to locate at the pore entrance/exit, but at a narrower position and therefore produced a stronger negative shift to the center electrostatic potentials and showed the biggest increase in γj among these three individual variants, while V53E is located in the middle of the GJ pore lumen. This variant produced a localized negative shift of center electrostatic potentials, but did not change the γj, indicating that increase in negative electrostatic potential in this part of the GJ pore alone had minimum influence in the ion permeation.

These results can be rationalized if one views the Cx50 GJ channel permeation passage as a chain of mirror symmetrical resistors connected in series. The G8, G46, and V53 are important residues partially responsible for the local resistance of their respective positions. Mutations of each of these residues to glutamate would be predicted to increase negative electrostatic potential locally, which in turn would increase local cation concentration. Depending on the residue position, side chain orientation, and the local pore diameter change before/after the mutations, these three mutants showed different abilities to increase γjs, with the following rank: G46E > G8E > V53E = Cx50 GJ, indicating that the local resistance at these positions are likely following the same ranking with the highest at the G46 position followed by the G8 position, and the least at the V53 position. This model is consistent with our experimental results that G46E produced the highest increase in γj and the V53E the least change in γj on Cx50 GJ. However, the double-mutant G8EG46E GJ γj is significantly lower than that of the triple mutant (G8G46V53), indicating that when the local resistance near the 8th and 46th positions substantially reduced (G8EG46E), an additional reduction in local resistance at the 53rd position (V53E) will further increase the γj. This is a simple qualitative interpretation of our results based on our homology models and a contingent series resistor model.

Intracellular magnesium modulation on GJs

Here, we demonstrated that increasing [Mg2+]i dose-dependently reduced γj of Cx50 GJ channel and a more pronounced reduction was observed in the triple variant (G8G46V53) GJ (Figure 5). The molecular mechanism of how [Mg2+]i affects γj is not clear. However, the effect of increase in [Mg2+]i is consistent with a role for the negative electrostatic surface potential in facilitating γj. Intracellular Mg2+, with its extremely high charge density, is expected to interact with negatively charged pore-lining residues. Binding of Mg2+ by pore-lining residues could directly slow down ion permeation due to less mobility of Mg2+ along the pore, as well as a change in the local electrostatic profile in the pore, which could form an energy barrier to reduce ion permeation especially for the cation preferred Cx50 (or its variants) GJs [22,24]. A recently resolved Ca2+-bound crystal structure of Cx26 GJ revealed that Ca2+ binds to a site in the Cx26 GJ pore enriched with negatively charged residues, including E42, E47, and carbonyl group of G45 [46]. Two of these corresponding residues in Cx50 are conserved and could function to bind Mg2+ to reduce GJ permeation. In addition, the single or triple Cx50 variants introduced one or more negatively charged glutamate residue(s), which would facilitate more interactions between Mg2+ and their respective GJ pore, leading to elevated sensitivity to Mg2+ modulation on γj, as was observed.

Conclusion

Our study demonstrates that collectively increasing negative charges at the 8th, 46th, and 53rd positions through a triple glutamate substitution substantially increases Cx50 GJ γj without much change in Vj-gating properties. Our homology structure models suggest that the increased γj was likely due to an increase in negative electrostatic potential in the pore to enhance permeation of cations. Furthermore, elevated [Mg2+]i decreased γj of Cx50 and the triple variant in a dose-dependent manner without much change in their Vj-gating properties.

Abbreviations

     
  • Ca2+

    calcium

  •  
  • CT

    carboxyl-terminus

  •  
  • Cx50

    connexin50

  •  
  • DMEM

    Dulbecco's Modified Eagle Medium

  •  
  • E1

    first extracellular domain

  •  
  • Gj,ss

    normalized steady-state junctional conductance

  •  
  • ICS

    intracellular solution

  •  
  • Ij

    macroscopic junctional current

  •  
  • ij

    single channel junctional current

  •  
  • Mg2+

    magnesium

  •  
  • NT

    amino-terminal

  •  
  • Vj

    transjunctional voltage

  •  
  • γj

    unitary channel conductance

Author Contribution

M.G.T., S.S., and N.K.K. designed and performed all experiments, analyzed data, and wrote an early draft of the manuscript. H.A. developed homology models and calculated pore surface potentials. B.H.S. calculated pore center potentials. D.B. designed the project, supervised data analysis, and critically revised the manuscript.

Funding

This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada [288241 to D.B. and 217494 to B.H.S.]. This work was also supported by Grants-in-Aid for Scientific Research [26440029 to H.A.] from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Acknowledgments

We thank Professor Tomitake Tsukihara for his generous help on developing homology models, technical help from Honghong Chen for generating all variant cDNA constructs, and some dual patch-clamp recordings from Artur Santos-Miranda.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Goodenough
,
D.A.
and
Paul
,
D.L.
(
2009
)
Gap junctions
.
Cold Spring Harb. Perspect. Biol.
1
,
a002576
2
Goldberg
,
G.S.
,
Lampe
,
P.D.
and
Nicholson
,
B.J.
(
1999
)
Selective transfer of endogenous metabolites through gap junctions composed of different connexins
.
Nat. Cell Biol.
1
,
457
459
3
Sáez
,
J.C.
,
Berthoud
,
V.M.
,
Brañes
,
M.C.
,
Martínez
,
A.D.
and
Beyer
,
E.C.
(
2003
)
Plasma membrane channels formed by connexins: their regulation and functions
.
Physiol. Rev.
83
,
1359
1400
4
Bai
,
D.
and
Wang
,
A.H.
(
2014
)
Extracellular domains play different roles in gap junction formation and docking compatibility
.
Biochem. J.
458
,
1
10
5
Sohl
,
G.
and
Willecke
,
K.
(
2004
)
Gap junctions and the connexin protein family
.
Cardiovasc. Res.
62
,
228
232
6
Kreuzberg
,
M.M.
,
Willecke
,
K.
and
Bukauskas
,
F.F.
(
2006
)
Connexin-mediated cardiac impulse propagation: connexin 30.2 slows atrioventricular conduction in mouse heart
.
Trends Cardiovasc. Med.
16
,
266
272
7
Veenstra
,
R.D.
,
Wang
,
H.Z.
,
Beyer
,
E.C.
,
Ramanan
,
S.V.
and
Brink
,
P.R.
(
1994
)
Connexin37 forms high conductance gap junction channels with subconductance state activity and selective dye and ionic permeabilities
.
Biophys. J.
66
,
1915
1928
8
Banach
,
K.
,
Ramanan
,
S.V.
and
Brink
,
P.R.
(
2000
)
The influence of surface charges on the conductance of the human connexin37 gap junction channel
.
Biophys. J.
78
,
752
760
9
Gonzalez
,
D.
,
Gomez-Hernandez
,
J.M.
and
Barrio
,
L.C.
(
2007
)
Molecular basis of voltage dependence of connexin channels: an integrative appraisal
.
Prog. Biophys. Mol. Biol.
94
,
66
106
10
Hille
,
B.
(
2001
)
Ion Channels of Excitable Membrane
,
Sinauer Associates, Inc.
,
Sunderland, MA
11
Oh
,
S.
,
Ri
,
Y.
,
Bennett
,
M.V.
,
Trexler
,
E.B.
,
Verselis
,
V.K.
and
Bargiello
,
T.A.
(
1997
)
Changes in permeability caused by connexin 32 mutations underlie X-linked Charcot-Marie-Tooth disease
.
Neuron
19
,
927
938
12
Gong
,
X.Q.
and
Nicholson
,
B.J.
(
2001
)
Size selectivity between gap junction channels composed of different connexins
.
Cell Commun. Adhes.
8
,
187
192
13
Weber
,
P.A.
,
Chang
,
H.C.
,
Spaeth
,
K.E.
,
Nitsche
,
J.M.
and
Nicholson
,
B.J.
(
2004
)
The permeability of gap junction channels to probes of different size is dependent on connexin composition and permeant-pore affinities
.
Biophys. J.
87
,
958
973
14
Veenstra
,
R.D.
,
Wang
,
H.Z.
,
Beblo
,
D.A.
,
Chilton
,
M.G.
,
Harris
,
A.L.
,
Beyer
,
E.C.
et al. 
(
1995
)
Selectivity of connexin-specific gap junctions does not correlate with channel conductance
.
Circ. Res.
77
,
1156
1165
15
Kronengold
,
J.
,
Trexler
,
E.B.
,
Bukauskas
,
F.F.
,
Bargiello
,
T.A.
and
Verselis
,
V.K.
(
2003
)
Single-channel SCAM identifies pore-lining residues in the first extracellular loop and first transmembrane domains of Cx46 hemichannels
.
J. Gen. Physiol.
122
,
389
405
16
Kronengold
,
J.
,
Srinivas
,
M.
and
Verselis
,
V.K.
(
2012
)
The N-terminal half of the connexin protein contains the core elements of the pore and voltage gates
.
J. Membr. Biol.
245
,
453
463
17
Xin
,
L.
and
Bai
,
D.
(
2013
)
Functional roles of the amino terminal domain in determining biophysical properties of Cx50 gap junction channels
.
Front. Physiol.
4
,
373
18
Xin
,
L.
,
Nakagawa
,
S.
,
Tsukihara
,
T.
and
Bai
,
D.
(
2012
)
Aspartic acid residue d3 critically determines cx50 gap junction channel transjunctional voltage-dependent gating and unitary conductance
.
Biophys. J.
102
,
1022
1031
19
Xin
,
L.
,
Gong
,
X.Q.
and
Bai
,
D.
(
2010
)
The role of amino terminus of mouse Cx50 in determining transjunctional voltage-dependent gating and unitary conductance
.
Biophys. J.
99
,
2077
2086
20
Tong
,
X.
,
Aoyama
,
H.
,
Sudhakar
,
S.
,
Chen
,
H.
,
Shilton
,
B.H.
and
Bai
,
D.
(
2015
)
The first extracellular domain plays an important role in unitary channel conductance of Cx50 gap junction channels
.
PLoS ONE
10
,
e0143876
21
Hu
,
X.G.
,
Ma
,
M.Y.
and
Dahl
,
G.
(
2006
)
Conductance of connexin hemichannels segregates with the first transmembrane segment
.
Biophys. J.
90
,
140
150
22
Tong
,
X.
,
Aoyama
,
H.
,
Tsukihara
,
T.
and
Bai
,
D.
(
2014
)
Charge at the 46th residue of connexin50 is crucial for the gap-junctional unitary conductance and transjunctional voltage-dependent gating
.
J. Physiol.
592
,
5187
5202
23
Srinivas
,
M.
,
Kronengold
,
J.
,
Bukauskas
,
F.F.
,
Bargiello
,
T.A.
and
Verselis
,
V.K.
(
2005
)
Correlative studies of gating in Cx46 and Cx50 hemichannels and gap junction channels
.
Biophys. J.
88
,
1725
1739
24
Srinivas
,
M.
,
Costa
,
M.
,
Gao
,
Y.
,
Fort
,
A.
,
Fishman
,
G.I.
and
Spray
,
D.C.
(
1999
)
Voltage dependence of macroscopic and unitary currents of gap junction channels formed by mouse connexin50 expressed in rat neuroblastoma cells
.
J. Physiol.
517
,
673
689
25
Palacios-Prado
,
N.
,
Hoge
,
G.
,
Marandykina
,
A.
,
Rimkute
,
L.
,
Chapuis
,
S.
,
Paulauskas
,
N.
et al. 
(
2013
)
Intracellular magnesium-dependent modulation of gap junction channels formed by neuronal connexin36
.
J. Neurosci.
33
,
4741
4753
26
Palacios-Prado
,
N.
,
Chapuis
,
S.
,
Panjkovich
,
A.
,
Fregeac
,
J.
,
Nagy
,
J.I.
and
Bukauskas
,
F.F.
(
2014
)
Molecular determinants of magnesium-dependent synaptic plasticity at electrical synapses formed by connexin36
.
Nat. Commun.
5
,
4667
27
Maeda
,
S.
,
Nakagawa
,
S.
,
Suga
,
M.
,
Yamashita
,
E.
,
Oshima
,
A.
,
Fujiyoshi
,
Y.
et al. 
(
2009
)
Structure of the connexin 26 gap junction channel at 3.5 Å resolution
.
Nature
458
,
597
602
28
Suga
,
M.
,
Maeda
,
S.
,
Nakagawa
,
S.
,
Yamashita
,
E.
and
Tsukihara
,
T.
(
2009
)
A description of the structural determination procedures of a gap junction channel at 3.5 Å resolution
.
Acta Crystallogr. D Biol. Crystallogr.
65
,
758
766
29
Baker
,
N.A.
,
Sept
,
D.
,
Joseph
,
S.
,
Holst
,
M.J.
and
McCammon
,
J.A.
(
2001
)
Electrostatics of nanosystems: application to microtubules and the ribosome
.
Proc. Natl Acad. Sci. U.S.A.
98
,
10037
10041
30
Pellegrini-Calace
,
M.
,
Maiwald
,
T.
and
Thornton
,
J.M.
(
2009
)
Porewalker: a novel tool for the identification and characterization of channels in transmembrane proteins from their three-dimensional structure
.
PLoS Comput. Biol.
5
,
e1000440
31
DeLano
,
W.L.
(
2006
)
The PyMOL Molecular Graphics System, Version 1.5.0.4
,
Shrodiner, LLC
,
New York
32
Bai
,
D.
,
del Corsso
,
C.
,
Srinivas
,
M.
and
Spray
,
D.C.
(
2006
)
Block of specific gap junction channel subtypes by 2-aminoethoxydiphenyl borate (2-APB)
.
J. Pharmacol. Exp. Ther.
319
,
1452
1458
33
Harris
,
A.L.
,
Spray
,
D.C.
and
Bennett
,
M.V.
(
1981
)
Kinetic properties of a voltage-dependent junctional conductance
.
J. Gen. Physiol.
77
,
95
117
34
Bai
,
D.
and
Cameron
,
J.
(
2016
)
Patch Clamp Analysis of gap Junction Channel Properties
,
CRC Press Taylor & Francis Group
,
London, U.K.
35
Wilders
,
R.
and
Jongsma
,
H.J.
(
1992
)
Limitations of the dual voltage clamp method in assaying conductance and kinetics of gap junction channels
.
Biophys. J.
63
,
942
953
36
Ramanan
,
S.V.
,
Brink
,
P.R.
,
Varadaraj
,
K.
,
Peterson
,
E.
,
Schirrmacher
,
K.
and
Banach
,
K.
(
1999
)
A three-state model for connexin37 gating kinetics
.
Biophys. J.
76
,
2520
2529
37
White
,
T.W.
,
Bruzzone
,
R.
,
Wolfram
,
S.
,
Paul
,
D.L.
and
Goodenough
,
D.A.
(
1994
)
Selective interactions among the multiple connexin proteins expressed in the vertebrate lens: the second extracellular domain is a determinant of compatibility between connexins
.
J. Cell Biol.
125
,
879
892
38
White
,
T.W.
,
Bruzzone
,
R.
,
Goodenough
,
D.A.
and
Paul
,
D.L.
(
1992
)
Mouse Cx50, a functional member of the connexin family of gap junction proteins, is the lens fiber protein MP70
.
Mol. Biol. Cell
3
,
711
720
39
Dong
,
L.
,
Liu
,
X.
,
Li
,
H.
,
Vertel
,
B.M.
and
Ebihara
,
L.
(
2006
)
Role of the N-terminus in permeability of chicken connexin45.6 gap junctional channels
.
J. Physiol.
576
,
787
799
40
Brelidze
,
T.I.
,
Niu
,
X.
and
Magleby
,
K.L.
(
2003
)
A ring of eight conserved negatively charged amino acids doubles the conductance of BK channels and prevents inward rectification
.
Proc. Natl Acad. Sci. U.S.A.
100
,
9017
9022
41
Imoto
,
K.
,
Busch
,
C.
,
Sakmann
,
B.
,
Mishina
,
M.
,
Konno
,
T.
,
Nakai
,
J.
et al. 
(
1988
)
Rings of negatively charged amino acids determine the acetylcholine receptor channel conductance
.
Nature
335
,
645
648
42
MacKinnon
,
R.
,
Latorre
,
R.
and
Miller
,
C.
(
1989
)
Role of surface electrostatics in the operation of a high-conductance Ca2+-activated K+ channel
.
Biochemistry
28
,
8092
8099
43
Wilson
,
G.G.
,
Pascual
,
J.M.
,
Brooijmans
,
N.
,
Murray
,
D.
and
Karlin
,
A.
(
2000
)
The intrinsic electrostatic potential and the intermediate ring of charge in the acetylcholine receptor channel
.
J. Gen. Physiol.
115
,
93
106
44
Grubbs
,
R.D.
(
2002
)
Intracellular magnesium and magnesium buffering
.
Biometals
15
,
251
259
45
Luthi
,
D.
,
Gunzel
,
D.
and
McGuigan
,
J.A.
(
1999
)
Mg-ATP binding: its modification by spermine, the relevance to cytosolic Mg2+ buffering, changes in the intracellular ionized Mg2+ concentration and the estimation of Mg2+ by 31P-NMR
.
Exp. Physiol.
84
,
231
252
PMID:
[PubMed]
46
Bennett
,
B.C.
,
Purdy
,
M.D.
,
Baker
,
K.A.
,
Acharya
,
C.
,
McIntire
,
W.E.
,
Stevens
,
R.C.
et al. 
(
2016
)
An electrostatic mechanism for Ca2+-mediated regulation of gap junction channels
.
Nat. Commun.
7
,
8770