Cx43 Channel Gating and Permeation: Multiple Phosphorylation-Dependent Roles of the Carboxyl Terminus

Abstract

Connexin 43 (Cx43), a gap junction protein seemingly fit to support cardiac impulse propagation and synchronic contraction, is phosphorylated in normoxia by casein kinase 1 (CK1). However, during cardiac ischemia or pressure overload hypertrophy, this phosphorylation fades, Cx43 abundance decreases at intercalated disks and increases at myocytes' lateral borders, and the risk of arrhythmia rises. Studies in wild-type and transgenic mice indicate that enhanced CK1-phosphorylation of Cx43 protects from arrhythmia, while dephosphorylation precedes arrhythmia vulnerability. The mechanistic bases of these Cx43 (de)phosphoform-linked cardiac phenotypes are unknown. We used patch-clamp and dye injection techniques to study the channel function (gating, permeability) of Cx43 mutants wherein CK1-targeted serines were replaced by aspartate (Cx43-CK1-D) or alanine (Cx43-CK1-A) to emulate phosphorylation and dephosphorylation, respectively. Cx43-CK1-D, but not Cx43-CK1-A, displayed high Voltage-sensitivity and variable permselectivity. Both mutants showed multiple channel open states with overall increased conductivity, resistance to acidification-induced junctional uncoupling, and hemichannel openings in normal external calcium. Modest differences in the mutant channels' function and regulation imply the involvement of dissimilar structural conformations of the interacting domains of Cx43 in electrical and chemical gating that may contribute to the divergent phenotypes of CK1-(de)phospho-mimicking Cx43 transgenic mice and that may bear significance in arrhythmogenesis.

Keywords: arrhythmia; casein kinase 1; channel gating; gap junction permeability; phosphorylation.

Figure 1

Cx43-CK1-D displays stronger V_j-sensitivity than Cx43-CK1-A. (A) Five individual I_j responses to V_j = ±80 mV pulses (left) and sum (right) of 38 similar traces from Cx43-CK1-A cell pairs; (B) Five individual I_j responses to V_j = ±80 mV pulses (left) and sum (right) of 30 similar traces from Cx43-CK1-D cell pairs. For (A,B), tau values of I_j inactivation (2nd order exponential decays) are shown and the fits (white) displayed over the corresponding sum traces; (C) V_j-dependence of G_j from individual experiments in Cx43-CK1-A (g_j = 2.8 ± 1.0; n = 5) cell pairs; (D) V_j-dependence of G_j from individual experiments in Cx43-CK1-D (g_j = 1.7 ± 1.1; n = 7) cell pairs. For (C,D), g_j from each experiment was normalized as described in the Methods and the Boltzmann fits are shown in solid black lines; (E) Average V_j-dependence for CK1-A (gray circles) and CK1-D (black triangles) and their corresponding Boltzmann fits (dashed lines). Fast inactivation and V₀ values were different between Cx43-CK1-D and Cx43-CK1-A. For fitting parameters, see Table 1, Tables S1 and S2.

Figure 2

Cx43-CK1-A and Cx43-CK1-D display fully open, V_j-sensitive gap junction channels. (A,B,D,E) Illustrative traces of channel activity from Cx43-CK1-A (A,B) and Cx43-CK1-D (D,E) expressing cell pairs, at 40 mV (A,D) and 80 mV (B,E) transjunctional gradients. For all traces: zero current (long-dashed line) and the most evident I_j levels (short-dashed lines) are indicated; when present, downward arrows mark the beginning (black) and end (gray) of pulses; plots at right are the all-points histogram for the displayed record segment, showing the fraction of time at each I_j level; numbers indicate the conductance change between current levels. Notice that channel transitions often occur between the identified levels. (C,F) Average transition amplitude histograms at 40 and 80 mV V_j values from Cx43-CK1-A (C) and Cx43-CK1-D (F). Peak fits indicated by solid black lines. Likely transitions between channel states (see main text for further explanation) are indicated by double arrowed vertical lines. Transition amplitude distributions of Cx43-CK1-D and Cx43-CK1-A differed from each other and from Cx43WT, at both Vj values of 40 and 80 mV. However, at V_j = 40 mV, both mutants displayed transitions amplitudes compatible with O↔C and O↔R transitions (if O = 150 pS and R = 30 pS). Transitions larger than 150 pS were documented for Cx43-CK1-D. At V_j = 80 mV, O-R and R-C transitions were more evident for both mutants. However, at both 40 and 80 mV, transitions between closed and levels smaller than fully open states were observed, suggesting substates (S). For each group, the number of experiments (n) and measured transitions (N) were respectively, as follows: For CK1-A, 6 and 1867 at 40 mV, 5 and 1080 at 80 mV. For CK1-D, 4 and 1369 at 40 mV, 6 and 1032 at 80 mV.

Figure 3

Permselectivity of dephospho-mimicking Cx43-CK1-A GJs is lower and less variable than that of phospho-mimicking Cx43-CK1-D GJs. (A) Rate constant of transjunctional dye diffusion vs. junctional conductance for the indicated mutants. Each symbol represents a single experiment; (B) Distribution (box plots) of collected permselectivity values for Cx43-CK1-D (0.074 ± 0.022; n = 16) and Cx43-CK1-A (0.017 ± 0.001; n = 7); Cx43WT data (from [14]) shown in light gray for comparison. Note that permselectivity values of Cx43-CK1-D (and Cx43WT) do not display a normal distribution. ** Median and variance of Cx43-CK1-A are different from Cx43-CK1-D and Cx43-WT (p < 0.05). See text for further explanation.

Figure 4

Cx43-CK1-D and Cx43-CK1-A mutants form gap junctions resistant to closure by low pH. Superfusion (starting at time 0) with a bicarbonate solution buffered at pH = 6.0–6.2 caused slow g_j decrease in Cx43-CK1-D (black triangles; n = 6), and no g_j decrease in Cx43-CK1-A (gray circles; n = 4) during an observation period of ≥ 20 min. The response of Cx43WT junctions to similar treatment (white triangles; n = 5) is reproduced from [14] for comparison.

Figure 5

Cx43-CK1-D and Cx43-CK1-A expressing cells display frequent connexin HCh activity. Plasma membrane channel transitions compatible with opening of undocked connexons from single cells expressing Cx43-CK1-D (A–C) or Cx43-CK1-A (D–F). Lines, marks, plots, and numbers as in Figure 2. (A,D) are 20 s samples of longer recordings; (B,E) are expanded displays of the 2 s interval marked by black lines at the bottom left of (A,C); (C,F) Further examples of transitions recorded during 5-s pulses. Hemichannel activity from Cx43-CK1-D is usually well defined (as in A–C), in contrast to Cx43-CK1-A (D–F).

Figure 6

Residue S368 of Cx43 is phosphorylated in Rin cells expressing Cx43-WT, Cx43-CK1-A or Cx43-CK1-D. (A,B,D,E,G,H) Fluorescent (upper) and fluorescent merged with the corresponding differential interference contrast (DIC, lower) images of cells stained with a Cx43-pS368 phospho-specific antibody (pS368). Cx43-pS368 (green) was found in junctional plaques in groups and isolated pairs of Cx43WT (A,B), Cx43-CK1-A (D,E) and Cx43-CK1-D (G,H) cells. (C,F,I) DIC and fluorescent images of cells stained simultaneously with a polyclonal Cx43 antibody (“total” Cx43, tCx43) and a phospho-specific antibody against Cx43-p^CK1 (pCK1). In all groups, total Cx43 (green) was found in junctional plaques and other cell areas (C,F,I, upper left panels). Bona fide CK1-phosphorylated Cx43 (red) was found only in Cx43WT cells (C, bottom right) colocalized with a membrane fraction of total Cx43 at junctional plaques (C, upper right, yellow). The pCK1 antibody labels a non-specific nuclear signal in the dephospho- (F) and phospho-mimicking (I) mutant expressing cells that is also found in parental Rin cells, devoid of connexins (Figure S5). Calibration bars (pink lines, 25 μm) apply to each column.