Coupling asymmetry of heterotypic connexin 45/ connexin 43-EGFP gap junctions: Properties of fast and slow gating mechanisms

Abstract

Although fast and slow gating mechanisms have been described in gap junctions (GJs), their relative contributions to dependence on transjunctional voltage, V_j, is still unclear. We used cell lines expressing wild-type connexin 45 (Cx45) and connexin 43 fused with enhanced green fluorescent protein (Cx43-EGFP) to examine mechanisms of gating in homo- and heterotypic GJs formed of these connexins. Macroscopically Cx45/Cx45 channels show high sensitivity to V_j. Cx45 channels demonstrate two types of gating: fast transitions between open and residual states and slow transitions between open and completely closed states. Single-channel conductance of the Cx45 channel is ≈32 pS for the open state and ≈4 pS for the residual state. Cx45/Cx43-EGFP heterotypic junctions exhibit very asymmetrical V_j gating with the maximum junctional conductance shifted to V_j positive on the Cx45 side. Conductance of single Cx45/Cx43-EGFP channels is ≈55 pS for the open state and ≈4 pS for the residual state, values consistent with the simple-series connection of Cx45 and Cx43-EGFP hemichannels. At V_j = 0, the slow gate of many Cx45 hemichannels is closed in both homotypic Cx45/Cx45 and heterotypic Cx45/Cx43-EGFP junctions. Fast and slow V_j gates of both Cx45 and Cx43 hemichannels close for relative negativity at their cytoplasmic end. Coupling mediated by Cx45/Cx43-EGFP junctions can exhibit asymmetry that can be strongly modulated by small changes in difference of holding potentials. Cx45/Cx43 junctions are likely to be found in brain and heart and may mediate rectifying electrical transmission or modulatable chemical communication.

The connexins are a large family of homologous integral membrane proteins that form gap junction (GJ) channels which provide a pathway for electrical signaling and metabolic communication (1, 2). Each GJ channel is composed of two hemichannels, each formed of six connexin (Cx) monomers. GJ channels generally respond to transjunctional voltage, V_j, with decay of junctional conductance (g_j) to some steady state, g_j∞ (3). Although sensitivity and kinetics depend on the connexin type, most homotypic junctions show a maximum g_j at V_j = 0 and symmetric g_j–V_j dependence, which has been ascribed to the presence of a gate in each hemichannel and their opposite orientation with respect to V_j. In general, as V_j is increased, g_j seems to decline to a voltage-insensitive or residual conductance, g_jmin, that ranges from ≈5 to 30% of g_j at V_j = 0 mV, depending on the connexin. This property has been explained by single-channel studies showing that GJ channels close from the main open state with conductance, γ_open, to a subconducting or residual conductance state, γ_res (4–7). Now, apparently an additional gating mechanism exists in each GJ hemichannel characterized by slower kinetics and full open/closed transitions which themselves are slow, often taking tens of milliseconds to complete (8, 9). We termed the two V_j-sensitive gating mechanisms “fast” and “slow”, with the fast gating mechanism referring to that which closes channels to the residual state and the slow gating mechanism that which closes channels completely. Two distinct gating mechanisms have also been reported in Cx40 (10), Cx30 (11), and Cx57 (12) channels, as well as in unapposed Cx46 and chimeric Cx32*Cx43E1 hemichannels (8, 13). Fast and slow gates have been supported by molecular studies, which show that fast gating can be abolished with little effect on slow gating (9, 14).

Here we report on the properties of Cx45 GJ channels and heterotypic channels formed by pairing of Cx45 and Cx43-enhanced green fluorescent protein (EGFP) expressing cells. Cx45 channels have both fast and slow gating mechanisms, and both gates are closed by the same polarity of V_j, relative negativity at the cytoplasmic end of the channel. The two gates per hemichannel operate as if they are in series where the state of one gate affects the voltage across the other gate (“contingent gating”). In Cx45/Cx45 junctions at V_j = 0, a significant fraction of slow gates are closed, and the slow gate rather than the fast gate seems to be responsible for the strong V_j dependence. In heterotypic Cx43/Cx43-EGFP junctions, the g_j–V_j relation is very asymmetric, and this property can result in very marked directional asymmetry in electrotonic transmission across the junction. Moreover, small differences in resting potential between the coupled cells strongly modulate the degree of asymmetry, and electrical transmission across the junctions can be switched from bidirectional to unidirectional. Modulation of electrical and chemical communication across these junctions may have important functional consequences.

Materials and Methods

cDNAs, Cell Lines, Culture Conditions, and Transfections.

Experiments were performed on HeLa cells (ATCC no. CCL-2) stably transfected with cDNAs encoding rat Cx45 and rat Cx43-EGFP. Cx43-EGFP has the EGFP attached to its cytoplasmic carboxyl terminus, and its trafficking can be visualized in living cells (15). HeLa cells were grown in Dulbecco's medium supplemented with 10% FBS. To study homotypic junctions, cells of one type were seeded onto coverslips. To study Cx45/Cx43-EGFP heterotypic junctions, HeLaCx45 and HeLaCx43-EGFP cells were mixed together in equal quantities and seeded on coverslips.

Electrophysiological and Fluorescence Measurements.

For simultaneous electrophysiological and fluorescence recording, cells grown on coverslips were transferred to an experimental chamber mounted on the stage of an inverted Olympus IX-70 microscope equipped with OlymPix 2000 digital camera (Olympus, New Hyde Park, NY). Fluorescence of Cx43-EGFP was monitored by using UltraVIEW software (Perkin–Elmer Life Sciences). The chamber was perfused with a modified Krebs–Ringer solution [140 mM NaCl/4 mM KCl/2 mM CaCl₂/1 mM MgCl₂/5 mM Hepes/5 mM glucose/2 mM pyruvate (pH = 7.4)]. Patch pipettes were filled with a solution containing (in mM) NaCl (140), KCl (0.2), CaCl₂ (1), MgCl₂ (3), MgATP (5), Hepes (3, pH 7.2) EGTA (2, [Ca²⁺]_i = 5 × 10⁻⁸ M). Junctional conductance was measured by using a dual whole-cell patch clamp as described (9). For data acquisition and analysis we used a MIO-16X A/D converter (National Instruments, Austin TX) and our own software.

Results

Macroscopic Voltage Gating of Cx45 Channels.

Fig. 1A shows a record from a Cx45 cell pair in response to a long V_j step of −100 mV (see V_j trace; negative V_j is defined for negative voltage step). Small, brief V_j ramps (V₁ was changed linearly from −18 to 18 mV in 0.8 s with 0.2 s between ramps) were applied to measure g_j before the long V_j step and after g_j had reached steady state during the step. During the V_j step, g_j (top trace) decreased from ≈8 to 1.2 nS. After the −100-mV step was terminated, g_j increased, slowly reaching a maximum of ≈9 nS, i.e., exceeding the control value, and then subsided back to the initial level. The “overshoot” of g_j, g_jover, during recovery was greater for larger V_j and somewhat greater for positive than for negative V_j (Fig. 1C).

Figure 1.

Macroscopic measurement of V_j gating of homotypic Cx45 junctions. (A) V_j gating in response to a long V_j step of −100 mV applied to cell 1. Repeated (1 Hz) V_j ramps (from −18 to +18 mV in 0.8 s) were applied before, during, and after the V_j step to monitor g_j. The g_j time plot determined from I_j and V_j records shows that during recovery from the step, g_j transiently exceeded the initial value; the maximum was termed g_jover. (B) Pooled data of normalized steady state g_j (G_j) vs. V_j measured in 15 Cx45 cell pairs. G_j was measured at the end of V_j steps ≈30 s in duration. The solid line is a fit of all of the data points by a four-state contingent gating model containing one gate per hemichannel (see below) with sensitivity parameters A = 0.3 mV⁻¹ and V₀ = 8.9 mV for negative V_j and A = 0.17 mV⁻¹ and V₀ = 7.5 mV. (C) Pooled data of normalized g_jover vs. V_j.

The relation between normalized g_j at steady state (G_j∞) and V_j was determined in 15 Cx45 cell pairs (summarized in Fig. 1B). The value of g_j declined markedly for V_j of either sign, with slight asymmetry around V_j = 0 mV ascribable to dependence on V_m (16). g_j reached a minimum at ±60 mV and then increased at higher V_j values. The solid line is a fit to the data with a model in which each hemichannel is assumed to contain a single gate whose state is dependent on the state of the other gate (contingent gating) (3, 17). Details of this model and the secondary increase in g_j are discussed with respect to Figs. 3 and 6.

Figure 3.

V_j dependence of Cx45/Cx43-EGFP heterotypic junctions. (A) Phase contrast (Left) and fluorescence (Right) images of a Cx45/Cx43-EGFP cell pair. The Cx43-EGFP cell is identified by its fluorescence. The arrow indicates a junctional plaque. (B) Changes in I_j in response to positive and negative 60-mV V_j steps (positive V_j is defined as relatively positive on the Cx43-EGFP side). I_j increased slowly from an initial level of ≈13 pA during the negative V_j step and decreased rapidly during the positive V_j step. g_j decreased by about 30% between the steps. (C) Normalized g_j–V relation for Cx45/Cx43-EGFP heterotypic junctions in 14 cell pairs. The thin black line is a fit of a 4-state contingent gating model with one gate in each hemichannel to all data points (open and center-dotted circles). The parameters are A= 0.18 mV⁻¹ and V₀ = 1 mV for the Cx45 hemichannel and A = 0.03 mV⁻¹ and V₀ = 26 mV for the Cx43-EGFP hemichannel. The thick gray line is a fit of the data points indicated by dotted circles with a four-state contingent gating model of the channel containing fast and slow gates only in the Cx45 hemichannel. The parameters are A = 0.27 mV⁻¹ and V₀ = 3 mV for the slow gate and A = 0.46 mV⁻¹ and V₀ = 10 mV for the fast gate. Normalized g_j decreased rapidly with voltage for relative positivity on the Cx43-EGFP side; the relatively high sensitivity indicates that this decrease is mediated by the Cx45 hemichannel and that its gating polarity is negative. An expanded view (Inset) demonstrates that the contingent model can explain the secondary increase in g_j at large V_j (see Discussion and Fig. 6).

Figure 6.

Summary of the fit to the experimental data obtained in heterotypic Cx45/Cx43-EGFP junctions by a four-state contingent gating model. (A) V_j sensitivity of open probability of fast (dashed line) and slow (solid line) gates that operate only in the Cx45 hemichannel. (B) At V_j = 0 ≈70% of Cx45 hemichannels are closed solely by the slow gate (G_j normalized to value at V_j = 0 is shown by the thick gray line and P_o of the slow gate is shown by the solid line) and that the fast gate operates only at V_j values > ≈30 mV. (C and D) Heterotypic channels with identical (C) or different (D) conductances of hemichannels containing one slow gate (arrows) per hemichannel. (E) G_j–V_j dependence calculated for a four-state contingent gating model in which the two hemichannels have V_j sensitivity like that of Cx45 and Cx43 but are identical in unitary conductance that is equal to 230 pS (dashed line, C) or that they differ in both V_j sensitivity and unitary conductance (64 and 230 pS for the Cx45 and Cx43 hemichannels, respectively; solid line, D).

Fig. 1C shows the g_jover vs. V_j plot. g_jover was determined at the peak of the transient increase and normalized to g_j at V_j = 0 mV. At V_j = ±150, g_jover increased to ≈1.3. These results can be explained for the channels in which a substantial fraction of the hemichannels are closed at V_j = 0. V_j of one polarity tends to close gates in one hemichannel and open them in its partner hemichannel. Because at V_j = 0 no voltage occurs across either hemichannel, they might be expected to relax to their final open probability, P_o, independently. g_jover will be greater if the open hemichannels close more slowly than the closed hemichannels open.

Single Cx45 Channels Show Fast and Slow V_j Gating Mechanisms.

We examined V_j gating of single Cx45 channels in weakly coupled cell pairs with only one or two operational channels. Fig. 2A shows single-channel gating in response to V_j steps of −65 and +65 mV. At the onset of the voltage step one channel was open and underwent frequent unresolved closures until it closed to γ_res (dashed line); then several gating transitions of ≈1.7 pA (≈28 nS) occurred between the open (γ_open) and residual states. The expanded record (Inset; I_j record was low-pass filtered at 200 Hz and sampled with 2-ms intervals) shows that 28-pS transitions between open and residual states were fast (no more than one point was recorded during each transition), and transitions from the residual to the closed state and back were slow (see arrows). The open-state conductance was ≈32 pS, and the residual state conductance was ≈4 pS. When V_j was reversed, the channel was still closed and after about 0.2 s opened slowly to the residual state. It then opened fully, closed to a level lower than the residual state, opened to the residual state, and then had a burst of poorly resolved transitions near the fully open state. In other cell pairs I_j–V_j relations of single channels were examined by applying voltage ramps (from +70 to −70 mV for the first two ramps and from +100 to −100 mV for the second two ramps; Fig. 2B). The channel closed after application of the positive shoulder of ramps and then opened again as V_j decreased, and then closed again for large negative V_j. Closures occurred to the residual state during the first two ramps and to the fully closed state in the second two, as indicated by arrows. Open-channel current increased linearly with voltage yielding a slope conductance of ≈32 pS. The residual state also seemed ohmic (see second ramp), although the resolution was compromised by the low value of the conductance.

Figure 2.

V_j gating of single Cx45 channels. (A) A negative and then a positive step of 60 mV were applied. At the beginning of the negative step a single channel was open (solid line) that showed closure to a substate of ≈4 pS (dashed lines), as well as a burst of apparently partial closings that were not well resolved. The closures were fast compared with the sampling interval of 2 ms. (B) Successive ramps repeated at 1/s showed fast gating to the residual state (first two ramps, dashed lines) and slow gating to the fully closed state (second two ramps, arrows). The channel was typically open at the onset of the ramp, closed at short latency, and then reopened (solid lines) at V_j approaching near zero.

Macroscopic V_j Gating of Heterotypic Cx45/Cx43-EGFP Channels.

These experiments utilized Cx45/Cx43-EGFP cell pairs, which exhibited EGFP fluorescence only in one cell, and had at least one fluorescent plaque present at the apposition between the cells (Fig. 3A; the arrow indicates a junctional plaque). Fig. 3B illustrates I_j changes in response to V_j steps of ±60 mV. For relative negativity on the Cx43-EGFP side, I_j increased over several seconds from about 13 pA to 28 pA (from ≈6 to ≈15 channels). During a brief stop at 0 mV, g_j decreased by about 30%. During the following positive V_j step, I_j rapidly decreased to near zero; the residual current corresponded to g_j = ≈16 pS, indicating that most of the channels had closed completely and ≈4 channels had closed to γ_res. In simple series placement, γ_res in homotypic Cx45 and heterotypic Cx45/Cx43-EGFP junctions would be given by the series sum of the hemichannel conductance of the residual state, γ_hem,res, and that of the open state, γ_hem,open. Thus, γ_res = γ_hem,res n/(n + 1), where n is the ratio, γ_hem,open/γ_hem,res. Because n is substantially greater than unity for homotypic Cx45 and even greater for Cx45/Cx43-EGFP heterotypic channels closed to the residual state on the Cx45 side, then γ_res ≈ γ_hem,res, i.e., γ_hem,resCx45 should be ≈4 pS.

We studied g_j–V_j dependence in 14 Cx45/Cx43-EGFP cell pairs in which g_j ranged between 0.7 and 23 nS. The g_j vs. V_j plot normalized to g_j at V_j = 0 shows marked asymmetry (Fig. 3C). G_j decreases sharply with V_j values relatively positive on the Cx43-EGFP side followed by a small secondary increase at V_j values >40 mV. For the opposite polarity of V_j, G_j increases and decreases again at larger V_j values. Considering that Cx43-EGFP hemichannels only possess the slow V_j gate, the reduction in G_j on application of large negative V_j values, V_j positive to −40 mV (open circles in Fig. 3C) is attributable to closure of the slow gate in Cx43-EGFP hemichannels. The remaining conductance changes (indicated by dotted circles in Fig. 3C) are thus attributable to fast and slow gating of Cx45 hemichannels. We modeled the conductance changes by considering two series gates in Cx45 hemichannels that are in series with open Cx43-EGFP hemichannels. Assuming that γ_hem,open = ≈2γ_open, Cx45 and Cx43-EGFP hemichannels were assigned conductances of 64 pS and 220 pS, respectively. A conductance of 4 pS was assigned to a channel with a closed Cx45 fast gate, and zero conductance was assigned to a channel with a closed Cx45 slow gate. The resulting state diagram consists of a circular reaction scheme containing four states: both fast and slow Cx45 gates open, only the Cx45 fast gate closed, only the Cx45 slow gate closed, and both fast and slow Cx45 gates closed; the slow gate of the Cx43-EGFP hemichannel was assumed to always remain open. In contingent gating, the gates interact such that the state of one gate alters the open probability of the other gate by changing the V_j it senses.

A fit of the data from Cx45/Cx43-EGFP junctions by the contingent gating model suggests that the slow gate tends to close at smaller V_j values than the fast gate; that a substantial fraction of slow gates are closed at V_j = 0; and that most fast gates are open at V_j = 0 (see Fig. 6 in Discussion). As V_j is increased, G_j decreases to near zero as slow gates tend to close channels completely. When V_j is in a range that closes fast gates, their closure reduces V_j at the slow gates, which tends to open them, giving rise to the secondary increase in G_j. On the basis of the asymmetry of g_j–V_j dependence shown in Fig. 3C and gating asymmetry at the single channel level (see Fig. 4), we conclude that both fast and slow gates of Cx45 have negative gating polarities, i.e., close in response to V_j negative at their cytoplasmic ends. Our studies indicate that the slow gates in Cx45 hemichannels are largely responsible for closure of Cx45/Cx43-EGFP channels at V_j = near zero. Thus, we conclude that increase of G_j at negative V_j values is due to the slow gates opening in Cx45 hemichannels and the G_j decrease at higher negative V_j values is due to the slow gates closing in Cx43-EGFP hemichannels.

Figure 4.

Asymmetric V_j gating in heterotypic Cx43/Cx43-EGFP junctions at the single-channel level. (A) Example of opening of Cx45/Cx43-EGFP channels during a V_j step of −60 mV applied to the Cx43-EGFP cell. Opening and closing transitions were between open and closed states. During a subsequent V_j step of +60-mV channels closed completely at short latency. (B) During negative V_j steps, the single channel gated between the open state with a conductance of ≈55 pS and the fully closed state. During positive pulses, the channel gated to the residual state (first pulse) or to the closed state (second and third pulses). (C) Ramps at 1 Hz from +100 to −100 mV show closure from the open state (solid lines) to the residual or to the closed state. As V_j became less positive, the channel opened and remained open during the negative part of the ramp.

V_j Gating of Single Cx45/Cx43-EGFP Channels.

We studied single-channel conductance of Cx45/Cx43-EGFP heterotypic junctions in five weakly coupled cell pairs and in seven more strongly coupled cell pairs during recovery from uncoupling induced by CO₂. Fig. 4A shows I_j and V_j records in a cell pair with at least four functional channels. A V_j step negative with respect to the Cx43-EGFP cell showed single-channel activity at the start of the pulse with gating transitions of ≈55 pS between open and fully closed states and then full opening of three additional channels. The amplitude of the gating transitions was consistent with the prediction that γ_open of heterotypic channels results from the two hemichannels in series, i.e., γ_{Cx45/Cx43-EGFP,open} = 1/(1/230 + 1/64) = ≈50 pS. An oppositely oriented V_j step closed the channels so fast that it was difficult to distinguish their numbers and closing times. These data demonstrate that at V_j = 0, a substantial fraction (three fourths in this case) of Cx45/Cx43-EGFP channels were closed. We conclude that it is the slow gate of Cx45 that closes the channels, because gating transitions were between open and fully closed states, and Cx43-EGFP hemichannels have a high open probability at V_j = 0 (9). In the record in Fig. 4B, apparently one functional channel was open at V_j = 0 when the negative V_j steps were applied on the Cx43-EGFP side, and gating transitions were between open and closed states. In response to positive V_j steps, the channel closed to a residual state (first pulse, dashed line) or flickered with unresolved degrees of closure and then closed completely (second and third pulses, solid line). Similar results were obtained by using a voltage ramp protocol (Fig. 4C). For a ramp beginning at V_j = 100 mV, positive on the Cx43-EGFP side, the single active channel quickly closed to the residual and then to the closed state. As V_j became less positive and then negative the channel opened and typically remained open for the remainder of the ramp.

Asymmetry of Signal Transfer in Heterotypic Cx45/Cx43-EGFP Junctions.

In the next series of experiments, we examined if asymmetry of the g_j–V_j relation of Cx45/Cx43-EGFP junctions could give rise to directional asymmetry in cell–cell coupling. In seven Cx45/Cx43-EGFP cell pairs, we examined electrical cell–cell coupling by voltage clamping one cell and current clamping the other. Fig. 5 A and B illustrate the experiment in which the potential in the Cx43-EGFP cell (V₁) was clamped to repeated (3 Hz) positive or negative electrical pulses 30 ms in duration and 100 mV in amplitude from a holding potential (V_h1) of 0 mV. The amplitude of the electrotonic response in the Cx45 cell (V₂) was much greater for negative than for positive pulses. Because g_j is sensitive only to V_j, negative potentials in the Cx43-EGFP cell are equivalent to positive potentials in the Cx45 cell, and transmission of depolarization would be greater from the Cx45 cell to the Cx43-EGFP cell than in the opposite direction. Before the record in Fig. 5A, the holding current in the Cx45 cell was set to keep its holding potential, V_h2, at about +5 mV. During application of −100 mV pulses to cell 1, the potential in cell 2, V₂, reached about −60 mV and decayed back to near that between pulses; the coupling coefficient for the pulses, k_1–2,− = V₂/V₁, was ≈0.6. Because k_1-2 is given by g_j/(g_j + g₂), where g₂ is the conductance of cell 2, g_j was about twice g₂. When positive pulses were applied in cell 1, the initial response in cell 2 was about 20 mV, and later responses became smaller over several seconds to reach a steady-state peak-to-trough value of ≈10 mV superimposed on a steady potential of ≈8 mV resulting in k_1–2,+ = ≈0.1 for peak to trough and ≈0.18 from peak to baseline. The slower decay from the peak to a nonzero level of depolarization was caused by an increase in the coupling time constant, because relative negativity on the Cx45 side decreased g_j. The ratio of coupling coefficients for positive and negative V_j, k_1–2,+/k_1–2,−, which we defined as the electrical coupling asymmetry coefficient (K_asym), for peak to trough in V₂ was ≈0.15. When the holding current in cell 2 was reversed (after ≈20 s; in Fig. 5A), so that V_h2 in the steady state was about −8 mV, the first response in cell 2 was ≈20 mV and the following responses increased over the next several seconds so that the peak was ≈50 mV superimposed on the steady voltage. When the V₁ pulses were reversed to +100 mV, the peak-to-trough value of V₂ was rapidly decreased to about ≈3 mV, so that K_asym = 0.05. These data demonstrate that the difference between holding potentials of two cells, ΔV_h = V_h1 − V_h2, strongly affects asymmetry of electrical cell–cell coupling (coupling asymmetry is lower for smaller V₁ pulses). When the Cx45 cell was voltage clamped and pulsed and the Cx43-EGFP cell was current clamped, coupling was greater for positive pulses than for negative (K_asym was defined as k_1–2,−/k_1–2,+; not illustrated). Fig. 5C shows summarized data illustrating K_asym dependence on ΔV_h measured in seven Cx45/Cx43-EGFP cell pairs when either the Cx43-EGFP cell (solid circles) or the Cx45 cell (open circles) was pulsed. ΔV_h strongly affected electrical coupling asymmetry. K_asym increased to near 1 (almost no asymmetry) when the Cx45 cell was made more positive and decreased to near 0 (maximal asymmetry) when the Cx45 cell was made more negative.

Figure 5.

Asymmetry of electrical coupling mediated by heterotypic Cx45/Cx43-EGFP junctions. (A) Asymmetry of electrical coupling in a Cx45/Cx43-EGFP cell pair. The Cx43-EGFP cell (V₁) was voltage clamped to positive or negative 100 mV, 30-ms rectangular pulses at 3 Hz from a holding potential of 0 mV. The Cx45 cell (V₂) was current clamped, and the steady-state holding potential was varied from ≈5 mV (from 0 to 18 s) to about −8 mV (from 18 to 62 s). Coupling was much greater for the negative pulses, and coupling asymmetry increased by making the Cx45 side more negative. (B) Schematics of cell pair; pulses were applied to HelaCx43-EGFP cell. (C) Pooled data of the electrical coupling asymmetry coefficient, K_asym, vs. ΔV_jh measured in seven Cx45/Cx43-EGFP cell pairs. Filled and open circles correspond to the experiments in which the Cx43-EGFP or Cx45 cell was voltage clamped and pulsed, respectively. Solid line shows data fit to sigmoid function, K_asym = A/{1 + exp[b(ΔV_jh − ΔV_jho)]}, with parameters A = 0.9338, b = 0.24 mV⁻¹, ΔV_jho = −6.4 mV.

Discussion

Voltage-gating properties of Cx45 junctions have been examined in transfected cell lines and Xenopus oocytes, and these junctions have been shown to be one of the more sensitive to V_j and also to have sensitivity to V_m (16, 18, 19). Until recently, V_j sensitivity in all connexin channels was assumed to occur by a single mechanism. Complex kinetics of changes in g_j in response to V_j and the observation that V_j closes some GJ channels completely suggested that there is more than one V_j-dependent gating mechanism (20–22). In correlating macroscopic and single-channel studies of Cx43 channels, we demonstrated that the distinct fast and slow kinetic components evident in macroscopic recordings segregated with transitions to γ_res and γ_closed (9). Mutational studies demonstrated that the fast mechanism is governed by charged residues in the N-terminal domain (13, 23, 24). Moreover, C-terminal modifications of Cx43 including deletion (19) or attachment of aequorin or EGFP (25, 26), selectively abolish the fast gating component and transitions to γ_res (26). In summary, the existing data indicate that fast and slow gates are molecularly distinguishable and may well be common to all connexins.

Here we demonstrated that Cx45 channels possess both gating mechanisms and that the slow V_j gating mechanism is more sensitive to V_j than in Cx43 channels. In Cx43 junctions neither mechanism of V_j gating would have much influence on cell–cell coupling, except at large V_j values (> ≈60 mV). In Cx45 junctions, small V_j values can cause substantial changes in cell–cell coupling, and slow V_j gating makes a major contribution to these changes. Heterotypic Cx45/Cx43-EGFP junctions exhibit asymmetric V_j gating similar to that reported for Cx45/Cx43 junctions (18, 19). Only complete closures underlie the weak V_j dependence at large negative V_j values on the Cx43-EGFP side in Cx45/Cx43-EGFP junctions, consistent with closure of the slow V_j gate in Cx43-EGFP hemichannels on relative negativity. Thus, we could deduce that both the fast and slow V_j gates of Cx45 close on relative negativity on the Cx45 side. Complete opening and closings underlie the V_j dependence about V_j = 0 mV indicating that closure of the slow gate of Cx45 is responsible for the submaximal conductance at V_j = 0 mV. Opening of the slow gate of Cx45 at modest V_j values negative on the Cx43-EGFP side gives an increase in g_j that peaks and declines at larger V_j values as the slow gate of Cx43-EGFP begins to close.

Because of the substantial fraction of channels closed by the slow gate of Cx45 at V_j = 0 mV, one might have predicted that in Cx45/Cx45 junctions, a still higher fraction of channels would be closed by this gate in response to V_j leaving little or no residual conductance (g_min). Nonetheless, we and others observed a residual conductance at large V_j values. At the single-channel level, we found γ_open and γ_res of homotypic Cx45 channels to be ≈32 pS and ≈4 pS, respectively, in agreement with ref. 19, and in the absence of a slow gate, g_min should approximate the ratio γ_res/γ_open = 4/32 = 0.13. However, at V_j = ≈60 mV, g_j decreased to a value lower than predicted and then increased with increasing V_j up to about ±120 mV. The conductance increase at high V_j can be explained by having in Cx45 hemichannels fast and slow gates in series as described for Cx43 hemichannels (9). Macroscopic and single-channel data indicate that the slow gates of Cx45 hemichannels dominate at small V_j values. Higher V_j values should activate the fast gates, which have faster kinetics. This mechanism would tend to close preferentially channels to γ_res resulting in a large fraction of the V_j gradient to drop across the fast gate and near collapse of the V_j gradient at the sensor for the slow V_j gate. Because Cx43-EGFP hemichannels only contain the slow gate, which is weakly sensitive to V_j, we reasoned that Cx45/Cx43-EGFP channels provide a V_j range in which gating of Cx45 hemichannels can be examined in the absence of gating in the partner hemichannel. With data from Cx45/Cx43-EGFP junctions throughout the V_j range of −30 to −120 mV, a fit of the data by a contingent gating model showed that the presence of a residual conductance can be explained by interacting fast and slow gates (gray line in Fig. 3C). The fitted Boltzmann relations for each of the gates, if they acted alone, are plotted in Fig. 6A and show that the slow gate is more sensitive to V_j than the fast gate with a P_o at V_j = 0 of ≈0.7. However, the fast gate has a steeper dependence on V_j. In contingent gating, P_o of the slow gate decreases with V_j and then increases, because V_j is large enough to close the fast gate (Fig. 6B). This interaction produces a substantial fraction of channels residing in γ_res at larger V_j values. A plot of the predicted conductance normalized to that at V_j = 0 mV shows that the conductance changes at smaller V_j values closely follow P_o of the slow gate. At higher V_j, conductance deviates from P_o of the slow gate as fast gates close. Although the data can be explained by having two distinct gates that interact through their effects on the electric field inside the channel, we cannot rule out interactions that may occur through an allosteric mechanism or through sharing some molecular components.

A series arrangement of gap junction hemichannels, which can give rise to contingent gating, can also lead to differences in V_j sensitivity of the hemichannels in heterotypic channels as compared to that in homotypic channels. Consider a heterotypic channel formed of two series hemichannels with different open state conductances but with the same sensitivity to the voltage across them as they have in the homotypic channels. For V_j values where the open probability of the higher conductance hemichannels is large, the V_j sensitivity of the lower conductance hemichannel will be increased, because a larger fraction of V_j will appear across it. Conversely, when the lower conductance hemichannel is open, the V_j sensitivity of the higher conductance channel will be decreased. These effects are illustrated in Fig. 6E for a hypothetical heterotypic junction. The solid and dashed lines are G_j vs. V_j curves generated by a contingent gating model in which a single gate is assigned to each hemichannel. The gates on either side differ in V_j sensitivity giving rise to asymmetry in the G_j–V_j relation; the parameters used were those describing V_j dependence of Cx45 and Cx43-EGFP homotypic junctions. The dashed line (Fig. 6E) is generated when both hemichannels have identical conductances, illustrated schematically as hemichannels having the same pore diameter (Fig. 6C). With hemichannel conductances differing in the same ratio as those of Cx45 and Cx43-EGFP, but with the same sensitivites to the voltage developed across each hemichannel (Fig. 6D), V_j dependence in the hemichannel with the larger conductance is reduced, because a greater fraction of V_j drops across the smaller conductance hemichannel (solid line, Fig. 6E). Conversely, V_j dependence in the hemichannel with the smaller conductance is increased. At V_j = 0, series conductance has no effect on gating, and the open probabilities are the same in each case.

Elenes et al. (19) did not observe fast gating of the Cx43 hemichannel in Cx43/Cx45 junctions and concluded that docking of Cx43 with Cx45 blocks fast V_j gating in the Cx43 hemichannels. However, because the conductance of the Cx45 hemichannel is considerably smaller than that of the Cx43 hemichannel, ≈75% of V_j should act on the Cx45 hemichannel and ≈25% on the Cx43 hemichannel. The V_j sensitivity of the Cx43 hemichannel would be reduced relative to that in Cx43/Cx43 junctions, which could be interpreted as loss of fast V_j gating on the Cx43 side.

Cx45/Cx43-EGFP heterotypic junctions demonstrate marked asymmetry of electrical coupling for large values of V_j, largely because of the slow gating mechanism of Cx45 (Fig. 5). Moreover, the slow gate can operate relatively rapidly in response to V_j values comparable to amplitudes of action potentials (<0.1 s; see Figs. 3B and 4 A and B). Small variations of ΔV_h of ±10 mV around zero very effectively modulate the degree of asymmetry, and at ΔV_jh = ≈5–15 mV positive on the Cx43-EGFP side, electrical coupling is almost unidirectional for repeated signals comparable in amplitude to action potentials. Thus, a synapse with Cx45 on the presynaptic side and Cx43 on the postsynaptic side could be highly rectifying. We also observed markedly asymmetric electrical signaling in Cx45/Cx47 heterotypic junctions (data not shown) and in Cx32/37 junctions (F.F.B. and R. Weingart, unpublished data). Transmission at synapses using this mechanism would differ from known electrical synapses that are involved in escape systems, where the rectification is very rapid (<0.1 ms; refs. 27–29), and presumably is caused by single-channel rectification as seen in Cx26/Cx32 junctions (30, 31) and Cx46 hemichannels (8). It is interesting that the postsynaptic neuron may be at a more positive potential resulting in ΔV_jh of ≈10 mV (32, 33).

In summary, asymmetric electrical signaling can arise in heterotypic junctions that demonstrate marked asymmetry in V_j gating, which can be highly increased if the hemichannel with higher V_j sensitivity has a smaller unitary conductance than the partner hemichannel. Cx45/Cx43 junctions are possible between oligodendrocytes and astrocytes (34), between Schwann cells and fibroblasts (which show asymmetric coupling like that in Fig. 3) (35), and in the conductive system of the heart and atrial and ventricular myocardium (36). Cardiac action potentials are long enough to alter coupling at the asymmetric junctions and may play a role in coupling asymmetry between cells of conductive and working myocardium. Astrocytes can be markedly depolarized by increases in extracellular potassium during electrical activity in the central nervous system. If oligodendrocytes were not depolarized to the same extent, their junctions with astrocytes would become lower in conductance and permeability. It appears that small changes in ΔV_jh can be highly effective in regulation of electrical and metabolic communication between different cells and in different organ systems.

Abbreviations

Cx

connexin

EGFP

enhanced green fluorescent protein

GJ

gap junction