The unitary conductances and permeability sequences of the rat connexin40 (rCx40) gap junction channels to seven monovalent cations and anions were studied in rCx40-transfected neuroblastoma 2A (N2A) cell pairs using the dual whole cell recording technique. Chloride salt cation substitutions (115 mM principal salt) resulted in the following junctional maximal single channel current-voltage relationship slope conductances (γj in pS): CsCl (153), RbCl (148), KCl (142), NaCl (115), LiCl (86), TMACl (71), TEACl (63). Reversible block of the rCx40 channel was observed with TBA. Potassium anion salt γj are: Kglutamate (160), Kacetate (160), Kaspartate (158), KNO3 (157), KF (148), KCl (142), and KBr (132). Ion selectivity was verified by measuring reversal potentials for current in rCx40 gap junction channels with asymmetric salt solutions in the two electrodes and using the Goldman-Hodgkin-Katz equation to calculate relative permeabilities. The permeabilities relative to Li+ are: Cs+ (1.38), Rb+ (1.32), K+ (1.31), Na+ (1.16), TMA+ (0.53), TEA+ (0.45), TBA+ (0.03), Cl (0.19), glutamate (0.04), and NO3− (0.14), assuming that the monovalent anions permeate the channel by forming ion pairs with permeant monovalent cations within the pore thereby causing proportionate decreases in the channel conductance. This hypothesis can account for why the predicted increasing conductances with increasing ion mobilities in an essentially aqueous channel were not observed for anions in the rCx40 channel. The rCx40 effective channel radius is estimated to be 6.6 Å from a theoretical fit of the relationship of relative permeability and cation radius.

Introduction

Gap junction channels are formed by twelve connexin protein subunits, six in each cell membrane (Peracchia, 1973; Caspar et al., 1977; Makowski et al., 1977), which allow the cell-to-cell passage of ions and small molecules (Bennett and Spray, 1965; Loewenstein, 1981; Hertzberg and Johnson, 1988; Beyer, 1993). Connexins contain highly conserved transmembrane and extracellular domains, but unique cytoplasmic regions (Beyer and Veenstra, 1994). To date at least 16 different connexin channels have been identified which are classified according to their molecular mass. What little is known about their functional differences is limited primarily to a comparison of their unitary conductances, voltage-dependent gating properties, or regulation by intracellular kinases, protons, and calcium. These unique physiological properties are attributed to the connexin-specific differences in primary amino acid sequences (Spray and Burt, 1990). In this study, the conductance and selectivity sequences of rat connexin40 (rCx40, molecular mass ≈40 kD)1 gap junction channels are investigated. Cx40 was chosen because it is highly expressed in Purkinje fibers, atrial myocardium, sinoatrial node, and cardiac atrioventricular conduction system in relative abundance while connexin43 (Cx43), another cardiovascular connexin, is highly expressed in ventricular myocardium and coexpressed with Cx40 in Purkinje fibers, atrial myocardium, and vascular endothelium and smooth muscle (Gourdie et al., 1993; Davis et al., 1994). These expression patterns are of particular physiological relevance since the conduction properties of these tissues vary, and it is known that Cx40 and Cx43 are not capable of forming heteromeric or heterologous gap junction channels (Bruzzone et al., 1993; Davis et al., 1994). What role the distinct channel properties of Cx40 and Cx43 play in the conduction properties of the respective tissues is not understood since relatively little is known about the distinct permeability properties of these two major cardiovascular connexins.

While ion selectivity and permeation have been studied extensively in a variety of membrane channels (Myers and Haydon, 1972; Hille, 1975; Eisenman et al., 1978; Adams et al., 1980; Dwyer et al., 1980; Blatz et al., 1984; Hess et al., 1986; Bormann et al., 1987; Franciolini and Nonner, 1987, 1994), the ionic permeabilities of connexin channels remain relatively unknown. Typically, much less is known about ion transfer in channels permeable to both cations and anions relative to cation-selective or anion-selective channels. The anion-to-cation selectivity of these channels as determined by asymmetric salt reversal potential measurements is usually not high, which may lead to the conclusion that the ability of these channels to transport ions of any valence is due to the relatively large diameter of the pores, although accounts of more complex ion–ion and ion–site interactions do exist (Franciolini and Nonner, 1994; Borisova et al., 1986). Gap junction channels with estimated pore radii ≥5 Å fall into this class. Experimental measurements of the relative ionic permeabilities from reversal potential measurements provide a theoretical method for determining the mechanism of selectivity or lack thereof. These measurements provide functional information necessary for structure-function analysis of the channel and the mechanisms that account for its selective permeability.

In this study, as in the preceding manuscript on the rat Cx43 channel (Wang and Veenstra, 1997, appearing in this issue), we used the dual whole cell recording technique to obtain single rCx40 gap junction channel currents. Using asymmetric ionic electrode solutions in each cell of an attached pair, we determined the relative permeability from the change in reversal potential measured with LiCl as the major salt in one electrode and with each of the seven cations, Cs+, Rb+, K+, Na+, TMA+, TEA+, and TBA+ as the test cation (as the Cl salt) in the second electrode. The ionic permeabilities were compared to the conductance ratios measured with symmetric solutions. We estimated the limiting pore radius of the rCx40 channels from the relationship between the relative ionic permeabilities and the effective ionic radii for the monovalent alkali cations and the organic alkylammonium ions TMA+, TEA+, and TBA+. Only minor differences were noted in the cation conductance and permeability ratios of the rCx40 channel relative to the rCx43 channel, however, the estimates of pore radii and anionic permeability sequences suggest different mechanisms of counterion transport could be involved for these two connexin channels. These data indicate that the molecular composition and macromolecular permeability of the connexin pores could be more variable than previously thought.

Methods

Cell Culture and Transfection

Genomic DNA containing the coding sequence of rat Cx40 was cloned into the EcoRI site of the eukaryotic expression vector pSFFV-neo and transfected into mouse Neuro2A (N2A) neuroblastoma cells as described previously (Beblo et al., 1995). N2A cell cultures were maintained as in the preceding manuscript (Wang and Veenstra, 1997).

Electrophysiological Recordings and Solutions

rCx40 connexin-transfected N2A cell cultures were plated at low density (2 × 105 cells/35 mm dish) for 24 h, washed with HEPES-buffered saline (in mM: 142 NaCl, 1.3 KCl, 0.8 MgSO4, 0.9 NaH2PO4, 1.8 CaCl2, 5 CsCl, 2 TEACl, 5.5 dextrose, and 10 HEPES, pH 7.4) and examined electrophysiologically on the stage of an inverted phase-contrast light microscope (IMT-2; Olympus Corp., Lake Success, NY). The compositions of the IPSs and bath solution are listed in Table I. The osmolarity of all IPS and bath solutions (310 ± 5 mosm) were measured before each experiment (Model 3W2 osmometer; Advanced Instruments Inc., Needham Heights, MA), and all experiments were performed at room temperature (20–22°C). The manufacture of the patch electrodes and analysis of junctional currents were performed following the same procedures described in the preceding manuscript (Wang and Veenstra, 1997). Double whole cell patch clamp recordings were obtained using two Axopatch-1D amplifiers (Axon Instruments, Inc., Foster City, CA). All current and voltage records were stored on VCR tape using a 4-channel digitizing unit (DR-484 Neuro-corder; Neuro Data Instruments Corp., NY) and VCR tape recorder (JVC HR-D600U) for off-line analysis.

Channel Current Analysis

Junctional channel currents were displayed as paired whole cell currents or as individual ΔI2 traces. All points current amplitude histograms were compiled from the −ΔI2 (≈Ij) trace for each experiment (Veenstra and Brink, 1992). Off-line analysis was completed following the same procedures described in the preceding manuscript (Wang and Veenstra, 1997) using the DOSTAT analysis software (Brink and Fan, 1989; Manivannan et al., 1992; Ramanan and Brink, 1993).

Cation and Anion Selective Permeability Calculations

To measure ionic selectivity reversal potentials for junctional currents in rCx40 gap junction channels were determined with 10 different asymmetric salt solution sets. Since each solution contained several ions (Table I), Erev is a function of all ion permeabilities according to the Goldman-Hodgkin-Katz voltage equation (see Wang and Veenstra, 1997). The ionic permeability ratios (Pion/PLi, see Table V) were calculated simultaneously using a matrix which contained the concentrations and ionic reversal potentials for all asymmetric solutions. The set of permeability ratios satisfied all of the Erev data listed in Table IV. Divalent cations, Mg2+ and Ca2+, which constituted <3% of the IPS ionic concentration (Table I), were constant in all IPSs with the exception of IPS KF due to precipitation and were neglected in all calculations.

Results

Single Channel Cation Chloride Salt Conductances

The single channel current amplitudes of the rCx40 channel were measured under symmetric salt conditions for all test monovalent cations. Fig. 1,A shows a segment of a double whole cell recording of a rCx40 channel in IPS KCl during a 2-min, −40-mV transjunctional voltage (Vj) pulse. The current amplitude of each channel recording is obtained by fitting the all points (real time) histogram of the −ΔI2 trace with a probability density function (pdf) as illustrated in Fig. 1,B. To illustrate the symmetry of the rCx40 channel current amplitude with respect to Vj, the current recording and histogram of the same rCx40 channel obtained at +40 mV are also illustrated in Fig. 1, C and D. Junctional channel current amplitudes (ij) for each Vj are plotted in Fig. 1 E, and the linear regression fit of the single rCx40 channel current-voltage relationship yielded a single channel slope conductance (γj) for this cell pair of 139 pS (r = 0.99).

The IPS KCl single channel current-voltage relationships from six different cell pairs were pooled together and a linear regression analysis of the composite ij-Vj relation produced a slope conductance of 141.5 pS (Fig. 2,C). This is comparable to the mean value γj of 141.2 ± 4.4 pS calculated from the independent slope conductances of the six IPS KCl rCx40 cell pairs (137.8, 137.5, 147.4, 146.2, 138.7, 139.3). These procedures were repeated for the other test cations and the ij-Vj relationships for each test solution are illustrated in Fig. 2. The number of cell pairs, slope conductance of the composite ij-Vj relations, and the mean conductance obtained from the linear slopes of the individual ij-Vj relations are summarized in Table II.

TBA rCx40 Channel Block

In six cell pairs during more than 44 min of recording currents at various transjunctional voltages with 115 mM TBACl in both cell pipettes, junctional coupling was absent with the exception of two short duration (<50 ms) channel openings in one cell pair. To verify that TBA was blocking the rCx40 channel rather than affecting the recordings through nonspecific junctional or nonjunctional effects (i.e., seal breakdown), TBACl was added to only one pipette in decreasing concentrations until the presence or absence of junctional current activity could be verified using IPS KCl. With 115 mM TBA in one pipette and 115 mM KCl in the other, nonjunctional currents (as indicated by the lack of equal and opposite polarity currents in both cells) were observed in the TBACl cell when a Vj was applied such that TBA was driven into the channel. This occurred in conjunction with TBA cell swelling and KCl cell shrinkage.

In the presence of 5 mM TBACl, Vj polarity was varied in a stepwise manner, and the resultant junctional current activity was recorded. Fig. 3,A demonstrates the presence of junctional current activity during +Vj pulses and disappearance of the same channel activity during −Vj pulses. The Vj transitions from +35 to −40 mV and from −40 to +25 mV are illustrated at higher time resolution in Fig. 3, B and C, which indicates the disappearance of and recovery of discrete channel openings. The ij-Vj relationship for this experiment is shown in Fig. 3 D. The γj of 141 pS is equal to the mean conductance obtained with symmetric IPS KCl which is consistent with an insignificant TBA current. The Vj-dependent kinetics of the onset and removal of apparent TBA block were not examined in this investigation.

Single Channel Potassium Anion Salt Conductances

Equimolar ion substitution of seven different anions (Cl, Br, glutamate plus F, NO3, acetate, and aspartate, all 115 mM K+ salts) were also performed using identical procedures to those described above for cations. The single channel currents were analyzed as presented above, and the ij-Vj relationships for each test solution are illustrated in Fig. 4. The number of cell pairs, slope conductance of the composite ij-Vj relations, and mean slope conductance of the individual linear ij-Vj relations are summarized in Table III. The divalent cations Mg2+ and Ca2+ and phosphate anion (monobasic Kphosphate) were omitted from the potassium fluoride IPS (Table I) due to problems with coprecipitation. A 6.5% reduction in total conductance was expected by these alterations. A composite slope of 148.0 pS was obtained with IPS KF. However, accounting for the reduction in osmolarity, a slope conductance of 158.3 pS was used for comparison with other test anion IPS. Phosphate also caused Cs+ to precipitate and was omitted from IPS CsCl (Table I). This was estimated to reduce the IPS CsCl total conductance by <0.4% and was therefore neglected.

Relationship between Channel Conductance and Aqueous Mobility

The monovalent cation and anion channel conductances were plotted as a function of their aqueous mobilities in Fig. 5, A and B, respectively. For the test cations, the channel conductance-mobility plot (Fig. 5,A, solid line and open circles) approximates a straight line with a slope of 18.1 pS and y-intercept of 5.0 pS (r = 0.96). The y-intercept predicts the conductance of the rCx40 channel in the presence of an impermeable (inert) cation and 136 mM Cl, although it should be noted that 12 mM Na+, 5 mM Cs+, and 2 mM TEA+ are always present in each IPS. Hence, a surprisingly low Cl conductance is estimated from this approach. Based on the aqueous diffusion coefficients for the test cations and Cl alone (i.e., no channel ionic selectivity) a line with a slope of 7.9 pS and y-intercept of 81.8 pS is predicted (dashed lines and filled circles). Clearly the difference in the two lines suggests that ions are not diffusing through the pore with the same mobility they possess in bulk solution. Even though the rCx40 channel cation conductance sequence (Table II) qualitatively matches the aqueous mobility sequence, the differences in the ionic conductance-mobility plots are indicative of a reduction of mobility created either by solvent drag within a restricted space or weak electrostatic interactions which alter the aqueous state of the ion in the pore (e.g., partial dehydration by another ion or site within the pore).

An entirely different result was obtained when the test anion channel conductances (Table III) were plotted as a function of their aqueous mobilities (Fig. 5 B). Again, the predicted anion conductance-mobility plot for the test anions, based on their respective aqueous diffusion coefficients (and no channel ionic selectivity, dashed line and filled circles), is shown for comparison to the experimental results (open circles). The experimental data do not approximate a straight line with a positive slope as predicted by independent electrodiffusion theory. In fact, channel conductance remains essentially constant over a wide range of anionic aqueous mobilities with the exception of Cl and Br, where the conductance unexpectedly decreases. In a relatively nonselective pore with weak cation–anion and ion–site interactions as expected from the cation conductance sequence, Cl and Br should be more permeable than the other anions and, thus, produce a larger unitary current and channel conductance. Because of these initial rCx40 channel conductance measurements with Br, Cl, acetate, and glutamate, we also tested F, nitrate, and aspartate which yielded identical results. Since the anion conductance mobility plot produced an inverse relation to the aqueous mobility sequence, the Levitt approximation for estimating the reduction in ionic mobilities within the pore relative to bulk solution, as applied to the rCx43 channel (Wang and Veenstra, 1997), was not attempted for the rCx40 channel.

Relative Ion Permeability Ratios and Estimation of Pore Size

The monovalent cation and anion conductance ratios suggest that there are weak cation interactions within the rCx40 pore. Yet some highly selective mechanism which paradoxically reduces channel conductance for Cl and Br and permits an essentially constant (and higher) conductance in the presence of larger organic anions or F (a more hydrated anion) also exists. To determine whether the rCx40 channel's relative ionic permeabilities actually correlate with the relative channel conductance measurements, reversal potentials were measured from ij-Vj relationships in cell pairs that were dialyzed with different IPSs on each side of the junction. In a series of experiments, the asymmetric salt reversal potentials were obtained for all of the test cation-Cl salt solutions relative to 115 mM LiCl (IPS LiCl, Table I). Fig. 6,A illustrates the ij-Vj relationship obtained from four cell pairs using asymmetric IPS KCl/LiCl salt gradients. Vj was defined as the voltage of the IPS KCl cell relative to the LiCl cell (Vj = VKCl − VLiCl) even though both cells were pulsed at different times during the course of each experiment. This convention is necessary to avoid transposing the measured channel current amplitudes on the voltage axis when using asymmetric salt solutions (see Wang and Veenstra, 1997). The zero current intercept of the composite linear ij-Vj relationship equals the reversal potential (Erev) for the cell pairs with opposing K+ and Li+ concentration gradients and is in close agreement with the mean Erev of +5.0 ± 0.9 mV (mean ± SE) determined from the Erev of each of the four experiments. Erev was determined with each asymmetric solution set using the identical protocol illustrated in Fig. 6,A; the results are summarized in Table IV.

To determine the permeability ratio PCl/PLi, Erev was also obtained with different concentrations of LiCl on each side of the junction. One pipette contained 115 mM LiCl (IPS LiCl) and the other pipette a solution with the LiCl reduced to 30 mM and mannitol (145 mM) added to maintain osmotic balance with IPS LiCl. All other ionic components of the IPS remained unaltered.

Fig. 6,B illustrates the composite ij-Vj relationship obtained from six cell pairs with asymmetric LiCl solutions. The Erevs for each set of asymmetric solution experiments are listed in Table IV. To test whether mannitol permeates the rCx40 channel, one additional experiment was performed in which stachyose, a tetrasaccharide, replaced mannitol as the IPS osmotic balancing agent. The reversal potential of 16.3 mV with stachyose compared well with the average reversal potential measured with mannitol of 16.1 mV as listed in Table IV.

Asymmetric salt reversal potential experiments were also performed on two of the organic anions with the highest and lowest aqueous diffusion coefficients (∝ aqueous mobilities), nitrate and glutamate, to determine their permeabilities relative to Cl. Fig. 6 C illustrates the composite ij-Vj relationship obtained from five cell pairs with asymmetric IPS Kglutamate/KCl gradients across the junction.

Ionic relative permeabilities were determined using the Goldman-Hodgkin-Katz voltage equation (see Eq. 3 of Wang and Veenstra, 1997). For comparison, the junctional conductance ratios, theoretical aqueous permeabilities, and ionic permeability coefficients relative to Li+ are all listed in Table V. The permeability and conductance ratios both agree qualitatively with the relative cation mobility sequence of Cs+ > Rb+ > K+ > Na+ > Li+ > TMA+ > TEA+. This permeability sequence corresponds to an Eisenman I sequence. The observation that the permeability and conductance ratios calculated are significantly lower than the aqueous diffusion coefficient ratios suggests that there is some finite interaction (e.g., partial dehydration or steric hindrance) occurring between the monovalent cations and the pore of the rCx40 gap junction channel.

Channel radii have been estimated previously from permeability ratios (Renkin, 1954; Levitt, 1975, 1991; Dwyer et al., 1980) assuming a cylindrical mechanical pore model with radius r. The permeability ratios are described as a function of the radii ratio α, which is defined as a/r, where a is the ion spherical radius. Drag factors as a function of a or α have also been considered. We have applied the following function which is a solution to the hydrodynamic continuum equation (Dwyer et al., 1980) assuming that the ion travels along the axis in an infinite uniform cylinder:

\begin{equation*}P_{X}/P_{Li}=C(1-{\alpha})^{2}[1-2.105{\alpha}+2.0865{\alpha}^{3}-1.7068{\alpha}^{5}+0.72603{\alpha}^{6}/(1-0.75857{\alpha}^{5})],\end{equation*}

for a chi-squared fit of PX/PLi as shown in Fig. 7. The quantity C(1 − α)2 represents a volume displacement term which estimates the effective target areas for ion entry into the pore relative to their respective cross-sectional areas. The best fit of the permeability data was obtained with C = 32.6 and an estimated pore radius of 6.6 ± 0.7 Å (mean ± standard error of pore radius determined by chi-squared fit of the data).

Discussion

The purpose of this investigation was to determine the monovalent cation and anion selectivity of the rCx40 gap junction channel. In addition to electrostatic interactions at specific pore sites where the ion could be temporarily bound or be repulsed by like charges, other factors affecting ion permeability include increasing binding affinity (decreasing hydration) and pore size. Ion diffusion in a pore is classically described as being inversely related to frictional drag which depends on the ratio of the effective ion radius to the limiting pore radius (Anderson and Quinn, 1974; Levitt, 1975). If current is forced to flow by an applied voltage, electro-osmosis occurs whereby moving ions drag water molecules. Removal of water molecules from permeant ions (hydration-dehydration rates) and channel transformation (e.g., conductance state and kinetics) during ion transit may also affect permeability (Hille, 1992).

For many ion channels, ion permeation deviates from classical electrodiffusion theory due to loss of independence (ion–ion interactions) and binding or repulsion from specific sites within the pore (ion–site interactions). These interactions are enhanced by restricted space at the site of these interactions. As a consequence, the single channel conductance sequences and permeability sequences for a series of ions are not necessarily correlated. The lack of correspondence can occur in saturating systems because two coefficients, a binding constant and a rate constant, are necessary to define conductance. In contrast, only one constant (the product of the binding constant and rate constant) is needed to define permeability (Eisenman et al., 1978). Hess et al. (1988) reported that the monovalent cation sequence from Erev measurements stands in contrast to single channel conductance for Ca2+ channels in isolated guinea pig cardiac ventricular myocytes. Thus, it is necessary to obtain an independent measure of relative permeability, by measuring reversal potentials under asymmetric salt conditions, and/or by directly testing for independence (e.g., saturation etc.). These circumstances might be relevant to gap junctions because their limiting channel radii of ≥5 Å are presumed to accommodate both anions and cations simultaneously. Furthermore, the gap junction channel is a long pore with a length ≥150 Å. Flux coupling, a result of crowded conditions in a long, narrow pore, can reduce the effective mobility of each permeant ion, although this concept has not been generally applied to gap junctions. The permeability and conductance behavior produced by ion–ion interactions can manifest themselves as concentration-dependent permeability ratios (Sandblom et al., 1977; Eisenman et al., 1978) and “anomalous mole-fraction dependent” conductances (Andersen, 1975; Neher, 1975).

The two methods applied in this study are commonly used to determine ionic selectivity from voltage clamp measurements. The first method, comparing the conductance in a test and control solution, requires that the conductance measured with each solution corresponds to the same conductance state of each channel, and that there is no saturation or block of open channels by any ion in solution. In this investigation we used maximum single channel conductances. The second method uses the asymmetric salt reversal potential, obtained from plots of single channel peak current versus voltage, and the Goldman-Hodgkin-Katz voltage equation. Reversal potential measurements are not affected by the number of conducting channels, block, or simple saturation. Diffusion potential errors were minimized for all cation reversal potential experiments by keeping Cl constant (136 mM) in the Ag/AgCl half-cell reactions. Any offset potentials between the recording electrodes and ground were compensated by the junction potential potentiometer of the patch clamp amplifier before GΩ seal formation. Larger electrode offset potentials resulting from anion substitution, since Cl was reduced to 21 mM in these cases, were similarly compensated before GΩ seal formation and subsequent whole cell recording (see Wang and Veenstra, 1997). Hence, the asymmetric salt reversal potential of the corresponding ij-Vj relationship reflects the differences in ionic permeabilities across the rCx40 junction.

These results demonstrate that the conductance and permeability sequences for monovalent cations and anions are identical (Cs+ > Rb+ > K+ > Na+ > Li+ > TMA+ > TEA+ > TBA+ and glutamate ≥ acetate ≥ aspartate ≥ nitrate > F > Cl > Br). The cation sequence also follows the aqueous mobility sequence for the test cations with the exception of TMA+ being less permeable than Li+. This is equivalent to the Eisenman I sequence and suggests only weak electrostatic interactions with other ions with charged sites within the rCx40 pore. This is in slight contrast to the Eisenman II sequence observed for the rCx43 pore (Wang and Veenstra, 1997), although neither pore possesses a pronounced affinity for the monovalent cations as evidenced by their conductance and permeability sequences. The most notable difference between the rCx40 and rCx43 channels is in the relative conductance and permeability ratios for Li+ and Na+. In the rCx40 channel the conductance and permeability ratios of Li+ were less than those of Na+, but both were approximately equal for Na+ and Li+ in the rCx43 channel. However, in neither case do the differences in conductance mimic the differences in the respective ionic aqueous mobilities. The cation conductance- mobility (γjcation) plots for the rCx40 channel suggests that channel conductance is predominantly dependent on cationic flux with ≤5 pS of total channel conductance being attributed to 136 mM Cl (Fig. 5 A). The slope of the rCx40 γjcation relationship (18 pS) is nearly double that of the rCx43 channel, indicative of a more cation selective channel. However, the rCx40 PCl/ PLi = 0.19 determined from the asymmetric 115/30 mM LiCl experiments with mannitol is nearly identical to the PCl/PLi of 0.18 determined for the rCx43 channel using the same 115/30 mM LiCl gradient with raffinose. PCl/PLi ratios were also determined using stachyose for both rCx40 and rCx43 channels with identical results to those mentioned above.

The anion conductance and permeability sequences for the rCx40 channel are in sharp contrast to the aqueous mobility sequence. As opposed to the Eisenman I permeability sequence observed for the rCx43 channel, the rCx40 channel conductance remains essentially constant at 158–160 pS for all of the nonatomic anions (glutamate, aspartate, acetate, and nitrate). Only the halide anions Br, Cl, and F exhibited lower γj values, and their values were also the reverse of what is expected from their aqueous mobilities. It should be noted that the IPS solution composition with potassium fluoride differed slightly compared to the other salts due to precipitation (Table I). The necessary conductance scaling for comparison with other salt conductances may have affected the placement of fluoride in the overall monovalent anion sequence. The relative permeability of two oxy anions with the highest and lowest mobilities (nitrate and glutamate) relative to Cl was determined to test this phenomenon based on the conductance ratio determinations. Again, the permeability coefficients determined from the asymmetric salt reversal potential experiments (Table V and Fig. 6 C) confirmed the findings of the conductance ratio experiments.

From these observations, two possible mechanisms could account for the observed effects of anion substitution upon rCx40 channel conductance. First, higher channel conductances were observed for the oxy anions (O = C-O or O = N-O groups) which suggests that these groups are preferentially conducted as indicated by the permeability coefficient calculations (Table V). However, this explanation also requires that any oxy anion has a similar conductance irrespective of the size of the entire molecule. This is not likely to occur in a partially hydrated pore and is more consistent with the general characteristics of a facilitated anion transport mechanism. An attempt was made to address the issue of oxy anion versus halide anion by performing conductance measurements with F since its aqueous mobility is between that of nitrate and glutamate, but the results were inconclusive due to differences in the solubility of the various anions.

The second possible explanation is that the higher mobility anions actually do permeate through the pore of the rCx40 channel more readily than their lower mobility counterparts (nitrate through glutamate), but as a neutral ion pair associated with a permeant potassium ion, not as a counterion flow of opposite direction to the cation flux as is generally assumed for a nonselective aqueous pore. Hence, the more permeant anion reduces the total current rather than contributing to it as in the rCx43 channel. As we have proposed in the companion paper (Wang and Veenstra, 1997), the rCx43 channel exhibits a cation-dependent anion permeability where the anion transiently binds to a cation as it encounters a cation at a fixed anionic site in the pore. However, the anion in the rCx43 pore is moving in the opposite direction to the cationic flux in a manner analogous to the model proposed by Borisova et al. (1986). As an example, extrapolating the γjanion plot (Fig. 5,B) to the y-intercept provides a maximum conductance estimate of 160.7 pS. Using the relative permeability coefficients for K+, Cs+, Na+, TEA+, and Cl and their respective ionic concentrations in the IPS, Cl would be expected to contribute about 19.7 pS to the total conductance. Assuming glutamate is 5.28 times less permeable than Cl, which is the reciprocal of the Pglutamate/PCl value in Table V, then glutamate would be expected to contribute about 3.2 pS to the total conductance of the rCx40 channel. By assuming that the anion is forming anion-cation complexes in the pore, Cl and glutamate should be reducing the channel conductance by this equivalent amount, which gives a predicted conductance of 141 and 157 pS respectively. These theoretical calculations are very close to the corresponding experimental γj values of 142 and 160 pS. This proposed mechanism is similar to one described by Franciolini and Nonner (1994) for an anion channel where cation permeation was dependent on the formation of transient cation-anion complexes over the entire length of the channel, and reducing the total charge flux below that sustainable by anions alone. Since net current is reduced by anion permeation, this interpretation is consistent with the reversal potentials obtained under asymmetric anion salt conditions and translates into PY/PCl coefficients which are the reciprocal of those listed in Table V. This is a testable hypothesis since anion permeability ratios will vary depending on salt concentration and on which permeant cation is present. Furthermore, there should be no anion conductance in the presence of an impermeant cation for the rCx40 pore and the maximum conductance should be achieved in the presence of an impermeant anion. However, such ions have yet to be identified for gap junction channels.

Specific ionic block of a connexin channel also has not been demonstrated previously. Experiments with TBACl strongly suggest that TBA+ is capable of entering, but not traversing the rCx40 pore. Findings are presented (Fig. 3) which suggest that unilateral addition of only 2 or 5 mM TBACl leads to a block of rCx40 channel activity when the Vj polarity is net positive for the TBA+-containing cell. This process is slowly reversible upon switching the Vj polarity to net negative with respect to the TBA+-containing cell. This occurs without alteration of the single channel conductance for the events observed prior to or after removal of block. The concentration-dependence, kinetics (e.g., time to first opening after block), and Vj-dependence of TBA+ block were not examined in detail in this investigation. The data presented demonstrate the reproducibility and reversibility of the observed phenomenon and are consistent with the proposed mechanism of action.

The results with TBA+ and mannitol suggest that these two molecules are not permeable through the rCx40 pore. This is in contrast to the results with rCx43 where mannitol did not produce a sustained reversal potential in asymmetrical 115/30 mM LiCl. Brief records with symmetrical 115 mM TBACl also indicated the presence of rCx43 junctional channel activity although the nonjunctional membrane noise increased during the course of the experiments and eventually precluded the resolution of unitary junctional currents (Wang and Veenstra, 1997). TBA+ has an effective radius of 4.94 Å and mannitol, being more linear and containing fewer carbon atoms, should be even smaller. Taken together, these results suggest that the higher conductance rCx40 channel has a smaller limiting pore radius than the rCx43 channel. Estimates of pore radius for rCx43 were 6.3 ± 0.4 Å (Wang and Veenstra, 1997), and the same approach (i.e., cation permeability ratios and the hydrodynamic equation) gives an estimate of 6.6 ± 0.7 Å for the rCx40 pore. These values are essentially indistinguishable from each other. Identification of the largest permeant and smallest impermeant cations for the rCx40 channel will ultimately provide the best estimate of pore size and allow direct comparison to other connexin channels. In the present investigations, the largest permeant monovalent cation used in the conductance and permeability measurements had a relative ion/pore radius of ≤0.65 according to the estimates of pore radius. As modeled in the rCx43 channel (Wang and Veenstra, 1997), the limiting radius of ≈6.6 Å for the rCx40 pore is sufficient to produce a greater reduction in the mobility of TMA+ relative to the more hydrated Li+, resulting in a switch of positions in the conductance ratio sequences (Li+ ≃ TMA+) for both channels.

In summary, our investigation of monovalent cation and anion permeability and conductance ratios leads to the conclusion that the rCx40 channel is not a simple aqueous pore. Monovalent cations apparently permeate through the pore in a partially hydrated state with reduced mobilities determined by the effective ionic radius relative to the limiting pore radius of ≈6.6 Å. The rCx40 channel appears to have a low anionic permeability which could be due to one or more fixed anionic sites within the pore. The mechanism of anion permeation remains to be definitively determined although all results suggest that formation of cation-anion complexes is probable. The differences in cation, anion, and molecular (mono-, tri-, and tetrasaccharide) permeabilities suggest that differences exist in charge composition and pore size for rCx40 and rCx43 channels. These results should provide a framework for future examination of structure-function differences among connexin channels in the pore-lining amino acid sequences.

references

Adams
DJ
,
Dwyer
TM
,
Hille
B
The permeability of endplate channels to monovalent and divalent metal cations
J Gen Physiol
1980
75
493
510
[PubMed]
Andersen, O.S. 1975. Ion-specificity of gramicidin channels. In International Biophysics Congress. Copenhagen, Denmark. 112. (Abstr.).
Anderson
JL
,
Quinn
JA
Restricted transport in small pores. A model of steric exclusion and hindered particle motion
Biophys J
1974
14
130
150
[PubMed]
Beblo
DA
,
Wang
H-Z
,
Beyer
EC
,
Westphale
EM
,
Veenstra
RD
Unique conductance, gating, and selective permeability properties of gap junction channels formed by connexin40
Circ Res
1995
77
813
822
[PubMed]
Bennett, M.V.L., and D.C. Spray. 1965. Gap Junctions. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York.
Beyer
EC
Gap junctions
Int Rev Cytol
1993
137C
1
37
[PubMed]
Beyer, E.C., and R.D. Veenstra. 1994. Molecular biology and electrophysiology of cardiac gap junction proteins. In Handbook of Molecular and Cellular Physiology of Membrane Channels. C. Peracchia, editor. Academic Press, San Diego, CA. 379–401.
Blatz
AL
,
Magleby
KL
Ion conductance and selectivity of single calcium-activated potassium channels in cultured rat muscle
J Gen Physiol
1984
84
1
23
[PubMed]
Borisova
MP
,
Brutyan
RA
,
Ermishkin
LN
Mechanism of anion-cation selectivity of amphotericin B channels
J Membr Biol
1986
90
13
20
[PubMed]
Bormann
J
,
Hamill
OP
,
Sakmann
B
Mechanism of anion permeation through channels gated by glycine and γ-aminobutyric acid in mouse cultured spinal neurones
J Physiol (Lond)
1987
385
243
286
[PubMed]
Brink
PR
,
Fan
SF
Patch clamp recordings from membranes which contain gap junction channels
Biophys J
1989
56
579
593
[PubMed]
Bruzzone
R
,
Haefliger
J-A
,
Gimlich
RL
,
Paul
DL
Connexin40, a component of gap junctions in vascular endothelium, is restricted in its ability to interact with other connexins
Mol Biol Cell
1993
4
7
20
[PubMed]
Caspar
DLC
,
Goodenough
DA
,
Makowski
L
,
Phillips
WC
Gap junction structures. I. Correlated electron microscopy and x-ray diffraction
J Cell Biol
1977
74
605
628
[PubMed]
Davis
LM
,
Kanter
HL
,
Beyer
EC
,
Saffitz
JE
Distinct gap junction protein phenotypes in cardiac tissues with disparate conduction properties
J Am Coll Cardiol
1994
24
1124
1132
[PubMed]
Dwyer
TM
,
Adams
DJ
,
Hille
B
The permeability of the endplate channel to organic cations in frog muscle
J Gen Physiol
1980
75
469
492
[PubMed]
Eisenman
G
,
Sandblom
J
,
Neher
E
Interactions in cation permeation through the gramicidin channel Cs, Rb, K, Na, Li, Tl, H, and effects of anion binding
Biophys J
1978
22
307
340
[PubMed]
Franciolini
F
,
Nonner
W
Anion and cation permeability of a chloride channel in rat hippocampal neurons
J Gen Physiol
1987
90
453
478
[PubMed]
Franciolini
F
,
Nonner
W
A multi-ion permeation mechanism in neuronal background chloride channels
J Gen Physiol
1994
104
725
746
[PubMed]
Gourdie
RG
,
Severs
NJ
,
Green
CR
,
Rothery
S
,
Germroth
P
,
Thompson
RP
The spatial distribution and relative abundance of gap-junctional connexin40 and connexin43 correlate to functional properties of components of the cardiac atrioventricular conduction system
J Cell Sci
1993
105
985
991
[PubMed]
Hertzberg, E.L., and R.G. Johnson. 1988. Gap Junctions. Alan R. Liss, Inc., New York.
Hess
P
,
Lansman
JB
,
Tsien
RW
Calcium channel selectivity for divalent and monovalent cations. Voltage and concentration dependence of single channel current in ventricular heart cells
J Gen Physiol
1986
88
293
319
[PubMed]
Hille, B. 1975. Ionic selectivity of Na and K channels of nerve membranes. In Membranes: A Series of Advances, Vol. 3, Dynamic Properties of Lipid Bilayers and Biological Membranes. G. Eisenman, editor. Marcel Dekker, Inc., New York. 255–323.
Hille, B. 1992. Ionic Channels of Excitable Membranes. Sinauer Associates, Inc., Sunderland, MA.
Levitt
DG
General continuum analysis of transport through pores
Biophys J
1975
15
533
551
[PubMed]
Levitt
DG
General continuum theory for multiion channel II. Application to acetylcholine channel
Biophys J
1991
59
278
288
[PubMed]
Loewenstein
WR
Junctional intercellular communication. The cell-to-cell membrane channel
Physiol Rev
1981
61
829
913
[PubMed]
Makowski
L
,
Caspar
DLC
,
Phillips
WC
,
Goodenough
DA
Gap junction structures. II. Analysis of the x-ray diffraction data
J Cell Biol
1977
74
629
645
[PubMed]
Manivannan
K
,
Ramanan
SV
,
Mathias
RT
,
Brink
PR
Multichannel recordings from membranes which contain gap junctions
Biophys J
1992
61
216
227
[PubMed]
Myers
VB
,
Haydon
DA
Ion transfer across lipid membranes in the presence of gramicidin A. II. The ion selectivity
Biochem Biophys Acta
1972
274
313
322
[PubMed]
Neher
E
Ionic specificity of the gramicidin channel and the thallous ion
Biochem Biophys Acta
1975
401
540
544
Nightingale
ER
Phenomenological theory of ion solvation. Effective radii of hydrated ions
J Phys Chem
1959
63
1381
1387
Peracchia
C
Low resistance junctions in crayfish. II. Structural details and further evidence for intercellular channels by freeze-fracture and negative staining
J Cell Biol
1973
57
66
80
[PubMed]
Ramanan
SV
,
Brink
PR
Multichannel recordings from membranes which contain gap junctions. II. Substates and conductance shifts
Biophys J
1993
65
1387
1395
[PubMed]
Renkin
EM
Filtration, diffusion, and molecular sieving through porous cellulose membranes
J Gen Physiol
1954
38
225
243
[PubMed]
Robinson, R.A., and R.H. Stokes. 1955. Electrolyte Solutions. Academic Press, Inc., New York.
Sandblom
J
,
Eisenman
G
,
Neher
E
Ionic selectivity, saturation and block in gramicidin A channels. I. Theory for the electrical properties of ion selective channels having two pairs of binding sites and multiple conductance states
J Membr Biol
1977
31
383
417
[PubMed]
Spray
DC
,
Burt
JM
Structure-activity relations of the cardiac gap junction channel
Am J Physiol
1990
258
C195
C205
[PubMed]
Veenstra, R.D., and P.R. Brink. 1992. Patch clamp analysis of gap junctional currents. In Cell-Cell Interactions: A Practical Approach. Stevenson, B., D.L. Paul, and W. Gallin, editors. IRL Press, Oxford, U.K. 167–201.
Wang
H-Z
,
Veenstra
RD
Monovalent cation and anion selectivity sequences of the rat connexin43 gap junction channel
J Gen Physiol
1997
109
491
507
[PubMed]

The rat Cx40 transfected N2A cells were courteously provided by Dr. Eric C. Beyer, Washington University School of Medicine, St. Louis, MO 63110. We thank Mark G. Chilton for maintaining the N2A cell cultures. We wish to thank Dr. Peter R. Brink, Dept. of Physiology and Biophysics, SUNY Health Science Center, Stony Brook, NY 11794 for helpful discussions about the biophysical determinants of ionic permeability within a pore and Dr. Joseph D. Robinson for his helpful comments on the manuscript.

The research was supported by National Institute of Health grants HL-42220 and HL-45466.

1

Abbreviations used in this paper: N2A, neuroblastoma 2A; rCx40, rat connexin40.

Author notes

Address correspondence to Dr. Dolores A. Beblo, Department of Pharmacology, SUNY Health Science Center at Syracuse, 750 East Adams Street, Syracuse, NY 13210. Fax: 315-464-8014.