This Article Abstract Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Veenstra, R. D. Articles by Brink, P. R. Search for Related Content PubMed PubMed Citation Articles by Veenstra, R. D. Articles by Brink, P. R.

(Circulation Research. 1995;77:1156-1165.)
© 1995 American Heart Association, Inc.

Articles

Selectivity of Connexin-Specific Gap Junctions Does Not Correlate With Channel Conductance

Richard D. Veenstra, Hong-Zang Wang, Dolores A. Beblo, Mark G. Chilton, Andrew L. Harris, Eric C. Beyer, Peter R. Brink

From the Department of Pharmacology (R.D.V., H.-Z.W., D.A.B., M.G.C.), State University of New York Health Science Center at Syracuse; the Department of Biophysics (A.L.H.), Johns Hopkins University, Baltimore, Md; the Department of Pediatrics and Cell Biology and Department of Physiology (E.C.B.), Washington University School of Medicine, St Louis, Mo; and the Department of Physiology and Biophysics (P.R.B.), State University of New York Health Science Center at Stony Brook.

Correspondence to Dr Richard D. Veenstra, Department of Pharmacology, SUNY Health Science Center at Syracuse, Syracuse, NY 13210. E-mail veenstrr@vax.cs.hscsyr.edu.

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
Abstract Connexins form a variety of gap junction channels that vary in their developmental and tissue-specific levels of expression, modulation of gating by transjunctional voltage and posttranslational modification, and unitary channel conductance (_j). Despite a 10-fold variation in _j, whether connexin-specific channels possess distinct ionic and molecular permeabilities is presently unknown. A major assumption of the conventional model for a gap junction channel pore is that _j is determined primarily by pore diameter. Hence, molecular size permeability limits should increase and ionic selectivity should decrease with increasing channel _j (and pore diameter). Equimolar ion substitution of 120 mmol/L KCl for potassium glutamate was used to determine the unitary conductance ratios for rat connexin40 and connexin43, chicken connexin43 and connexin45, and human connexin37 channels functionally expressed in communication-deficient mouse neuroblastoma (N2A) cells. Comparison of experimental and predicted conductance ratios based on the aqueous mobilities of all ions according to the Goldman-Hodgkin-Katz current equation was used to determine relative anion-to-cation permeability ratios. Direct correlation of junctional conductance with dye transfer of two fluorescein-derivatives (2 mmol/L 6-carboxyfluorescein or 2',7'-dichlorofluorescein) was also performed. Both approaches revealed a range of selectivities and permeabilities for all five different connexins that was independent of channel conductance. These results are not consistent with the conventional simple aqueous pore model of a gap junction channel and suggest a new model for connexin channel conductance and permselectivity based on electrostatic interactions. Divergent conductance and permeability properties are features of other classes of ion channels (eg, Na⁺ and K⁺ channels), implying similar mechanisms for selectivity.

Key Words: gap junction • channel • selectivity • 6-carboxyfluorescein • 2',7'-dichlorofluorescein

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
Gap junction channels are formed by a 12-member family of connexin proteins that share many principal structural features (eg, transmembrane topology and regions of conserved amino acid residues).¹ Gap junction channels are composed of an aqueous pore with a slight negative charge that is 0.8 to 1.4 nm in diameter, which permits the passage of soluble molecules of up to 1 kD in size.^{2 3 4 5} Potentially permeable molecules include a wide variety of molecules (eg, cAMP, inositol tris-phosphate, Ca²⁺, ATP, and morphogens) besides the obvious exchange of ions between adjacent cells.^{6 7 8 9 10 11 12 13 14} The size (M_r or diameter) permeability limit to fluorescent tracers decreases reciprocally with increasing negative charge of the permeant dye molecule,^{2 5} suggesting that gap junction pores contain fixed anionic charges, an observation substantiated by low Cl^-/K⁺ permeabilities (0.52 to 0.69) relative to Na⁺/K⁺ permeabilities of 0.81 to 0.84 for the same gap junction channels as determined by _j ratios or bi-ionic reversal potentials.^{15 16 17}

These classic gap junction channel permeability studies predate the identification of the connexins, and similar permeability studies involving any of the connexin-specific junctional channels have not been performed. _j values of homotypic connexin channels (all 12 connexins are identical) are known to vary from 26 to 300 pS.^{17 18} Connexin channels also exhibit multiple conductances or subconductance states that may be regulated by V_j or connexin phosphorylation.^{19 20 21 22 23} For a right cylindrical (simple) aqueous pore, as gap junction channels are assumed to be, _j and permeability are both direct functions of the pore cross-sectional area.²⁴ If true, connexin channels with higher _j values would exhibit greater permeability to hydrophilic molecules of increasing size and valence. The physiological consequences of known connexin-specific channel conductance and subconductance states, namely, state-dependent changes in molecular permselectivity, are also minimized by the simple pore hypothesis. However, recent evidence suggests that low- and high-_j connexin-specific channels exhibit significantly lower anion and anionic dye permeabilities than previously attributed to gap junction channels.^{22 23} We have examined the relative anion-to-cation ratio and differential dye permeability of several connexin-specific channels with respect to their maximum _j and gating properties. These connexins include Cx43, Cx40, Cx45, and Cx37, which are predominantly expressed in specialized tissues of the mammalian cardiovascular system such as ventricular myocardium (Cx43), atrial, nodal, and His-Purkinje tissues (Cx40, Cx43, and Cx45), and vascular endothelium and smooth muscle (Cx40, Cx37, and Cx43). Cx43 is the most ubiquitous connexin in the mammalian body, since it is known to be expressed in a variety of epithelial and smooth muscle tissues in addition to the cardiovascular tissues.¹ Our results favor a new model based on electrostatic charge and relative pore-to-permeant molecular size ratio for determining the conductance and permeability properties of connexin-specific gap junction channels, which may influence the flux of physiologically relevant ions and second messengers between like or different cell types.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
Full-length connexin genomic or cDNAs were cloned into the EcoRI site of the pSFFV-neo expression vector.^{17 18} Mouse N2A neuroblastoma cells were transfected with 20 µg of linearized plasmid by use of the lipofectin reagent (GIBCO/BRL) according to manufacturer's directions, and stable neomycin-resistant colonies were selected in 0.5 mg/mL G418 (GIBCO/BRL). Connexin expression was verified by Northern blot analysis as previously reported for chicken Cx43 and Cx45,¹⁷ human Cx37,¹⁸ and rat Cx40 and Cx43 (data not shown). N2A cells were cultured as described previously.^{17 18}

I_js were obtained by using conventional double whole-cell patch-clamp procedures by stepping the holding potential of the prejunctional cell (V₁) from a common value (V₁=V₂=0 mV) to a test potential for durations 2 minutes.²⁵ N2A cells were rinsed with HEPES-buffered saline containing (mmol/L) NaCl 142, KCl 1.3, MgSO₄ 0.8, NaH₂PO₄ 0.9, CaCl₂ 1.8, dextrose 5.5, and HEPES 10, pH 7.2, with 1N NaOH. Patch electrodes had patch resistances of 2 to 5 M when filled with one of three internal pipette solutions (Table 1). CsCl (15 mmol/L)and TEA chloride (10 mmol/L) were added to the bath saline when IPS 2 or IPS 3 was used. Only those cell pairs in which single-channel activity could be resolved were analyzed (g_j, <0.5 nS). All-points current-amplitude histograms were compiled from the -I₂ trace excluding the first 10 seconds of each voltage pulse. All analog signals were low pass–filtered (eight-pole Bessel, LPF-30, WPI Inc) at 100 Hz and digitized at 2 kHz (DT2801A board, Data Translation Inc) installed in an IBM PC/AT clone (Everex 386SX/20). The dead time of the recording instrumentation was 1.8 milliseconds.

View this table: [in this window] [in a new window]   Table 1. Composition of Internal Pipette Solutions

The gaussian peaks observed in each current amplitude histogram were fitted with a pdf assuming n independent channels, where n is one less than the number of observed peaks. Each channel was assigned an open current amplitude, variance, and probability (cumulative open time=area under each peak) as described previously.^{22 23 26} The duration of each channel recording was 2 minutes unless otherwise indicated. Closed and open channel variances were 0.4±0.2 pA, and event counts, determined after the pdf fit of each histogram, typically ranged between 20 and 100 events for each 2-minute V_j pulse. Event counts were higher for multichannel records and were occasionally 20 (minimum, 1) for some single-channel records, although channel amplitude measurements (ie, pdf fits) are unaffected by the event count, since this is not a variable for the pdf. For some histograms, accurate pdf fits could not be generated, and only channel amplitudes and event counts were obtained. Junctional channel I-V curves were constructed for each experiment by plotting the current amplitudes observed at each V_j. _j values for each connexin examined were taken as the slope conductance of the I-V plot for each experiment.

The _j ratio method was used to estimate R_p by calculating theoretical _j values for each connexin channel when switching from IPS 1 or IPS 2 to IPS 3 (120 mmol/L glutamate^- for Cl^- substitution). An R_p of 1 indicates equal permeability for cations and anions, an R_p of <1 indicates a higher selectivity for cations, and an R_p of >1 indicates a higher selectivity for anions. R_p values were determined from the following expression:

(1)

where I_s is the current carried by each ion s, [S] is solute concentration, z_s is the valence of ion s, and P_s is D_sß_s/l=RTµ_sß_s/Fl=K_sµ_s. P_s is permeability of solute s, D_s is the diffusion coefficient, ß_s is the partition coefficient, µ_s is aqueous mobility, l is pore length, Ks=RTß_s/Fl=constant, and FE/RT has its usual meaning. The estimated effective R_p was determined by scaling an anionic permeability coefficient (x_a=R_p) until the theoretical results matched the experimental IPS 3–to–IPS 2 _j ratio. The D_s values used in the above expression were (10^-5 cm²/s): K⁺ 1.96, Cs⁺ 2.06, TEA⁺ 0.87, Cl^- 2.03, and glutamate^- 0.7. The R_p term assumes that there is no selectivity between Cl^- and glutamate^- other than their mobility difference. The essential assumptions of this model are that all ions perceive the pore as a solvent space (eg, aqueous environment) and that there is no multi-ion occupancy of specific sites within the pore.

The D_s value for glutamate^- is an extrapolation based on the relation between molecular weights and measured diffusion constants for glycine (0.97) and arginine (0.59), whose molecular weights bracket that of glutamate. Assuming a D_s of zero for glutamate^- would increase the predicted IPS 3–to–IPS 2 _j ratio to 1.72, thereby reducing the R_p estimates for all connexins accordingly (see Table 2). All relative permeability estimates assume that channel conductance is limited by the aqueous diffusion coefficients for all permeant monovalent ions.

View this table: [in this window] [in a new window]   Table 2. Open Probabilities and Event Counts for All Rat Cx43 Channel Recordings

Dye transfer of 6-CF (376 D, Molecular Probes) or diCl-F (401 D, Eastman Kodak Co) was assessed by adding 2 mmol/L dye to IPS 2 in one of the recording pipettes. For every experiment, g_j was assessed over a 10-minute observation period by applying 5-second -40-mV V_j pulses at a rate of 4/min (33% duty cycle) to the dye-containing cell. Dye (2 mmol/L) was prepared from 20 mmol/L stock solution (stored in the dark at -20°C) daily.²³ After a 10-minute recording period, the presence or absence of dye transfer was initially observed under epifluorescent illumination. Phase-contrast and fluorescent micrographs were taken with an automatic exposure Olympus 35-mm camera body attached to the microscope's camera port.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
The conventional model for a gap junction channel, based on structural and permeability data of a decade ago, is an aqueous right cylindrical pore 8 to 14 Å in diameter and 16 nm in length. The limiting conductance for such a pore is determined by the diffusion coefficients of the ions and the physical dimensions of the pore according to Ohm's law.²⁴ Fig 1 illustrates these elementary properties of an aqueous pore with a resistivity of 100 ·cm and the approximate physical dimensions of a gap junction channel. The expected range of conductances for the pore and the channel is given by the following equation:

(2)

where R_channel, R_pore, and R_access indicate channel, pore, and access resistance, respectively; l is length, a is radius, and is resistivity in ·cm. The difference between the pore and channel conductances illustrated in Fig 1B is due to the contribution of R_access to R_channel. Any reductions in the mobilities of a permeant monovalent ion due to increased solvent viscosity or frictional drag within the pore would require an increase in channel diameter to achieve similar conductance values.

View larger version (26K): [in this window] [in a new window]   Figure 1. Simple aqueous pore model for a gap junction channel. A, Schematic drawing of the longitudinal and cross-sectional profile of a gap junction channel. An 8-Å-diameter pore has a cross-sectional area of 50 Å2 (inner circle with 1-Å grid), whereas a 14-Å diameter pore has a cross-sectional area of 154 Å2. For comparison, a sodium or potassium ion hydrated by a single water molecule has a diameter of 5 or 6 Å, respectively. raccess and rpore indicate access and pore resistance, respectively. B, Graph of pore (dashed line) and channel (solid line) conductances for a 16-nm-long channel with resistivity () of 100 ·cm as determined by using Equation 2. For an 8- to 14-Å-diameter channel, predicted channel conductances range from 30 to 100 pS (crosshatched area). The observed range of conductances for transfected connexin channels ranges from 30 to 300 pS (stippled area).22 23

To test the basic assumptions of this model, we varied the monovalent ionic composition of the internal pipette solutions used in double whole-cell recordings of gap junction channel currents for all five functionally expressed connexins. All three internal pipette solutions use either potassium glutamate or KCl as the principal salt (Table 1). On the basis of the assumption that the permeability of each ion is directly proportional to its aqueous mobility (see "Materials and Methods"), a 64% or 36% increase in _j is expected when substituting IPS 3 for IPS 1 or IPS 2 according to Equation 1, respectively. Actual _j values were obtained for each connexin in IPS 2 (or IPS 1 in the case of chicken Cx43 and Cx45) and IPS 3. The experimental determination of both _j values for rat Cx43 are illustrated in Fig 2. Dual whole-cell current traces from two different rat Cx43–transfected N2A cell pairs are illustrated in Fig 2A and 2D. Channel activity was recorded at each V_j for 2 minutes, and junctional channel currents appear as simultaneous signals of equal amplitude and opposite polarity in the whole-cell currents of the prejunctional cell (pulsed, I₁) and the postjunctional cell (nonpulsed, I₂). Single-channel current amplitudes were determined for each recording by fitting the all-points current amplitude histogram (dots) with a pdf (solid line) that estimates the amplitude, variance, and open-state probability of every channel open-closed event (Fig 2B and 2E). These procedures were repeated at several V_j values, and a single-channel I_j-V_j relation was constructed for all rat Cx43 cell pairs examined (Fig 2C and 2F). Slope conductances of each I_j-V_j plot were used to determine the _j value of the rat Cx43 channel in IPS 2 and IPS 3. The open probabilities and event counts for each rat Cx43 channel in IPS 2 (experiments 1 to 4) and IPS 3 (experiments 5 and 6) are given in Table 2. In some cases, distinct open-channel peaks in the all-points current-amplitude histogram were readily observed for amplitude and event count determinations, but independent channel pdfs could not be obtained (because of either substate activity or cooperativity among multiple channels). Event counts for chicken Cx43 are also provided in Table 3.

View larger version (51K): [in this window] [in a new window]   Figure 2. A, Whole-cell currents from a rat Cx43 (rCx43)–transfected N2A cell pair during a -75-mV voltage step applied to cell 1 from a common holding potential of 0 mV, resulting in Vj of -75 mV. Ijs appear as equal amplitude and opposite polarity signals. The gaps in both current traces correspond to the omission of 16- and 71-second intervals, during which the channel remained open or closed, respectively. Three channels were open initially, and two closed within the first 2 seconds of the 260-second Vj pulse. Only two single-channel openings occurred during the remainder of the Vj pulse. B, All-points amplitude histogram compiled from the negative of the cell-2 current trace (-I2=Ij25 ) after low-pass filtering at 100 Hz and digitized at 1 kHz. A pdf (solid line) depicts a single open-channel current of -4.72 pA (63 pS) with a variance of 0.45 pA and an open probability of 0.055. C, Ij-Vj relation of the rCx43 channel in the presence of IPS 2. The Ij amplitudes were determined from histogram fits at the indicated Vj from four different cell pairs (different symbols). Each Vj pulse had a minimum duration of 2 minutes. The linear I-V relation has a slope conductance of 57 pS (r=.99). D, Similar equal amplitude and opposite polarity Ij signals from another rat Cx43 cell pair in the presence of IPS 3. Only one channel was observed during the 220-second -45-mV Vj pulse. E, All-points current amplitude histogram and pdf with an amplitude of -3.74 pA (83 pS), variance of 0.22 pA, and open probability of .948. There were 26 channel events. F, Ij-Vj relation taken from two cell pairs in the presence of IPS 3. The rat Cx43 I-V relation in KCl has a slope conductance of 80 pS (r=.99).

View this table: [in this window] [in a new window]   Table 3. Open Probabilities and Event Counts for All Chicken Cx43 Channel Recordings

The experimental IPS 3–to–IPS 2 _j ratio (80 pS/57 pS) for the rat Cx43 channel of 1.40 closely approximates the theoretical IPS 3–to–IPS 2 _j ratio of 1.36 and yields an R_p of 1.17. An R_p value close to 1.0 is indicative of a nonselective aqueous pore. These procedures were followed for all five distinct connexin channels examined, and the maximum _j values in each internal pipette solution, experimental _j ratio, and calculated R_p value are listed in Table 4. The maximum _j values and the corresponding R_p values, for five distinct connexin channels expressed in N2A cells, varied by more than one order of magnitude. Four of the connexins had R_p values of <1. Cx45, with the lowest _j, was the most selective (R_p, 0.12 [or cation selectivity, 8.3]), as might be expected if the low _j value were due to a narrow pore. In contrast, Cx37, with the highest _j, possessed an intermediate R_p of 0.38. Only the maximum _j state of each connexin channel was used for the R_p estimates, since any substate activity observed for a particular connexin was too infrequent to provide reliable I_j-V_j curves for slope conductance determinations, and any variations in conductance with ion substitution are within the noise limits (variance, 0.4 pA) of the recordings.

View this table: [in this window] [in a new window]   Table 4. Conductance and Permeability Ratios of Connexin-Specific Channels

To confirm the conductance ratio method for determining relative selectivities, we performed asymmetrical potassium glutamate/KCl experiments on rat Cx43 cell pairs with 1.2 mmol/L KCl or potassium glutamate added to IPS 2 or IPS 3, respectively. Single-channel recordings were obtained by using the same procedures described for Fig 2, and the composite I_j-V_j relation from two experiments are shown in Fig 3. When a negative V_j pulse is applied to the IPS 3 side, an I_j that consists primarily of a K⁺ influx from the partner cell and a Cl^- efflux from the KCl-containing cell is initiated. Conversely, when a positive V_j pulse is applied to the IPS 3 side, a glutamate^- influx and K⁺ efflux result. Linear regression analysis of the data under conditions favoring KCl I_js (negative V_j pulses applied to the IPS 3–containing cell or positive V_j pulses applied to the IPS 2–containing cell) produced a linear I_j-V_j relation with a slope of 93 pS. The corresponding linear I_j-V_j relation data under conditions favoring potassium glutamate^- I_js had a slope of 64 pS. The individual _j values for the two cell pairs were 92.6 and 92.8 pS or 64.1 and 66.3 pS under KCl or potassium glutamate I_j conditions (mean _j values, 92.7±0.1 or 65.2±1.1 pS, respectively). The IPS 3–to–IPS 2 conductance ratio for the asymmetrical solutions is 1.42, which corresponds with an R_p value of 1.34 pS and is in close agreement with the R_p estimate derived from independent experiments.

View larger version (26K): [in this window] [in a new window]   Figure 3. IPS 3–to–IPS 2 rCx43 conductance ratio in asymmetric solutions. Single-channel Ij-Vj relation is shown for all two-cell pairs using asymmetrical 120 mmol/L KCl/potassium glutamate (Kg) internal pipette solutions with 1.2 mmol/L of the opposite salt added to the other electrode. Bath compensation of the diffusion potential between the two electrodes was performed before gigaohm seal formation and averaged 28±1 mV, which closely approximates the predicted reversal potential of 29 mV when the bi-ionic potential equation is used (with the respective Cl- and glutamate- permeabilities set equal to their aqueous mobilities). Each symbol type refers to a different cell pair. The open symbols represent channel currents obtained when pulsing the IPS 3–containing cell, and the closed symbols represent the channel currents obtained when pulsing the IPS 2–containing cell. Linear regression fits of the channel amplitudes were obtained for the Vj pulses when Cl- or glutamate- was the permeant anion, as defined in the legend, and the corresponding slope conductances were 93 and 64 pS (r>.99).

These results are not consistent with the simple aqueous pore model of a gap junction channel for two reasons. First, the maximum _j values for rat Cx40, chicken Cx43, and human Cx37 in IPS 2 would require pore diameters of 18, 21, and 24 Å, respectively (see Fig 1B). These pore diameters exceed all estimates of pore size based on previous permeability studies and are not consistent with the selectivity observed (0.29R_p0.43) for these three connexins. Second, even though the maximum _j values and the R_p values varied by more than one order of magnitude among the five connexins studied, there was no correlation between these two parameters of the permeation pathway (Fig 4). The lack of correlation between conductance and charge selectivity demonstrates that form and function of the intercellular pathways formed by connexins have greater diversity than previously recognized.

View larger version (21K): [in this window] [in a new window]   Figure 4. Semilog plot of Rp and channel conductance of five connexin channels. The log of the Rp value for each connexin channel examined (chicken [c], rat [r], and human [h]) in Table 2 was plotted as a function of maximum j in the potassium glutamate internal pipette solution (IPS 2). For comparison, the scale for the corresponding cation-to-anion selectivity ratio is given on the right axis.

Previous fluorescent tracer studies using neutral or anionic dyes, from which the molecular permeability limit of 1 kD was derived,^{2 4 5} likely reflect the permeability properties of Cx43, since this is the most abundantly expressed connexin in mammalian cells and tissues.^{1 28} To further examine the permeability properties of the five disparate connexin channels in this investigation, 2 mmol/L 6-CF or diCl-F was added to one recording pipette in the double whole-cell recording configuration. These two fluorescent dyes are both negatively charged molecules with an unhydrated width of 10 Å yet dramatically different junctional permeabilities.⁵ The molecular structure and calculated 1.5 kT electrostatic charge surfaces of both dyes are shown in Fig 5. These two fluorescein derivatives vary only by substitution of two chlorides or an additional carboxyl group for hydrogens at the indicated positions of the fluorescein molecule (at pH 7.0, valences are -2 for 6-CF and -1 for diCl-F), yet the junctional permeability of diCl-F is 20-fold greater than that of 6-CF.⁵ The electrostatic surface charge profile of 6-CF exhibits a dramatically enlarged anionic charge with a shape not closely resembling the molecular backbone, increasing the polarity and hydration of this molecule. The difference in junctional permeabilities of 6-CF and diCl-F can likely be attributed to the dramatically different negative charge surfaces.

View larger version (13K): [in this window] [in a new window]   Figure 5. Structures and surface charge distribution of fluorescein derivatives. A, Structure of 6-CF. B, Structure of diCl-F. C, Calculated potential surface of 6-CF at 1.5 kT. The positive (blue) and negative (red) electrostatic potential surfaces at 1.5 kT are shown. Calculations were carried out by using the finite difference solution of the Poisson-Boltzmann equation using the Delphi module of INSIGHT II, version 2.3 (Biosym Corp). Charges were assigned according to ionization at pH 7.0, and the potentials were calculated for a medium of 140 mmol/L salt and a dielectric of 80. D, Calculated potential surface of diCl-F at 1.5 kT. Calculations were carried out as described for panel C for the 1.5-kT profile of diCl-F.i

The moderate differences in physical structure and very disparate electrostatic surface charge distribution between 6-CF and diCl-F make these two fluorescein derivatives useful probes for distinguishing between the physical and electrostatic determinants of pore permeability. Fig 6 illustrates the successful passage of 6-CF and diCl-F in two rat Cx43 cell pairs with similar g_j values of 0.5 nS. The lowest g_j rat Cx43 cell pairs are shown to illustrate the minimum conditions for successful 6-CF and diCl-F transfer observed with this connexin. Higher g_j rat Cx43 cell pairs exhibited more pronounced dye transfer.

View larger version (110K): [in this window] [in a new window]   Figure 6. Dye transfer in rat Cx43 N2A cell pairs. A, Phase-contrast micrograph of a rat Cx43-transfected N2A cell pair in the double whole-cell recording configuration containing IPS 2. gj was determined by applying 5-second -40-mV Vj pulses to cell 1 (2 mmol/L 6-CF dye-loaded cell [*]) at a rate of four pulses per minute. gj was monitored for 10 minutes as cells were dye-loaded, and dye transfer was immediately assessed under epifluorescent illumination.22 23 This pair had a measured gj of 0.5 nS for a 10-minute recording period. B, Epifluorescent micrograph of the same cell pair immediately after cessation of gj recording and removal of the patch electrodes to optimize the fluorescent image of the dye-filled cells. Dye transfer was evident in both cells before and after removal of both patch electrodes with a distinct dye diffusion gradient. C, Phase-contrast micrograph of a second rat Cx43-transfected N2A cell pair in the double whole-cell recording configuration using IPS 2. Cell 1 (*) contained 2 mmol/L diCl-F. gj of this cell pair was also 0.5 nS. D, Epifluorescent illumination of the same cell pair immediately after cessation of the 10-minute gj recording period. As with 6-CF, a distinct dye diffusion gradient is evident between both cells.

The 6-CF and diCl-F dye permeability results for all five connexins are summarized in Table 5. Only rat Cx43 (R_p>1, anion>cation permeability) exhibited successful dye transfer with both dyes at g_j of <1 nS, consistent with an R_p of 1. Chicken Cx43, with an R_p of 0.33 to 0.43, was also permeable to both dyes, although 6-CF dye transfer occurred only in the highest g_j cell pairs, whereas diCl-F transfer occurred with g_j as low as 0.2 nS. Human Cx37, the highest known _j connexin channel, has an R_p value (0.38) similar to that of chicken Cx43, yet dye transfer with both dyes was sporadic and inconsistent at all observed g_j values. Chicken Cx45 (R_p, 0.12) did not exhibit any permeability to 6-CF even with g_j of 30 nS (900 open channels) but was permeable to diCl-F. These data indicate that many connexin channels are able to discriminate between molecules of similar dimension (10 Å) because the differences in surface charge distribution are in a manner generally consistent with their measured R_p values. Interestingly, chicken Cx45 and rat Cx40 have the lowest R_p values, but Cx45 has a smaller _j and a higher diCl-F permeability than does Cx40 or Cx37. This suggests that factors other than the charge selectivity (open-state probability, permeability and open probability of any subconductance states, and transit time for large permeant molecules relative to mean channel open time) reflected in the R_p values can contribute significantly to selectivity among larger permeant molecules.

View this table: [in this window] [in a new window]   Table 5. Permeability Limits of Connexin-Specific Channels to Fluorescein Derivatives

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

 
In the present investigation, we have challenged the presently accepted simple right cylindrical pore model of a gap junction channel by examining the key assumption that channel conductance is directly related to pore diameter. It has often been inferred that gap junction channel ionic selectivity is negligible and that molecular permeability limits are high (hundreds of daltons in molecular mass and 8 Å in diameter) owing to the large diameter of the pore. The results of this investigation indicate that there is no direct relation between connexin channel conductance and ionic selectivity or dye permeability. The variability of the ionic selectivities and dye permeabilities with channel conductance suggests that there is considerably more diversity among the connexin pores than previously acknowledged. Since the greatest anion-to-cation selectivity is only 8:1, we also propose that weak electrostatic potentials associated with the pore of each connexin channel can account for the range of ionic selectivities observed in the present investigation. This updated interpretation of connexin-specific gap junction channels provides additional information that questions previous predictions about the pore-forming region of the connexins, which rest on the validity of the KCl/potassium glutamate ion substitutions as an indicator of the relative channel anion-to-cation selectivity.

The validity of the experimental-to-theoretical conductance ratio estimates of anion-to-cation selectivity is based on the differences between the observed conductance changes and the predicted changes for a nonselective aqueous pore. Hence, the accuracy of the R_p values depends principally on the chloride/glutamate mobilities. Whether the derived aqueous mobility for glutamate is applicable to its permeation through the various connexin channels is best addressed by the 6-CF and diCl-F permeabilities through each connexin channel. Any connexin channel that readily exhibits dye transfer as summarized in Table 5 must have a finite permeability to glutamate, since its valence is equal to diCl-F, whereas its molecular mass and dimensions are less than that of diCl-F. Still, the mobility of glutamate in the pore could be reduced relative to its aqueous mobility. However, reducing the glutamate permeability term to zero only affects the theoretical IPS 2 conductance measurement in a manner that requires even higher channel cationic conductances to match the experimental IPS 2 conductance value. Thus, if glutamate permeability is zero, all channels have a lower R_p value, and the present estimates provide a maximum anion-to-cation selectivity estimate for each connexin channel. R_p values can be revised for the glutamate-impermeable condition by substituting a theoretical IPS 3–to–IPS 2 _j ratio of 1.72 for 1.36 in Table 4 and recalculating accordingly (see "Materials and Methods"). The greatest impact would be on the rat Cx43 R_p value, which would become <1 if glutamate is assumed to be impermeant. However, this connexin channel has the highest relative permeabilities to 6-CF and diCl-F (Table 5) of the five connexins examined, which is not consistent with the latter assumption.

Other factors besides R_p may influence dye transfer, including channel open probability and gating to multiple open configurations of unknown size and selectivity relative to the main open state. With increasing V_j, the maximum conductance state of many connexin channels undergoes a decrease in open probability and often shifts to a higher open probability subconductance state (eg, Cx37).²² The low incidence of dye transfer with both dyes is consistent with the Cx37 subconductance state not being dye permeable, since the substate is the dominant conducting state at V_j of ±30 mV.²² The Cx37 channel is unique among the connexins we have examined to date because of the high open probability for a subconductance state and concomitant low maximum conductance state open probability even at low V_j values (30 mV). Subconductance state behavior has been observed for chicken Cx45²³ and rat Cx40. Initial estimates of open probability for the various connexins gave values ranging from .2 to .99 at V_j <±30 mV, although accurate single-channel open probability estimates require several hundred channel events at each V_j, which was not accomplished here because of the many connexins under investigation. In all cases, a decline in open probability was observed with increased V_j, consistent with the normalized macroscopic g_j-V_j relations for the respective connexins.^{17 19 29}

These data suggest that electrostatic field effects associated with charged amino acids within or near the connexin pore can alter the selectivity and conductance of the connexin-specific channels. It is possible that multiple conductance states of a given connexin channel can confer ionic and molecular permeabilities that are distinct from the maximum conductance state of the channel, although this point remains to be definitively demonstrated. This hypothesis is analogous to the cation selectivity and Mg²⁺-dependent block of the nicotinic acetylcholine receptor channel that is imparted by the presence of three rings of acidic amino acids and a central uncharged ring of amino acids located at the narrowest portion of the pore.^{30 31} The pore-lining sequence of the connexins remains to be determined through rigorous structure-function investigations of connexin channel permeability or conductance changes associated with site-directed mutagenesis, as has been performed on the P-segment of the ion-selective voltage-gated Na⁺, Ca²⁺, and K⁺ channels or on the M2 domain of the superfamily of ligand-gated channels.^{32 33} The favored hypothesis for the pore-forming segment of the connexins is that the third transmembrane domain, M3, lines the pore, since it is the most amphipathic helix predicted from the primary amino acid sequences of all known connexins.^{34 35} However, it should be noted that similar predictions of pore-forming amphipathic -helical transmembrane domains such as S4 of the Na⁺ channel and MA of the nicotinic acetylcholine receptor channel proved to be incorrect after rigorous structure-function analysis.^{36 37} The present study of five connexin channels provides the initial permeability characteristics necessary to perform detailed structure-function analyses of the pore-forming segments of the different connexins. For instance, chicken Cx43 has a maximum _j that is 2.5 times higher than the rat Cx43 channel, yet chicken Cx43 exhibits a twofold greater selectivity for cations (R_p, 0.50) than does rat Cx43. Preliminary comparisons of the primary amino acid sequences reveal 27 amino acid substitutions (overall identity, 93%) with only two involving (acidic-basic) charge differences and a single conservative I-V substitution in M4 being the only sequence variation in any of the transmembrane domains. Structure-function analyses of these sequence variations are in progress. One major difference in the tissue distributions is that chicken Cx42, not Cx43, is the predominant connexin in avian adult ventricular myocardium.³⁸ Additional ionic selectivity experiments are presently being performed in this laboratory to further define the cationic and anionic selectivities of selected connexin channels.

Our data have several implications regarding ionic and chemical signaling through gap junction channels composed of different connexins. First, the permeability properties of a gap junction channel are specific to the connexins expressed in a particular cell or tissue. Second, heterotypic channels (ie, formed by unlike hemichannels) can produce asymmetric junctional permeabilities if the two connexins possess opposing R_p values (<1 and >1). For example, one would predict that a negative voltage pulse applied to the cell containing a cation-selective hemichannel will attempt to drive anions through that hemichannel, whereas the countercurrent cationic flow will originate from the cell containing the anion-selective hemichannel, thus producing low current flows in either direction. Conversely, when a positive voltage is applied to the cation-selective hemichannel, a cationic flux is initiated from this cell, whereas the anionic flux will originate from the anion-selective hemichannel. Hence, depending on the voltage polarity between the two cells, ion entry into both sides of the gap junction channel pore is inhibited or favored as indicated by the hemichannel R_p values and can result in rectifying I-V relations. However, it should be noted that permeability can never be unidirectional (ie, irreversible). Apparent unidirectional transfer of fluorescent tracers like lucifer yellow, which readily passes from astrocytes expressing Cx43 to oligodendrocytes expressing Cx32, has been observed between different cell types.^{39 40} A proposed mechanism for this unidirectional transfer was based solely on asymmetric pore diameters and _js. Our data indicate that _j and pore diameter or dye permeability are not correlated and that only mammalian Cx43 channels have a _j and nonselectivity expected for a simple aqueous pore. Opposite R_ps on each side of the channel, indicative of direct charge interactions between permeant molecules and fixed charges associated with the channel, in addition to channel gating and substate properties are major determinants of dye transfer. Third, since most second messengers (eg, cAMP and inositol tris-phosphate) are anionic and possess greater molecular mass and diameters than 6-CF or diCl-F, it is reasonable to assume that many of the connexin-specific channels will be less permeable (or impermeant) to physiologically relevant second messengers than rat Cx43. This diversity of connexin channel properties is a likely consequence of variations in primary amino acid sequence and the resulting structure of the pore-forming region of the connexins, which remain unidentified to date.

	Selected Abbreviations and Acronyms

 

j = unitary channel conductance I2 = difference in holding current of cell 2 6-CF = 6-carboxyfluorescein Cx32, Cx37, Cx40, Cx43, and Cx45 = connexin32, connexin37, connexin40, connexin43, and connexin45, respectively diCl-F = 2',7'-dichlorofluorescein gj = junctional conductance I-V = current-voltage Ij = junctional current IPS 1, 2, and 3 = internal pipette solutions 1, 2, and 3, respectively N2A = Neuro2A pdf = probability density function Rp = relative anion-to-cation selectivity ratio TEA = tetraethylammonium Vj = transjunctional voltage

	Acknowledgments

 
This study was supported by grants from the National Institutes of Health (to Drs Veenstra, Beyer, Brink, and Harris) and the Office of Naval Research (to Dr Harris). Drs Veenstra and Beyer are Established Investigators of the American Heart Association. We thank E.M. Westphale for her technical assistance in transfecting and maintaining the N2A cell cultures.

Received April 5, 1995; accepted August 10, 1995.

	References

Top Abstract Introduction Materials and Methods Results Discussion References

 

1. Beyer EC. Gap junctions. Int Rev Cytol. 1993;137:1-37.
2. Flagg-Newton J, Simpson I, Loewenstein WR. Permeability of the cell-to-cell membrane channels in mammalian cell junctions. Science. 1979;205:404-407. [Abstract/Free Full Text]
3. Makowski L, Caspar DLD, Phillips WC, Goodenough DA. Gap junction structures, V: structural chemistry inferred from X-ray diffraction measurements on sucrose accessibility and trypsin susceptibility. J Mol Biol. 1984;174:449-481. [Medline] [Order article via Infotrieve]
4. Imanaga I, Kameyama M, Irisawa H. Cell-to-cell diffusion of fluorescent dyes in paired ventricular cells. Am J Physiol. 1987:252;H223-H232.
5. Brink PR, Dewey MM. Evidence for fixed charge in the nexus. Nature. 1980;285:101-102. [Medline] [Order article via Infotrieve]
6. Tsien RW, Weingart R. Inotropic effect of cyclic AMP in calf ventricular muscle studied by a cut end method. J Physiol (Lond). 1976;260:117-141. [Abstract/Free Full Text]
7. Lawrence TS, Beers WH, Gilula NB. Transmission of hormonal stimulation by cell-to-cell communication. Nature. 1978;272:501-506.[Medline] [Order article via Infotrieve]
8. Dunlap K, Takeda K, Brehm P. Activation of a calcium-dependent photoprotein by chemical signalling through gap junctions. Nature. 1987;325:60-62. [Medline] [Order article via Infotrieve]
9. Saez J, Connor JA, Spray DC, Bennett MVL. Hepatocyte gap junctions are permeable to the second messenger, inositol 1,4,5-trisphosphate, and to calcium ions. Proc Natl Acad Sci U S A. 1989;86:2708-2712. [Abstract/Free Full Text]
10. Hobbie L, Kingsley DM, Kozarsky KF, Jackman RW, Krieger M. Restoration of LDL receptor activity in mutant cells by intercellular junctional communication. Science. 1989:235:69-73.
11. Brink PR. Gap junction channels and cell-to-cell messengers in myocardium. J Cardiovasc Electrophysiol. 1991;2:360-366.
12. Mehta PP, Bertram JS, Loewenstein WR. Growth inhibition of transformed cells correlates with their junctional communication with normal cells. Cell. 1986:44:187-196.
13. Warner A. The gap junction. J Cell Sci. 1988;89:1-7. [Free Full Text]
14. Guthrie SC, Gilula NB. Gap junctional communication and development. Trends Neurosci. 1989;12:12-16. [Medline] [Order article via Infotrieve]
15. Neyton J, Trautmann A. Single channel currents of an intercellular junction. Nature. 1985;317:331-335. [Medline] [Order article via Infotrieve]
16. Brink PR, Fan S-F. Patch clamp recordings from membranes which contain gap junction channels. Biophys J. 1989;56:579-593. [Abstract/Free Full Text]
17. Veenstra RD, Wang H-Z, Westphale EM, Beyer EC. Multiple connexins confer distinct regulatory and conductance properties of gap junctions in developing heart. Circ Res. 1992;71:1277-1283. [Abstract/Free Full Text]
18. Reed KE, Westphale EM, Larson DM, Wang H-Z, Veenstra RD, Beyer EC. Molecular cloning and functional expression of human connexin37, an endothelial cell gap junction protein. J Clin Invest. 1993;91:997-1004.
19. Chen, DeHaan RL. Multiple-channel conductance states and voltage regulation of embryonic chick cardiac gap junctions. J Membr Biol. 1993;136:125-134. [Medline] [Order article via Infotrieve]
20. Takens-Kwak BR, Jongsma HJ. Cardiac gap junctions: three distinct single channel conductances and their modulation by phosphorylating treatments. Pflugers Arch. 1992:422;198-200.
21. Moreno AP, Sáez JC, Fishman GI, Spray DC. Human connexin43 gap junction channels: regulation of unitary conductances by phosphorylation. Circ Res. 1994;74:1050-1057. [Abstract/Free Full Text]
22. Veenstra RD, Wang H-Z, Beyer EC, Ramanan SV, Brink PR. Connexin37 forms high conductance gap junction channels with subconductance state activity and selective dye and ionic permeabilities. Biophys J. 1994;66:1915-1928. [Abstract/Free Full Text]
23. Veenstra RD, Wang H-Z, Beyer EC, Brink PR. Selective dye and ionic permeability of gap junction channels formed by connexin45. Circ Res. 1994;75:483-490. [Abstract/Free Full Text]
24. Hille B. Ionic Channels of Excitable Membranes. 2nd ed. Sunderland, Mass: Sinauer Associates Inc; 1992;268:337-360.
25. Veenstra RD, Brink PR. Patch clamp analysis of gap junction channel currents. In: Stevenson B, Paul DL, Gallin W, eds. Cell-Cell Interactions: A Practical Approach. Oxford, UK: IRL Press; 1992:167-201.
26. Manivannan K, Ramanan SV, Mathias RT, Brink PR. Multichannel recordings from membranes which contain gap junctions. Biophys J.. 1992;61:216-227. [Abstract/Free Full Text]
27. 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. [Abstract/Free Full Text]
28. Beyer EC, Paul DL, Goodenough DA. Molecular cloning of cDNA for rat liver gap junction protein. J Cell Biol. 1987;105:2621-2629. [Abstract/Free Full Text]
29. Wang H-Z, Li J, Lemanski LF, Veenstra RD. Gating of mammalian cardiac gap junction channels by transjunctional voltage. Biophys J. 1992;63:139-151. [Abstract/Free Full Text]
30. Imoto K, Busch C, Sakmann B, Mishina M, Konno T, Nakai J, Bujo H, Mori Y, Fukuda K, Numa S. Rings of negatively charged amino acids determine the acetylcholine receptor channel conductance. Nature. 1988;335:645-648. [Medline] [Order article via Infotrieve]
31. Revah F, Galzi J-L, Giraudat P-Y, Lederer F, Changeux J-P. The noncompetitive blocker [3H]chlorpromazine labels three amino acids of the acetylcholine receptor g subunit: implications for the a-helical organization of regions MII and for the structure of the ion channel. Proc Natl Acad Sci U S A. 1990;87:4675-4679. [Abstract/Free Full Text]
32. Tomaselli GF, Backx PH, Marban E. Molecular basis of permeation in voltage-gated ion channels. Circ Res. 1993;72:491-496. [Free Full Text]
33. Sakurada K, Masu M, Nakanishi S. Alteration of Ca²⁺ permeability and sensitivity to Mg²⁺ and channel blockers by a single amino acid substitution in the N-methyl-D-aspartate receptor. J Biol Chem. 1993;268:410-415. [Abstract/Free Full Text]
34. Milks LC, Kumar NM, Houghten R, Unwin N, Gilula NB. Topology of the 32-kd liver gap junction protein determined by site-directed antibody localizations. EMBO J. 1988;7:2967-2975. [Medline] [Order article via Infotrieve]
35. Bennett MVL, Barrio LC, Bargiello TA, Spray DC, Hertzberg E, Saez JC. Gap junctions: new tools, new answers, new questions. Neuron. 1991;6:305-320. [Medline] [Order article via Infotrieve]
36. Noda M, Shimizu S, Tanabe T, Takai T, Kayano T, Ikeda T, Takahashi H, Nakayama H, Kanaoka Y, Minamino N, Kangawa K, Matsuo H, Raftery MA, Hirose T, Inayama S, Hayashida H, Miyata T, Numa S. Primary structure of Electrophorus electricus sodium channel deduced from cDNA sequence. Nature. 1984;312:121-127. [Medline] [Order article via Infotrieve]
37. Mishina M, Tobimatsu T, Imoto K, Tanaka K, Fujita Y, Fukuda K, Kurasaki M, Takahashi H, Morimoto Y, Hirose T, Inayama S, Takahashi T, Kuno M, Numa S. Location of functional regions of acetylcholine receptor -subunit by site-directed mutagenesis. Nature. 1985;313:364-369. [Medline] [Order article via Infotrieve]
38. Minkoff R, Rundus VR, Parker SB, Beyer EC, Hertzberg EL. Connexin expression in the developing avian cardiovascular system. Circ Res. 1993;73:71-78. [Abstract]
39. Robinson SR, Hampson ECGM, Munro MN, Vaney DI. Unidirectional coupling of gap junctions between neuroglia. Science. 1993;262:1072-1074. [Abstract/Free Full Text]
40. Flagg-Newton JL, Loewenstein WR. Asymmetrically permeable membrane channels in cell junction. Science. 1980;207:771-773.[Abstract/Free Full Text]

This article has been cited by other articles:

				G. Kanaporis, G. Mese, L. Valiuniene, T. W. White, P. R. Brink, and V. Valiunas Gap Junction Channels Exhibit Connexin-specific Permeability to Cyclic Nucleotides J. Gen. Physiol., March 31, 2008; 131(4): 293 - 305. [Abstract] [Full Text] [PDF]

				N. S. Heyman and J. M. Burt Hindered Diffusion through an Aqueous Pore Describes Invariant Dye Selectivity of Cx43 Junctions Biophys. J., February 1, 2008; 94(3): 840 - 854. [Abstract] [Full Text] [PDF]

				T. Y. Li, D. Colley, K. J. Barr, S.-P. Yee, and G. M. Kidder Rescue of oogenesis in Cx37-null mutant mice by oocyte-specific replacement with Cx43 J. Cell Sci., December 1, 2007; 120(23): 4117 - 4125. [Abstract] [Full Text] [PDF]

				J. C. B. Jacobsen, C. Aalkjaer, H. Nilsson, V. V. Matchkov, J. Freiberg, and N.-H. Holstein-Rathlou A model of smooth muscle cell synchronization in the arterial wall Am J Physiol Heart Circ Physiol, July 1, 2007; 293(1): H229 - H237. [Abstract] [Full Text] [PDF]

				V. V. Matchkov, H. Gustafsson, A. Rahman, D. M. Briggs Boedtkjer, S. Gorintin, A. K. Hansen, E. V. Bouzinova, H. A. Praetorius, C. Aalkjaer, and H. Nilsson Interaction Between Na+/K+-Pump and Na+/Ca2+-Exchanger Modulates Intercellular Communication Circ. Res., April 13, 2007; 100(7): 1026 - 1035. [Abstract] [Full Text] [PDF]

				M. Wang, V. M. Berthoud, and E. C. Beyer Connexin43 increases the sensitivity of prostate cancer cells to TNF{alpha}-induced apoptosis J. Cell Sci., January 15, 2007; 120(2): 320 - 329. [Abstract] [Full Text] [PDF]

				K Boengler, R Schulz, and G Heusch Connexin 43 signalling and cardioprotection Heart, December 1, 2006; 92(12): 1724 - 1727. [Abstract] [Full Text] [PDF]

				L. Dong, X. Liu, H. Li, B. M. Vertel, and L. Ebihara Role of the N-terminus in permeability of chicken connexin45.6 gap junctional channels J. Physiol., November 1, 2006; 576(3): 787 - 799. [Abstract] [Full Text] [PDF]

X. F. Figueroa, B. E. Isakson, and B. R. Duling Vascular Gap Junctions in Hyper