The effect of CO2-induced acidification on transjunctional voltage (Vj) gating was studied by dual voltage-clamp in oocytes expressing mouse connexin 50 (Cx50) or a Cx50 mutant (Cx50-D3N), in which the third residue, aspartate (D), was mutated to asparagine (N). This mutation inverted the gating polarity of Cx50 from positive to negative. CO2 application greatly decreased the Vj sensitivity of Cx50 channels, and increased that of Cx50-D3N channels. CO2 also affected the kinetics of Vj dependent inactivation of junctional current (Ij), decreasing the gating speed of Cx50 channels and increasing that of Cx50-D3N channels. In addition, the D3N mutation increased the CO2 sensitivity of chemical gating such that even CO2 concentrations as low as 2.5% significantly lowered junctional conductance (Gj). With Cx50 channels Gj dropped by 78% with a drop in intracellular pH (pHi) to 6.83, whereas with Cx50-D3N channels Gj dropped by 95% with a drop in pHi to just 7.19. We have previously hypothesized that the way in which Vj gating reacts to CO2 might be related to connexin’s gating polarity. This hypothesis is confirmed here by evidence that the D3N mutation inverts the gating polarity as well as the effect of CO2 on Vj gating sensitivity and speed.
- cell communication
- gap junctions
- chemical gating
- channel gating
- Xenopus oocytes
gap junctions are regions of cell contact that mediate the exchange of small cytosolic molecules via cell-cell channels. A gap junction channel results from the extracellular interaction of two hemichannels (connexons), each a hexamer of connexin proteins. Connexins (Cx) contain four transmembrane domains, two extracellular loops, a cytoplasmic loop, a short NH2 terminus (NT), and a COOH terminus of variable length. The sequences of cytoplasmic loop and COOH terminus vary significantly among the members of the connexin family, whereas those of the other domains are relatively well conserved (see Ref. 14 for review).
Gap junction channels are gated by transjunctional voltage (Vj; 27) and increased intracellular [Ca2+] (10, 25) or [H+]i (28, 33) via molecular mechanisms that are still poorly defined (see Ref. 12 for review). At least two Vj-sensitive gates have been identified: fast and slow. On the basis of their behavior at the single channel level, fast Vj gate and chemical gate are believed to be distinct: the former closes rapidly (<1 ms) but incompletely, leaving a 20–30% residual conductance, whereas the latter closes slowly (8–10 ms) but completely (5). In contrast, slow Vj gate and chemical gate are likely to be the same (5, 17, 20). Slow and fast Vj gates are in series, and each hemichannel appears to have both gates. The slow gate closes at the negative side of Vj in all connexin channels tested, whereas the polarity of the fast Vj gating mechanism varies among connexin channels (see Ref. 8 for review).
In the past, chemical and Vj gating have been studied almost exclusively by testing chemicals and Vj gradients, respectively, whereas minimal attention has been devoted to potential effects of voltage on chemical gating or chemicals on voltage gating. In the early 1990s, CO2 application was reported to increase the Vj sensitivity of Cx32 channels (38). More recently chemical gating induced by CO2 was shown to be reversed by Vj gradients positive at the mutant side of heterotypic channels between Cx32 and various Cx32 mutants (17, 20), and chemical gating was reversed in insect cells by bilateral hyperpolarization (37). These observations suggest that chemical and voltage gates may be sensitive to Vj and CO2, respectively.
We have recently reported that the speed and sensitivity of Vj-dependent inactivation of junctional current (Ij) are increased by CO2 application in both Cx45 (21) and Cx32 (22, 41, 42) channels. Significantly, however, the effect of CO2-induced acidification on Vj gating differs among connexin channels, as CO2 decreases the Vj sensitivity of Cx40 (13), Cx26 (22, 42), Cx50 (present study), Cx37 (C. Peracchia, unpublished observations), and Cx38 (38) channels. This suggests that there are two distinct classes of connexin channels whose Vj gating sensitivity reacts in opposite ways to CO2-induced acidification.
As a hypothesis, we have proposed that the way in which the Vj gating of connexin channels responds to CO2 may be related to the polarity of Vj gating (13). Indeed, the fast Vj gates close at the positive side of Vj (positive gaters) in Cx26, -37, -38, -40, and -50 channels, and at the negative side in Cx32 and Cx45 channels (see Ref. 8 for review). To test this hypothesis, the present study has monitored the CO2 sensitivity of Vj gating in Cx50 channels (positive gaters; 4, 8, 35) and in channels made of a Cx50 mutant (Cx50-D3N) in which the third residue, aspartate (D), is replaced with asparagine (N). Cx50 is a connexin expressed in eye lens fibers whose high gating sensitivity to cytosolic acidification (9, 30, 40) has been proposed to play a role in the function of differentiating fibers (2).
The data show that CO2 decreases the Vj sensitivity of Cx50 channels, and that the point mutation D3N, which neutralizes the negative charge in third position, reverses both the Vj gating polarity of Cx50 channels (Cx50-D3N channels are negative gaters) and their Vj sensitivity to CO2. A preliminary account of this study has been published in abstract form (15).
MATERIALS AND METHODS
Oocyte Preparation and Microinjection
Oocytes were prepared as previously described (18). Briefly, adult female Xenopus laevis frogs were anesthetized with 0.3% tricaine (MS-222) and the oocytes were surgically removed from the abdominal incision. The oocytes were placed in ND96 medium containing (in mM) 96 NaCl, 2 KCl, 1.8 CaCl2, 1 MgCl2, and 5 HEPES (pH 7.6 with NaOH). Oocytes at stage V or VI were subsequently defolliculated in 2 mg/ml collagenase (Sigma, St. Louis, MO) for 80 min at room temperature in nominally Ca2+-free OR2 solution containing (in mM) 82.5 NaCl, 2 KCl, 1 MgCl2, and 5 HEPES (pH 7.6 with NaOH). The defolliculated oocytes were injected with 46 nl (0.25 μg/μl) of antisense oligonucleotide complementary to endogenous Xenopus Cx38: 5′-GCTTTAGTAATTCCCATCCTGCCATGTTTC-3′ (commencing at nt −5 of Cx38 cDNA sequence; Ref. 3) with a Nanoject apparatus (Drummond, Broomall, PA). The antisense oligonucleotide blocks completely the endogenous junctional communication within 48 h. From 24 to 72 h later, 46 nl of mouse Cx50 wild-type cRNA (∼0.4 μg/μl), or cRNA of a mouse Cx50 mutant in which the aspartate residue in position 3 was replaced with asparagine (Cx50-D3N), were injected into oocytes at the vegetal pole and the oocytes were incubated overnight at 18°C. The oocytes were mechanically stripped of their vitelline layer in hypertonic medium (18) and paired at the vegetal poles in conical wells of culture dishes (Falcon Products, Becton Dickinson Labware, Franklin Lakes, NJ) filled with ND96. All oocyte pairs were studied electrophysiologically 2–3 h after being paired.
Measurement of junctional conductance and uncoupling protocols.
The oocyte chamber was continuously perfused at a flow rate of 0.6 ml/min by a peristaltic pump (Dyamax model RP-1, Rainin Instrument, Woburn, MA). The superfusion solution was ejected by a 22-gauge needle placed near the edge of the conical well containing the oocyte pair. The level of the solution in the chamber was maintained constant by continuous suction. All of the experiments were performed using the standard double voltage-clamp procedure for measuring Gj (27). After the insertion of a current and a voltage microelectrode in each oocyte, both oocytes were individually clamped by two oocyte clamp amplifiers (model OC-725C, Warner Instrument, Hamden, CT) to the same holding potential, Vm1 = Vm2 (−20 mV), so that no junctional current would flow at rest (Ij = 0). For measuring junctional conductance (Gj) and CO2 sensitivity, a Vj gradient was created by imposing a voltage step (V1) to oocyte 1 while maintaining V2 at Vm; thus Vj = V1. The negative feedback current (I2), injected by the clamp amplifier in oocyte 2 for maintaining V2 constant at Vm, was used for calculating Gj, as it is identical in magnitude to Ij but of opposite sign (Ij = −I2); Gj = Ij/Vj. Pulse generation and data acquisition were performed with pCLAMP version 184.108.40.206 software (Axon Instruments, Foster City, CA) and a DigiData 1322A interface (Axon). Ij and Vj were measured with CLAMPfit (Axon), and the data were plotted with SigmaPlot (SPSS, Chicago, IL).
To test the effect of CO2 on Gj and Vj sensitivity, oocyte pairs were superfused for 10–80 min (0.6 ml/min) with ND96 solutions gassed with 2.5%, 5.0%, 10%, or 30% CO2, and Gj was measured by applying voltage steps of −60 mV (12-s duration) every 30 s to one oocyte, while maintaining the other oocyte at Vm. Gj peak (Gj,pk), Gj steady state (Gj,ss), and the ratio Gj,ss/Gj,pk were calculated in the presence and absence of CO2. For comparing Cx50 and Cx50-D3N channels in terms of Gj sensitivity to pHi, oocyte pairs were superfused for 30–40 min (0.6 ml/min) with ND96 solutions gassed with 2.5%, 5.0%, 10%, or 30% CO2 and Gj was measured by applying voltage steps of +20 mV (2 s duration) every 30 s to one oocyte while maintaining the other oocyte at Vm.
For studying voltage dependence of Gj in the presence and absence of CO2, a standard Vj protocol was used. Each oocyte was first voltage clamped at −20 mV. Voltage steps of −10 mV (100 mV Vj maximum) and 25-s duration were applied every 45 s to either oocyte of the pair, while maintaining the other at −20 mV. The voltage-insensitive junctional conductance (Gj,max) was calculated using Ij values elicited by Vj = −10 mV because at this Vj there is no Ij decay. To illustrate the relationship between Gj,ss and Vj, the ratio Gj,ss/Gj,max was plotted with respect to Vj. The curve was fitted to a two-state Boltzmann distribution of the form: (Gj,ss − Gj,min)/(Gjmax − Gj,ss) = exp[−A(Vj − V0)], where V0 is the Vj value at which Gj is one-half of the value of Gj,max − Gj,min, Gj,max is Gj at Vj = 0 mV and Gj,min is the theoretical minimum normalized Gj. A = ηq/kT is a constant expressing voltage sensitivity in terms of number of equivalent gating charges, η, moving through the entire applied field, where q is the electron charge, k is the Boltzmann constant, and T is the temperature in Kelvin. The time constants (τ) of Ij decay at Vj = 100 mV, for Cx50 channels, and Vj = 50 mV, for Cx50-D3N channels, were calculated by fitting each Ij curve to a two-term exponential function (τ1 and τ2), following baseline correction (CLAMPfit, Axon). Gj,ss was obtained from the exponential fit (parameter “C” of CLAMPfit, Axon).
Measurement of Intracellular pH
For testing the effect of different CO2 concentrations on intracellular pH (pHi), oocytes were superfused for 40 min (0.6 ml/min) with ND96 solutions gassed with 2.5%, 5.0%, 10%, or 30% CO2; these CO2 concentrations reduce the pH of ND96 from 7.6 to 7.4, 6.85, 6.26, and 5.75, respectively. pHi measurements were performed with the Dual-Wavelength Fluorescence Imaging and Photometry Systems (InCyt; Intracellular Imaging, Cincinnati, OH), as previously described (16). The fluorescein derivative pH indicator BCECF (model B-1151, Molecular Probes, Eugene, OR) was injected into oocytes in amounts sufficient to reach an intra-oocyte concentration of ∼100 μM. After the injection, the oocytes were placed in the same conical wells used for electrophysiology, modified by replacement of the plastic floor with a glass coverslip. The conical wells were mounted on the stage of a Nikon TMS microscope equipped for epifluorescence, and the oocytes were superfused with ND96 gassed with different CO2 concentrations (see above). Specimen observation and light measurements were performed with a Nikon Fluor ×10 objective. Light from a 300-W xenon arc illuminator passed through a computer-controlled filter changer and shutter unit, containing 440- and 490-nm band-pass filters and liquid light guide. Light emitted by the oocytes (focusing on their bottom surface, closest to the glass coverslip), was collected by a cooled charge-coupled device video camera (model 4922-2010; COHU, San Diego, CA). Pairs of images at the two wavelengths were collected in rapid succession, and [H+]i was computer calculated online (model GP7-450, Gateway, N. Sioux City, SD) by dividing the short wavelength by the long wavelength image, after subtraction of the respective backgrounds. Calibration curves were generated by rationing droplets of 0.1 M phosphate buffers (pH 8.0, 7.5, 7.0, 6.5, 6.0) containing 13 μM BCECF.
Homotypic Cx50 Channels
Gj sensitivity to 30% CO2.
The Gj sensitivity to CO2 of channels made of Cx50 was tested by applying −60 mV Vj pulses (12-s duration) at 30-s intervals. With 10-min exposures to 30% CO2, Gj,pk and Gj,ss dropped from 4.2 ± 1.4 μs and 1.8 ± 0.5 μs to 0.8 ± 0.2 μs and 0.7 ± 0.2 μs, respectively (means ± SE; n = 14; Fig. 1, A and B), and recovered to near control values at a similar rate. Exposure to CO2 greatly influenced both Vj sensitivity and Ij inactivation kinetics. Vj sensitivity decreased dramatically, as Gj,ss/Gj,pk increased by ∼78% (n = 14; Fig. 1B). Gj,pk, Gj,ss and Gj,ss/Gj,pk changed with similar time course. The progressive change in Vj sensitivity is clearly seen in the continuous Ij record displayed in Fig. 1A. Exposure to either 5% or 10% CO2 did not significantly affect Gj or Vj sensitivity (data not shown).
Effect of CO2 on Vj sensitivity.
To test the effect of CO2 on Vj sensitivity, the Vj protocol was applied before (Fig. 2B) and during (Fig. 2C) exposure to 30% CO2. Steady-state conditions were reached after 15–20 min of CO2 superfusion (Fig. 2A). After that, the CO2 superfusion was continued for as long as 45–60 min, during which time the oocytes were retested with the Vj protocol. At steady-state conditions, in 30% CO2, the channels displayed a large decrease in Vj sensitivity (Fig. 2, C and D) with respect to controls (absence of CO2; Fig. 2, B and D). In plots of the relationship between Gj,ss/Gj,max and Vj (Fig. 2D), the Boltzmann values were the following: V0 = 24.53 mV, η = 3.3 and Gj,min = 0.15, in the absence of CO2 (n = 11), and V0 = 72.4 mV, η = 1.4 and Gj,min = 0.32, in the presence of CO2 (n = 7; see Table 1). Although the 30% CO2 data did not reach an asymptote, the degree of confidence was >0.98 (χ2 = 0.00019 and 0.00054, for negative and positive Vj, respectively). In Fig. 2D, the fitted function for control conditions departs somewhat from the data points at Vj >40 mV. This is also observed in other Vj records of Cx50 (4, 29, 40) and in those of Cx40 (1, 13) and Cx38 (36). The voltage sensitivity under control condition is slightly lower than that reported by an earlier study on Cx50 expressed in oocytes (39), but higher than those reported for Cx50 expressed in N2A cells (29, 40). Vj sensitivity returned to control values after prolonged CO2 washout (data not shown). The mean values of Gj,pk and Gj,ss decreased slightly during the 30- to 60-min period of exposure to 30% CO2 during which the Vj sensitivity was tested by the Vj protocol, but the changes were not statistically significant; Gj,pk decreased from 1.77 ± 0.68 μs to 1.41 ± 0.56 μs (means ± SE; 20.3% drop; n = 9; P = 0.41) and Gj,ss from 1.26 ± 0.42 μs to 1.03 ± 0.37 μs (means ± SE; 18.3% drop; n = 9; P = 0.41); Gj,ss/Gj,pk remained virtually unchanged (0.71 and 0.73, respectively).
Exposure to 30% CO2 also decreased the speed of Ij inactivation, as the time constants (τ1 and τ2) of Ij decay at Vj = ± 100 mV increased from 6.7 ± 0.4 and 0.68 ± 0.03 s (means ± SE; n = 20) to 12.1 ± 1.9 and 1.1 ± 0.1 s (means ± SE; n = 11; P = 0.0007 and 0.0009, respectively), corresponding to 81% and 63% increase, respectively (Fig. 2C, inset).
Heterotypic Cx50/Cx50-D3N Channels
Heterotypic channels were generated by pairing oocytes expressing Cx50 with oocytes expressing Cx50-D3N. These channels displayed an asymmetrical Vj behavior consistent with a reversal of Vj polarity resulting from the D3N mutation (neutralization of the negative charge in position 3). With Cx50 hemichannels at the positive side of Vj, the magnitude of Ij inactivation increased progressively with increasing Vj gradients (Fig. 3A, bottom traces), whereas with Cx50 at the negative side of Vj, Ij inactivation was negligible (Fig. 3A, top traces). This indicates that the D3N mutation has switched the Vj gating polarity from positive (Cx50) to negative (Cx50-D3N), such that with Cx50 at the positive side of Vj the gates of both Cx50 and Cx50-D3N hemichannels are activated, whereas with Cx50 at the negative side of Vj neither hemichannel’s gate is activated. In plots of the relationship between Gj,ss/Gj,max and Vj, with Cx50-D3N at the negative side of Vj (Fig. 3B, right), the Boltzmann values were: V0 = 43.2 mV, η = 2.82, and Gj,min = 0.0035 (n = 11) (see Table 1). Note that Gj,min is nearly zero, which contrasts with the behavior of homotypic Cx50 channels (Fig. 2D).
Homotypic Cx50-D3N Channels
Gj sensitivity to 5% CO2.
The D3N mutation rendered Cx50-D3N channels considerably more sensitive to CO2 than wild-type Cx50 channels, such that even CO2 concentrations as low as 2.5% were sufficient to significantly affect Gj and Vj sensitivity. With 10 min, exposures to 5% CO2, Gj,pk and Gj,ss dropped from 1.7 ± 0.6 and 0.7 ± 0.2 μs to 0.45 ± 0.15 and 0.09 ± 0.02 μs, respectively (means ± SE; n = 6; Fig. 4, A and B), and recovered to near control values at a slower rate. The CO2 exposure increased significantly the Vj sensitivity because Gj,ss/Gj,pk decreased by ∼48% (n = 6; Fig. 4B). The time courses of Gj,pk and Gj,ss/Gj,pk are well matched (Fig. 4B), as seen with homotypic Cx50 channels as well (Figs. 1B and 2A). The progressive change in Vj sensitivity is clearly visible in the continuous Ij record shown in Fig. 4A.
Effect of CO2 on Vj sensitivity.
The effect of CO2 on the Vj sensitivity of Cx50-D3N channels was tested by applying the Vj protocol before (Fig. 5B) and during (Fig. 5C) exposure to 2.5% CO2. Steady-state conditions were reached after 15–20 min of CO2 superfusion (Fig. 5A). After that time, the CO2 superfusion was continued for as long as 45–60 min and the oocytes were retested with the Vj protocol while they were maintained in 2.5% CO2. At steady-state conditions, in 2.5% CO2, the channels displayed a significant increase in Vj sensitivity (Fig. 5, C and D) with respect to controls (absence of CO2; Fig. 5, B and D). In plots of the relationship between Gj,ss/Gj,max and Vj (Fig. 5D), the Boltzmann values were the following: V0 = 46.09 mV, η = 1.83 and Gj,min = 0.0085, in the absence of CO2 (n = 14), and V0 = 36.15 mV, η = 2.04 and Gj,min = 0.02, in the presence of CO2 (n = 12) (see Table 1); the ∼10 mV shift in V0 (Fig. 5D) is statistically significant (P = 0.0002 and 0.003, for negative and positive Vj, respectively). Vj sensitivity returned to control values after prolonged CO2 washout (data not shown). The mean values of Gj,pk and Gj,ss decreased slightly during the 30–60 min period of exposure to 2.5% CO2 during which the Vj sensitivity was tested by the Vj protocol, but the changes were not statistically significant; Gj,pk decreased from 2.28 ± 1.02 to 2.02 ± 0.91 μs (means ± SE; 11.4% drop; n = 6; P = 0.188) and Gj,ss from 0.86 ± 0.37 to 0.73 ± 0.33 μs (means ± SE; 15% drop; n = 6; P = 0.26); Gj,ss/Gj,pk remained virtually unchanged (0.38 and 0.36, respectively).
Exposure to 2.5% CO2 increased the speed of Ij inactivation. The slow time constant (τ1), measured at Vj = ± 50 mV, decreased from 9.5 ± 1.0 s (n = 8) to 5.6 ± 0.9 s (means ± SE; n = 11; P = 0.01), corresponding to a 41% drop (Fig. 5C, inset). In contrast, τ2 did not change significantly, being 1.2 ± 0.2 s (means ± SE; n = 8) under control conditions and 1.4 ± 0.1 s (means ± SE; n = 11) in 2.5 CO2. Note that Gj,min is nearly zero both in the presence and absence of CO2. In this, homotypic Cx50-D3N channels behave like heterotypic Cx50/Cx50-D3N channels (Fig. 3B), but differ from homotypic Cx50 channels (Fig. 2D).
Gj Sensitivity of Cx50 and Cx50-D3N Channels to pHi
To test the effect of different CO2 concentrations on pHi, oocytes were superfused for 40 min with ND96 solutions gassed with 2.5%, 5.0%, 10%, or 30% CO2 (Fig. 6A). Steady state was reached after 15–20 min of CO2 superfusion (Fig. 6A). pHi reversibly dropped from control values of 7.73 ± 0.05 (means ± SE, n = 14) to 7.37 ± 0.02 (means ± SE; n = 4), 7.19 ± 0.06 (means ± SE; n = 2), 6.98 ± 0.07 (means ± SE; n = 3), and 6.83 ± 0.03 (mean ± SE; n = 5), respectively (Fig. 6A). The mean pHi of 7.73 measured under control conditions in these experiments is close to that previously reported by us (pHi = 7.63; Ref. 17) and Sasaki et al. (pHi = 7.69; Ref. 26). At control conditions, the pHi of oocytes perfused with ND96 solutions (pH 7.6) varied among batches taken from different Xenopus females, as reported by Cicirelli et al. (7).
The Gj of Cx50-D3N channels was significantly more sensitive to lowered pHi than that of Cx50 channels (Fig. 6B). After 30- to 40-min superfusion of ND96 gassed with different CO2 concentrations, Gj of Cx50 channels dropped to 22.2 ± 4.4% (means ± SE; n = 12) and 81.8 ± 2.1% (means ± SE; n = 5) of control values with 30% and 10% CO2, respectively, whereas CO2 concentrations of 5% (n = 3) and 2.5% (n = 2) had no effect of Gj (Fig. 6B). In contrast, in Cx50-D3N channels Gj dropped to 4.8 ± 1.2% (means ± SE; n = 5) and 44.9 ± 5.2% (means ± SE; n = 6) of control values with 5% and 2.5% CO2, respectively, whereas CO2 concentrations of 10% (n = 3) and 30% (n = 2) caused complete cell-cell uncoupling (Fig. 6B).
This study describes the effect of CO2 on chemical and Vj gating sensitivities of channels made of Cx50 wild-type or a Cx50 mutant (Cx50-D3N) in which the aspartate (D) in position 3 was mutated to asparagine (N). The data show that this mutation inverts the Vj gating polarity of Cx50 from positive (4, 8, 35) to negative and that CO2 decreases both Vj gating sensitivity and inactivation speed of Cx50 channels but increases those of Cx50-D3N channels. Evidence from this study strengthen the hypothesis that the response of Vj gating sensitivity to CO2-induced cytosolic acidification may be related to the Vj gating polarity of the connexin (13).
During CO2 superfusion, the time course of the Gj,ss/Gj,pk ratio matched closely that of Gj,pk in both Cx50 wild-type and Cx50-D3N channels. Because the drop in Gj,pk reflects the number of channels closed by CO2 (chemical gating), a possible interpretation is that the changes in Vj sensitivity are mechanistically related to chemical gating. Perhaps there is an interplay between the chemical (slow) gate and the Vj sensor of the fast Vj gate.
In both Cx50 and Cx50-D3N channels, the Vj-dependent Ij decay is best fit by a two-term exponential function, suggesting the presence of two gating components. In Cx50 channels both time constants (τ1 and τ2) are CO2 sensitive as they increase significantly with CO2, indicating that CO2 reduces the speed as well as the sensitivity of Cx50 Vj gating. In contrast, in Cx50-D3N channels τ1 is significantly decreased by CO2, indicating that both the speed and sensitivity of Vj gating are increased by CO2.
The Boltzmann fits correspond well to the data points in Vj records of Cx50-D3N channels (Figs. 3B and 5D), but diverge somewhat at Vj gradients higher than ± 40 mV in control records of Cx50 channels (Fig. 2D). This divergence is also seen in other records of Cx50 channels (4, 29, 40) and in those of Cx40 (1, 13) and Cx38 (36) channels, but not in records of Cx32 (36) and Cx45 (21) channels. Because Cx50, Cx40, and Cx38 channels are positive gaters, whereas Cx32, Cx45, and Cx50-D3N channels are negative gaters, this phenomenon may also be related to gating polarity. Perhaps it involves competition vs. cooperation between fast and slow Vj gates, as in positive gaters; fast and slow Vj gates are believed to operate at opposite ends of the channel, whereas in negative gaters they are likely to operate at the same end of the channel.
As previously reported (9, 30, 40), Cx50 channels are among the most CO2-sensitive connexin channels tested. Indeed, Gj drops by as much as 78% with exposures to just 30% CO2, corresponding to a drop in pHi to 6.83. However, even more striking is the CO2 sensitivity of the Cx50-D3N mutant channels, as Gj drops by as much as 95% with CO2 concentrations as low as 5%, corresponding to a drop in pHi to just 7.19. Because longer exposures to 5% CO2 would have closed all of the Cx50-D3N channels, a concentration as low as 2.5% CO2 was needed to enable us to reach steady-state conditions and yet have a sufficient number of operational channels for measuring the effect of CO2 on Vj sensitivity. This is the reason why the effect of CO2 appears less striking with Cx50-D3N channels than with Cx50 channels. Very likely, a higher CO2 concentration might have had greater effects on both Vj sensitivity and gating speed, if it had been possible to test it.
The high CO2 sensitivity of Cx50-D3N channels may make them the most sensitive connexin channels yet known. Indeed, at sea level (PB = 760 torr, gas saturated with water vapor at 37°C) 5% and 2.5% CO2 correspond to Pco2 values of 36 and 18 torr, respectively. Because cells of average human tissue operate at Pco2 values of ∼48 torr (Pco2 of mixed venous blood), if this connexin mutant were present in a living organism, cells expressing Cx50-D3N would be expected to be constantly in uncoupled state. The reason for this high CO2 sensitivity is unclear, but it might relate to the loss of the negative charge in position 3 resulting from D/N mutation. This also points to the extreme relevance of NH2 terminus domain to chemical gating. Significantly, this domain is likely to be a calmodulin (CaM) binding site, and a model of chemical gating that envisions a direct CaM-Cx interaction has been proposed (for reviews, see Refs. 12 and 19). In Cx32, this NH2 terminus domain has been found to bind CaM in Ca2+-dependent way with a Kd of 27 nM (32). Because CaM interacts best with basic amphiphilic helical domains, perhaps the removal of a negative charge from this site enhances its CaM binding efficiency.
Evidence that the D3N mutation inverts the gating polarity of Cx50 from positive to negative confirms similar data on Cx26 (D2N mutation; 11, 42). Significantly, in Cx32 the reverse mutation (Cx32-N2D) inverts the polarity of Cx32 from negative to positive (11, 23, 34, 42). The data on Cx50 and Cx50-D3N are consistent with the hypothesis that the manner in which Vj gating responds to CO2 may be related to the gating polarity of the connexin (13). Recently, this hypothesis has also been strengthened by evidence that Cx32-N2D mutant channels (positive gaters) decrease in Vj gating sensitivity during CO2 application (42).
The D3N mutation in Cx50 enables the channels to close completely at high voltages: with Vj = 100 mV, Gj,min is virtually zero in both heterotypic Cx50/Cx50-D3N and homotypic Cx50-D3N channels. This is interesting, because gating to full channel closure is believed to reflect the behavior of the slow Vj gate (5, 17, 20). The function of the slow Vj gate is usually hidden, in the absence of chemical uncouplers, in all connexin channels except in Cx45 channels (6), but it manifests itself in various disparate channels made of connexin mutants (17, 20, 23, 31). This suggests that the D3N mutation of Cx50 may unmask or unlatch the slow Vj gate. Of course, the potential effect of the D3N mutations on slow gating could only be properly evaluated at the single channel level.
The reason why CO2 decreases the Vj sensitivity of positive gaters, such as Cx26 (42), Cx40 (13), Cx37 (C. Peracchia, unpublished observations), Cx38 (38), Cx50 (this study) and Cx32-N2D (42), and increases that of negative gaters, such as Cx32 (42), Cx45 (21), and Cx50-D3N (this study), is unclear. The molecular basis of gating polarity is also only partly understood, but there is good evidence that charged residues at the initial segment of NT play a role (34). A standing hypothesis envisions the NT domain as the voltage sensor, located within the channel’s mouth (11, 23, 24, 34). Positive and negative gaters would have an acidic or a basic residue at the initial sequence of NT, respectively. With the establishment of a Vj gradient, the NT of positive gaters would move away from the channel’s mouth (toward the cytoplasm) at the positive side of Vj, enabling channel gating by another connexin domain or an accessory molecule. Conversely, in negative gaters NT would move away from the channel’s mouth at the negative side of Vj. Positive gaters would be so by virtue of the acidic residue in the third position (D3), which is absent in negative gaters, except in Cx43.
With CO2-induced cytosolic acidification, histidine residues are likely to be protonated, resulting in the addition of positive charges to relevant domains of connexins and/or accessory proteins. A possibility is that protonated histidines affect the function of the Vj sensor in opposite ways depending on the gating polarity of the connexin. However, preliminary experiments with the histidine reagent diethylpyrocarbonate (DEPC) do not support this hypothesis, as 20-min applications of 0.5–1 mM DEPC (pH 6.0) to oocytes expressing Cx50 channels altered neither the CO2 sensitivity of Vj gating nor the effect of CO2 on Gj. If indeed histidine protonation does not play a role, an alternative mechanism could involve an interplay between the Vj sensor and chemical gate. Because the chemical gate is likely to be negatively charged (17, 20), if this gate were to come in proximity of the Vj sensor, before closing the channel completely, it could potentially hinder or facilitate the Vj-induced displacement of the sensor from the channel lumen in positive or negative gaters, respectively. Significantly, the absence of a DEPC effect on CO2 induced uncoupling efficiency seriously questions the potential role of histidine protonation in the uncoupling mechanism as well.
In conclusion, this study shows that CO2 induced cytosolic acidification decreases both sensitivity and speed of Vj gating in channels made of mouse Cx50 and increases those of channels made of a Cx50 mutant (Cx50-D3N), in which the aspartate in position 3 is mutated to asparagine. In addition, the data indicate that the D3N mutation greatly increases chemical gating sensitivity to CO2 and may unmask the activity of the slow Vj gate. Because the D3N mutation inverts the gating polarity of Cx50 from positive (4, 8, 35) to negative, the opposite effect of CO2 on Cx50 and Cx50-D3N Vj gating sensitivity confirms the hypothesis that the way, in which Vj gating responds to CO2 may be related to the polarity of Vj gating.
This study was supported by National Institute of General Medical Sciences Grant GM-20113.
The authors thank Joey T. Chen for excellent technical assistance.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society