Role of the Carboxyl Terminal of Connexin43 in Transjunctional Fast Voltage Gating
Previous studies show that chemical regulation of connexin43 (Cx43) gap junction channels depends on the integrity of the carboxyl terminal (CT) domain. Experiments using Xenopus oocytes show that truncation of the CT domain alters the time course for current inactivation; however, correlation with the behavior of single Cx43 channels has been lacking. Furthermore, whereas chemical gating is associated with a “ball-and-chain” mechanism, there is no evidence whether transjunctional voltage regulation for Cx43 follows a similar model. We provide data on the properties of transjunctional currents from voltage-clamped pairs of mammalian tumor cells expressing either wild-type Cx43 or a mutant of Cx43 lacking the carboxyl terminal domain (Cx43M257). Cx43 transjunctional currents showed bi-exponential decay and a residual steady-state conductance of approximately 35% maximum. Transjunctional currents recorded from Cx43M257 channels displayed a single, slower exponential decay. Long transjunctional voltage pulses caused virtual disappearance of the residual current at steady state. Single channel data revealed disappearance of the residual state, increase in the mean open time, and slowing of the transition times between open and closed states. Coexpression of CxM257 with Cx43CT in a separate fragment restored the lower conductance state. We propose that Cx43CT is an effector of fast voltage gating. Truncation of Cx43CT limits channel transitions to those occurring across the higher energy barrier that separates open and closed states. We further propose that a ball-and-chain interaction provides the fast component of voltage-dependent gating between CT domain and a receptor affiliated with the pore.
Gap junction channels provide a pathway for direct cell-to-cell communication between adjacent cells and are involved in a number of biological functions. Mutations of their constituent proteins have been associated with nonsyndromic deafness, Charcot-Marie-Tooth, and congenital cataracts, among other congenital human diseases.1 Each gap junction channel is composed of 2 identical hexameric structures called connexons that dock across the extracellular space and form a permeable pore isolated from the extracellular region or gap. Each connexon results from oligomerization of an integral membrane protein called connexin (Cx).2 Once assembled, connexins cross the cell membrane 4 times, thus rendering 3 intracellular domains: the amino terminal, the intracellular loop, and the carboxyl terminal. The most abundant connexin isotype in the heart, brain, and other tissues is the 43-kDa protein connexin43 (Cx43).
Gating of gap junction channels is regulated via chemical mechanisms including phosphorylation3–5⇓⇓ and low intracellular pH.6,7⇓ A number of structure-function studies show that, for Cx43, the carboxyl terminal (CT) region acts as the major regulatory domain. Truncation of Cx43 interferes with the chemical gating induced by low intracellular pH, as well as by insulin, insulin-like growth factor,8 and v-src.9 Moreover, the regulatory function can be rescued when CT is coexpressed as a separate fragment, leading to the hypothesis that chemical gating follows a “ball-and-chain” model, similar to the one proposed for voltage gating of potassium channels.10 According to that hypothesis, CT acts as a gating particle that, under the appropriate conditions, binds to a receptor affiliated with the pore.6
Gap junction conductance can also be regulated by trans-junctional or transmembrane voltage. Initial mutagenesis studies suggested that the CT domain played only a minor role in the voltage gating of Cx43.11 However, recent studies by Revilla et al,12 conducted in Xenopus oocyte pairs, showed that the CT domain of Cx43 is clearly part of the voltage-gating mechanism. Their data revealed that the time course of current relaxation during transjunctional voltage pulses is best described by 2 exponentials where only the fast component is eliminated after truncation of the CT domain. Furthermore, the total amplitude of current relaxation increases. These results strongly suggested the presence of 2 separate gating mechanisms, one of them dependent on the CT domain. Still lacking, however, is information about the single channel properties of truncated Cx43 and the intimate mechanisms responsible for the changes in current relaxation observed in the oocyte expression system. Moreover, in the context of the ball-and-chain model for chemical gating,13 the observations of Revilla et al12 lead to the question as to whether the fast voltage-gating mechanism could be restored by coexpression of the CT fragment.
Single channel analysis reveals that some connexins (including Cx43) present conductance levels intermediate between the open and the closed configurations.14,15⇓ Recent studies have linked the existence of a residual state in the unitary channel behavior of connexins with the residual conductance observed from macroscopic gap junction currents during voltage gating (also referred to as Gmin).16 Furthermore, Anumonwo et al17 recently reported that truncation of the CT domain of Cx40 was associated with the elimination of the residual state. The present study addresses the following questions: what are the single channel properties of a Cx43 channel after truncation of its CT domain? Is the interaction of the CT domain with the channel pore consistent with the idea of the CT fragment as an independent gating particle? And finally, can the changes observed at the single channel level be sufficient to explain the changes in macroscopic currents reported by Revilla et al?12 The results presented in this study have led us to propose a model where the CT domain acts as an effector of fast voltage gating. Whether chemical and voltage gating are interdependent in these channels remains to be determined.
Materials and Methods
Cell Culture and Transfection of Neuroblastoma Cells
Murine neuroblastoma (N2a) cells stably transfected with plasmids containing cDNA for wild-type Cx43 or Cx43M257 were used for this study. Additional experiments were conducted with N2a cells containing regulatable (LacSwitch) Cx43 plasmid (Strategene, Calif). Stable Cx43M257 transfectants were transiently transfected with pIRES plasmid encoding for amino acids 259 to 382 of Cx43 and green fluorescent protein (GFP). GFP-positive cells were observed using epifluorescent illumination provided by a 100-watt mercury lamp and the appropriate set of filters. Only cell pairs in which both cells demonstrated GFP were used. Mammalian cells were grown in Dulbecco’s modification of Eagle’s medium (DMEM, Life Technologies) supplemented with 10% Fetal Bovine Serum (Atlanta Biologicals), l-glutamine and 5000 U/L penicillin, and 5 mg/L streptomycin. The cells were maintained at 37°C in 5% CO2.
Cloning and Transfection
Cx43 was inserted into pCDNA3.1 (Invitrogen); cDNA for Cx43M257 was cloned into pEFBOS18 vector, and cotransfected with pRSVNeo. LipofectAmine (Life Technologies) was used for transfections, following manufacturer’s directions. Stable transfectants were selected with 800 μg/mL G418 (Geneticin, Life Technologies).
Cx43CT was inserted into the bicistronic expression vector (pIRES), as previously reported.19 This vector produced simultaneous expression of Cx43CT and EGFP for identification of positive cells. High-efficiency (>70%) transient transfection was obtained using Effectene (Qiagen), following the manufacturer’s protocol (0.5 μg of plasmid DNA).
The dual-whole-cell voltage clamp technique was used to record gap junction currents in N2a transfectants. Transjunctional currents (Ij) were measured in one cell of the pair (held at constant voltage) while voltage steps (from −100 to 100 mV, in 10 or 20 mV increments) of 10 to 20 seconds in duration were applied to the other (using pClamp software; Axon Instruments). Dependence of gap junctions to transmembrane potential was tested by changing the membrane potential of both cells simultaneously to values between −80 and 80 mV for 20 seconds. Patch pipettes were filled with a solution containing cesium (in mmol/L: 130 CsCl; 0.5 CaCl2; 10 Hepes; 10 EGTA; pH 7.2). During recording, cells were kept at room temperature in a cesium-containing solution (in mmol/L: 160 NaCl; 7 CsCl; 2.0 CaCl2; 0.6 MgCl2; 10 Hepes; pH 7.4). Only electrodes of less than 5 MΩ and cell pairs with initial junctional conductance no larger than 10 nS were used in gating calculations to minimize the effects of electrode resistance. Series resistance was compensated up to 70% at the beginning of all experiments. All current traces were digitized (Neurodata) and stored on VCR tapes.
Recording and Analysis of Macroscopic Currents
All macroscopic voltage-dependent current traces were digitized at 11 kHz and acquired at 1 kHz for analysis. To avoid under-sampling fast current decay, the first 100 ms of tracings were acquired at 10 kHz. The steady state current (iss) was obtained 10 to 20 seconds after initiation of the pulse. Initial conductance (Gi) and steady state conductance (Gss) measured at each pulse were calculated as ratios of currents to transjunctional voltages: iinst/Vj and iss/Vj. The steady-state voltage sensitivity (Gss/Vj) of gap junction channels was normalized to the average conductance measured by a hyperpolarizing 5-mV prepulse of 200 ms. The function describing normalized Gss as a function of Vj was best fit by a single Boltzmann equation.20 The kinetics of junctional current relaxation were described by single- or bi-exponential functions, based on a Marquadt protocol involving X2 minimization (Origin; Microcal).
Recording and Analysis of Single-Channel Currents
Unitary junctional currents were obtained during long voltage steps applied to one of the cells. Current traces from cells with fewer than 10 functional gap junction channels were used to calculate single channel gating properties. Transitions between conductive and nonconductive states were measured with a digitizing board (Summagraphics). Gaussian distribution best fits were calculated for all-event histograms using Origin software (Microcal). The unitary conductance in these histograms was determined by calculating the difference between means from Gaussian curves calculated for the open and residual or closed states. Transition times between current levels were measured from amplified traces filtered at 1 kHz and digitized at 2 kHz. Open time histograms were calculated using event-sampling protocols (pClamp) in 1- to 3-minute traces at transjunctional voltages of 40 to 60 mV.
Statistical analysis of differences between parameters utilized the Student’s t test; a value of P<0.05 was considered significant. All data are expressed as the mean±SEM. N refers to the number of cell pairs; n, the number of transition events.
The methods for cell preparation, mRNA injection, and protein expression have been described in detail elsewhere.6
Unitary and Residual Conductances of Cx43
We characterized (without pharmacological uncouplers) single channel behavior of Cx43 channels to establish a baseline to compare with the results of mutant constructs. In agreement with previous reports,14,15⇓ Cx43 channels gated between 3 states: closed, open, and residual. Figure 1A (left panel) illustrates a typical recording from an N2a cell pair expressing Cx43. The horizontal dotted lines in the expanded trace indicate that several conducting states were detected. Also consistent with previous reports,14,21⇓ the amplitude of residual conductance was about one third of that of the main open state. The channel dwelled at the residual state for long periods, particularly at Vj>60 mV, and most transitions occurred between residual and open states. The all-points histogram (Figure 1A, right) shows 2 large peaks, corresponding to dwelling of the channel at closed and residual states. The amplitudes of the closed-to-open and residual-to-open transitions depicted were 119 and 102 pS, respectively. The dominance of the residual-to-open transitions can be observed in the histogram of events shown in Figure 1A right panel, which was best fit by 3 Gaussian curves centered at 35±5, 102±18, and 118±14 pS (N=6, n=415).
Effect of CT Truncation on Single- Channel Conductance
Unitary Conductances of Cx43 and Cx43M257: Elimination of the Residual State
Truncation of the CT domain did not significantly modify the magnitude of the main unitary conductance of Cx43 channels. However, subconductance levels (including the residual state) were no longer observed. A trace recorded from an N2a cell pair expressing Cx43M257 channels is shown on the left side of Figure 1B. All transitions were of similar amplitude (between 100 and 120 pS), suggesting the presence of several channels with only one conductive state. A histogram of events depicting the amplitude of channel transitions recorded from 10 different cell pairs is presented on the right side of Figure 1B. Contrasting with data obtained from Cx43, all events recorded from Cx43M257 centered on a single peak best described by a single Gaussian function centered at 110±10 pS (n=258), indicating that channels no longer dwell at the residual state but transit directly between open and closed states in response to transjunctional voltage gradient. These observations are consistent with those of Anumonwo et al17 showing that truncation of the CT domain of Cx40 eliminates the residual state. It has been proposed that the steady-state junctional conductance observed in macroscopic junctional currents is a consequence of the existence of a residual state.14 Our data help explain the reduction in steady-state junctional conductance of Cx43M257 channels observed by Revilla et al.12
CT Domain as an Independent Voltage-Gating Particle
Coexpression of Cx43CT is known to restore chemical gating of truncated Cx43 channels.13 We tested whether coexpression of Cx43CT would also restore intermediate conductive states. N2a cells stably transfected with Cx43M257 were transiently transfected with a bicistronic plasmid encoding the CT fragment and, as a separate protein, GFP. Control experiments were conducted on cells transfected with the same plasmid but encoding only GFP protein. Only cell pairs positive for GFP were selected for double whole cell voltage clamping. Figure 2A shows examples of channel activity recorded from Cx43M257 after coexpression of Cx43CT. The traces clearly show several levels in which the current dwelled temporarily that were not multiples of the main open state. These levels are reflected in the histogram displayed in Figure 2B (lower panel). When fitted to Gaussian distributions (with a minimal X2 of 0.0065), the main peaks appeared at 36±4, 50±7, 61±17, 84±8, 100±16, and 112±7 pS (N=6, n=349). This multimodal histogram contrasts with data obtained from N2a cells expressing Cx43M257 and GFP, but not Cx43CT (Figure 2B; upper panel). In cells without the CT fragment, a single average unitary conductance of 100±10 pS was observed (N=6, n=249). This value is the same as that recorded from Cx43M257 channels in the absence of GFP (see Figure 1B). Overall, the data show that Cx43CT is capable of interacting (directly or indirectly) with the pore-forming region of Cx43 to regulate channel conductance.
Effect of CT Truncation on Channel Kinetics
Open and Closed Times of Cx43 and Cx43M257
Our data shows that truncation of the CT domain locks the Cx43 channel into only 1 conducting state. For the next step, we assessed whether truncation of the carboxyl terminal domain affects channel kinetics. At a single channel level, the most appropriate way to assess these effects is to determine the mean open/close time for events recorded. This is a difficult task in gap junction channels because of their slow voltage-dependent kinetics. Nonetheless, even with a small number of events (N=2, n=240), we were able to determine that the mean open time of Cx43 channels at Vj=60 mV, analyzed at 3 levels of maximal conductance and excluding the residual conductance, was 126±20 ms (Figure 3A). Recordings at 40 mV (not shown) yielded a mean open time of 306±50 ms (N=2, n=135). In contrast, the mean open time of Cx43M257 channels was considerably prolonged (2450±200 ms; see Figure 3B). Estimation of closed times for Cx43M257 channels yielded a bi-modal distribution, with average values of 42±10 and 3399±300 ms (N=3, n=72; see Figure 3B, lower right). Estimations of closed times for Cx43 channels were not possible because of the prolonged residence of the channel in the residual state.14 Our data show that truncation of the CT domain caused a significant prolongation of the open time and a bimodal distribution of closed times. This behavior may explain, at least in part, the longer relaxation time course recorded from Cx43M257 channels in Xenopus oocytes.12
Time Course of Transitions Between States
In addition to dwell times, we tested whether truncation of the CT domain altered the time course for transitions between conducting and nonconducting states. For Cx43 channels recorded at Vj=60 mV, the transition times between residual and open states and between closed and open states were consistently shorter than 20 ms (N=3, n=34; see Figure 4A). Very slow transitions between residual and closed states were infrequently detected, ranging from 50 to 350 ms (not shown). In contrast, only 20% of the transitions between closed and open states measured in Cx43M257 channels showed a single step lasting less than 10 ms (transitions labeled “1” and “3” in Figure 4B). The rest of the transitions (eg, those labeled 2, 4, 5, 6, and 7 in Figure 4B) were longer than 20 ms and displayed an irregular behavior where the current trace appeared to reach several intermediate levels before achieving the main unitary conductance (N=3; n=26). Altogether, the data indicate that truncation of the CT domain impedes transition between states and tends to stabilize the channel in either the open or closed configuration. These results are consistent with prolonged open and closed times recorded from Cx43M257 channels (Figure 3) and provide single-channel kinetics that correlate with the macroscopic current kinetics previously reported.12 As a final step in our study, we assessed whether these modifications in channel kinetics translated into changes in macroscopic currents consistent with those observed in Xenopus oocyte pairs.
Effect of CT Truncation on the Voltage Dependence of Macroscopic Junctional Currents
We studied time and voltage dependence of junctional currents from Cx43-transfected mammalian cells. The traces in Figure 5A show that the time course and amplitude of Ij relaxation is a function of transjunctional voltage. The value of normalized Gss decreased as the Vj amplitude increased, producing the sigmoidal Gss-Vj relationship depicted by open circles in Figure 5D; data from 6 different cell pairs were averaged and fit using a Boltzmann relationship, producing the parameters shown in Table 1. The time course for current relaxation during Vj pulses of 60 mV or more was best fit by a bi-exponential function. Time constant values are presented in Table 2. All measured parameters were consistent with those previously reported for connexin43 expressed in mammalian cells.14
Junctional currents recorded in N2a cells transfected with Cx43M257 behaved differently than those observed in wild-type Cx43 transfectants. As shown in Figure 5B, current relaxation followed a slower time course, but the amplitude of the voltage-dependent current decay was larger than that of Cx43 channels. Gss actually reached near-zero values when long-duration pulses (30 seconds or more) to Vj of 100 mV were applied (see Figure 5C). The Gss-Vj relationship for Cx43M257 is shown in Figure 5D (filled symbols), and the Boltzmann parameters calculated from currents recorded during 20-second voltage steps are shown in Table 1. V0 was significantly larger for Cx43M257 than for Cx43 channels (V0Cx43≈56,55 mV; V0Cx43M257≈73,76 mV). It should be noted that, in some cases the calculated V0 values might be overestimated, because junctional currents did not reach a true steady state at the end of the long-duration voltage pulse. In contrast with results obtained from Cx43, the time course for current relaxation of Cx43M257 channels was best fit by a single exponential function. For Vjs of 80 mV and 100 mV, the time constant of that exponential was larger than the slow component of the bi-exponential equation describing the behavior of Cx43 channels, reflecting the fact that truncation of the CT domain leads to slower current kinetics. Quantitative parameters are presented in Table 2. Overall, the data obtained from macroscopic currents is consistent with that recorded by Revilla et al in Xenopus oocytes.12 More importantly, the behavior of the macroscopic currents is consistent with changes in unitary currents brought about by truncation of the CT domain.
For comparative purposes, the role of CT on Vj dependence was addressed in Xenopus oocytes. Cx43, Cx43M257, or Cx43M257+Cx43CT were expressed in different pairs of oocytes. The change in Gss for each of the groups was recorded from 5 experiments, and the best fit for each of these groups was obtained with the parameters shown in Table 1. V0 calculated for each of the different groups was not statistically different, although Gmin for Cx43M257 and Cx43M257+Cx43CT was significantly reduced when compared with wild-type Cx43. As shown in Table 1, no statistically significant differences were observed between their voltage dependence parameters. Expression of Cx43CT was confirmed by conventional immunoprecipitation for 2 distinct amounts of mRNA injected (not shown).
There is increasing evidence that the CT region is the major regulatory domain of Cx43. Here, we report data on the role of the CT domain on single-channel behavior and voltage-gating characteristics for Cx43. Our recordings of Cx43 channels showed the characteristic unitary conductance profile, wherein intermediate (subconducting) levels could be identified between open and closed states.14 In contrast, single channels recorded from Cx43M257 showed a single conducting (open) state. Interestingly, subconductance states could be seen in the same Cx43M257 cells if Cx43CT was coexpressed as a separate fragment. Kinetic analysis of single channel recordings showed that CT truncation led to significant prolongation of open times and a large increase in transition times. These data were consistent with the characteristics of macroscopic junctional currents recorded from Cx43M257 transfectants. Our observations show that changes in voltage-gating kinetics reported by Revilla et al12 are not exclusive to the Xenopus oocyte system. More importantly, our data provide necessary single channel correlation to explain changes seen in large-channel populations and emphasize the relevance of the CT domain in voltage gating as an independent particle. Based on these observations, we propose that the CT domain is an effector for fast voltage gating, in a manner consistent with the ball-and-chain model of channel regulation10 (see also Anumonwo et al17).
Exogenous expression of Cx43 in N2a cells allows characterization of single-channel behavior as well as macroscopic voltage dependence. As in any expression system (including Xenopus oocytes), variability in expression of the exogenous product is expected between experiments. Our studies were conducted in cells expressing junctional conductances within the resolution of our dual voltage clamp system. For experiments involving coexpression of Cx43 and GFP, we chose pairs where fluorescence intensities were similar. In previous studies this method was reliable in detecting Gj values within limited ranges.22 It is extremely unlikely that our results are a consequence of differences in protein expression levels.
Although subconductance states were not observed in channels without CT, the main unitary conductance of Cx43M257 channels was similar to that of wild-type Cx43. A similar effect was observed after truncation of CT in Cx40,18 suggesting that the CT domain is not a major component of the ion-conducting pathway. These observations are consistent with recent high-resolution structural data showing that the narrowest portion of the channel is located near the docking region.23 However, the CT domain may interact with pore-affiliated regions to modify the pore structure as it transits from its closed to its open state.
Our data show that Cx43M257 channels dwell in the open or the closed states for longer periods than wild-type Cx43. CT truncation led to a prolongation of the transition times between closed and open states, although fast transitions were still detected. These results are consistent with the hypothesis that truncation of the CT domain unmasks a slow-gating mechanism that functions across a higher energy barrier and is structurally independent of the CT domain.12 Further studies will be necessary to identify the structural bases for slow Vj gating of Cx43. Our data are consistent with the suggestion24 that the sensor for slow Vj gating may be, as in the case of other connexins, in the NT or loop domains.25,26⇓
Previous studies using Xenopus oocytes have suggested that in addition to the Vj gates (fast and slow), Cx43 contains a slow gate that is responsive to transmembrane voltage (Vm) and is independent of the integrity of the 258 to 382 fragment.24 In our studies, we failed to detect dependence of either Cx43 or Cx43M257 channels on Vm. The source of this apparent inconsistency is unclear, but previous studies show that the voltage-gating behavior of some connexins varies depending on the expression system (compare, eg, Beblo et al27 with White et al28; see also Anumonwo et al17). This may result from intrinsic differences in channel properties or on ancillary proteins affiliated with the channel as they assemble in different cells. Whether endogenous gap junctions show Vm gating remains to be determined. It is tempting to speculate that, if present in cardiac myocytes, Vm gating may lead to a decrease in electrical coupling during rapid pacing or fast idiopathic rhythms. This may explain early observations showing rate-dependent uncoupling in cardiac tissues29,30⇓ and may also be part of the mechanism responsible for self-perpetuation of fibrillatory rhythms in mammalian hearts.31
Previous studies showed that coexpression of the CT domain17 recovered the residual state of truncated Cx40 channels, but that the frequency of transition to the residual state in the coexpression system was lower than in wild-type Cx40. In contrast, Cx43CT coexpression in Cx43M257 cells produced a complex histogram of events, with proportionately large peaks at several intermediate states. Because they are the same cell type, expression plasmids and transfection protocols were used for both connexins; it is unlikely that these contrasting results were caused by different transfection/expression efficiencies. Rather, it seems to indicate intrinsic differences in gating properties of one connexin versus the other.
Our data show that independent expression of Cx43CT as a separate fragment produces more than one transitional state between fully open and closed conductances. The nature of these intermediate states is unclear. Using the ball-and-chain model, we speculate that these residual states result from interaction of the CT domain with a separate region of the molecule that acts as a receptor for the gating particle.10,17⇓ In this scenario, a covalently attached CT domain may be more constrained structurally, and may provide a more stable particle-receptor interaction and one residual conductive state. In contrast, a free, less-organized CT fragment may have multiple, but less stable, points of interaction with the receptor, producing multiple conductive states. Further experiments will be necessary to assess this or alternative hypotheses. Our results do show that the CT fragment is capable of interacting with a separate region of the molecule to modify the conductive state of the channel. The analogies and/or similarities between this behavior and the one described for chemical Cx43 gating13 remain to be determined.
A number of studies have explored the structural bases for voltage gating. Modifications in different connexin domains have led to alterations in gating behavior for several connexins.17,32⇓ Our study demonstrates a role for the CT domain in the prevalence of the residual state and temporal control of state-to-state transitions. It is clear that the CT domain is not the sole component of the process, but an element in a complex sequence of events. Our results, when compared with those previously published,6,12⇓ lead us to propose that the CT domain acts as a gating particle for both chemical and voltage regulation of the Cx43 channel. Whether the sensor and/or the transducer of the voltage-gating process are also part of the CT structure remains to be determined.
This work was supported by grants from the National Institutes of Health (R01-HL 463469 to A.M.; R01 GM57691 and P01 HL39707 to M.D.) and by the Swiss National Science Foundation (32-55745.98 to M.C.).
Original received July 19, 2001; revision received January 14, 2002; accepted January 14, 2002.
- ↵Abrams CK, Bennett MVL. Hereditary Human Diseases Caused by Connexin Mutations.In: Peracchia, C, ed. Gap Junctions: Molecular Basis of Cell Communication in Health and Disease. New York, NY: Academic Press; 2000; 423–459.
- ↵Beyer EC, Goodenough DA, Paul DL. The connexins, a family of related gap junction proteins.In: Hertzberg EL, Johnson RG, eds. Gap Junctions. New York, NY: Alan R. Liss Inc; 1988; 167–175.
- ↵Moreno AP, Saez JC, Fishman GI, Spray DC. Human connexin43 gap junction channels: regulation of unitary conductances by phosphorylation. Circ Res. 1994; 74: 1050–1057.
- ↵Kwak BR, Hermans MM, De Jonge HR, Lohmann SM, Jongsma HJ, Chanson M. Differential regulation of distinct types of gap junction channels by similar phosphorylating conditions. Mol Biol Cell. 1995; 6: 1707–1719.
- ↵Kanemitsu MY, Loo LW, Simon S, Lau AF, Eckhart W. Tyrosine phosphorylation of connexin 43 by v-Src is mediated by SH2 and SH3 domain interactions. J Biol Chem. 1997; 272: 22824–22831.
- ↵Homma N, Alvarado JL, Coombs W, Stergiopoulos K, Taffet SM, Lau AF, Delmar M. A particle-receptor model for the insulin-induced closure of connexin43 channels. Circ Res. 1998; 83: 27–32.
- ↵Zhou L, Kasperek E, Nicholson B. Dissection of the molecular basis of pp60v-src induced gating of connexin 43 gap junction channels. J Cell Biol. 1999; 144: 1033–1042.
- ↵Hoshi T, Zagotta WN, Aldrich RW. Biophysical and molecular mechanisms of Shaker potassium channel inactivation. Science. 1990; 250: 533–538.
- ↵Fishman GI, Moreno AP, Spray DC, Leinwand LA. Functional analysis of human cardiac gap junction channel mutants. Proc Natl Acad Sci U S A. 1991; 88: 3525–3529.
- ↵Valiunas V, Bukauskas FF, Weingart R. Conductances and selective permeability of connexin43 gap junction channels examined in neonatal rat heart cells. Circ Res. 1997; 80: 708–719.
- ↵Harris AL, Spray DC, Bennett MV. Kinetic properties of a voltage-dependent junctional conductance. J Gen Physiol. 1981; 77: 95–117.
- ↵Anumonwo JM, Taffet S, Gu H, Chanson M, Moreno AP, Delmar M. The carboxyl terminal domain regulates the unitary conductance and voltage dependence of connexin40 gap junction channels. Circ Res. 2001; 88: 666–673.
- ↵Mizushima S, Nagata S. pEF-BOS, a powerful mammalian expression vector. Nucleic Acids Res. 1990; 18: 5322.
- ↵Spray DC, Harris AL, Bennett MV. Equilibrium properties of a voltage-dependent junctional conductance. J Gen Physiol. 1981; 77: 77–93.
- ↵Unger VM, Kumar NM, Gilula NB, Yeager M. Three-dimensional structure of a recombinant gap junction membrane channel. Science. 1999; 283: 1176–1180.
- ↵Revilla A, Bennett MV, Barrio LC. Molecular determinants of membrane potential dependence in vertebrate gap junction channels. Proc Natl Acad Sci U S A. 2000; 97: 14760–14765.
- ↵Oh S, Abrams CK, Verselis VK, Bargiello TA. Stoichiometry of transjunctional voltage-gating polarity reversal by a negative charge substitution in the amino terminus of a Cx32 chimera. J Gen Physiol. 2000; 116: 13–31.
- ↵Beblo DA, Wang HZ, 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.
- ↵White TW, Paul DL, Goodenough DA, Bruzzone R. Functional analysis of selective interactions among rodent connexins. Mol Biol Cell. 1995; 6: 459–470.
- ↵Allessie MA, Boyden PA, Camm AJ, Kleber AG, Lab MJ, Legato MJ, Rosen MR, Schwartz PJ, Spooner PM, Van Wagoner DR, Waldo AL. Pathophysiology and prevention of atrial fibrillation. Circulation. 2001; 103: 769–777.