INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

GAP JUNCTION CHANNELS are critical for the passage of current between cells in the working myocardium and specialized conducting tissues of the heart. Gap junctions are also present and implicated in the function of other cardiovascular cell types including endothelial cells and vascular smooth muscle cells. A gap junction channel is formed by the meeting of two hemichannels (connexons) located in the plasma membranes of apposed cells. A hemichannel is composed of six subunit proteins [connexins (Cx)]. Several different connexins are expressed in cardiovascular cells, including connexin Cx37, Cx40, Cx43, and Cx45. Each of these connexins can form functional gap junctions by themselves (homomeric/homotypic channels). However, many cells coexpress more than one connexin, suggesting that they might participate in the formation of mixed channels. One specialized mixed channel (a heterotypic channel) could be formed by the pairing of two hemichannels that contain different connexins. A more complicated type of mixing would occur if multiple connexins mixed within the same hemichannel (a heteromeric channel).

Biochemical studies have demonstrated the formation of heteromeric gap junction channels in expression systems and in various mature tissues. Stauffer (35) expressed recombinant Cx32 and Cx26 in insect cells and used gel filtration to detect mixing in detergent-solubilized connexons, whereas Jiang and Goodenough (21) used coimmunoprecipitation to show mixing of Cx46 and Cx50 in the lens. Bevans et al. (4) found heteromeric mixing of Cx26 and Cx32 in the rodent liver when Cx26 was coisolated with Cx32 by affinity chromatography on an anti-Cx32 column. He et al. (18) showed coimmunoprecipitation of Cx40 with Cx43 in A7r5 cells.

The observation of heteromeric forms has lead to speculation that some or all of the biophysical properties of the resultant gap junction channels might be different from homomeric gap junction hemichannels containing only one connexin type. We (5) previously utilized electrophysiological studies of Cx37 and Cx43 cotransfected into N2A cells to show the existence of a population of heteromeric Cx37-Cx43 channels. Dual whole cell patch-clamp analysis of coexpressing cells demonstrated that the voltage dependence was weaker and the range of single channel conductances was broader than could be explained as arising from conventional homotypic or heterotypic gap junction channels. Other studies have suggested that Cx43 and Cx40 can also make heteromeric channels. In A7r5 rat vascular smooth muscle cells, which normally coexpress Cx40 and Cx43, the macroscopic voltage dependence was weaker than that of either homotypic form, and some of the single channel conductances could not be explained easily as homotypic Cx43 or Cx40 forms (18). These authors concluded that the single channel data provided evidence for putative heteromeric Cx40-Cx43 gap junction channels. However, they did not consider heterotypic Cx40-Cx43 channels, whose existence has been subsequently established (37).

A tempting hypothesis is that the disparate properties of heteromeric channels relative to homotypic and/or heterotypic channels could affect multicellular function. For example, in excitable cells (such as the myocytes found in cardiac conducting and pacemaker regions), the type of gap junction channel (homotypic, heterotypic, or heteromeric) could potentially have a profound influence on action potential propagation (22, 34, 37).

The purpose of the current study was to examine biochemically the possible heteromeric mixing of Cx40 with Cx43 and to analyze the electrophysiological properties of gap junction channels formed between cells coexpressing these two connexins. We studied these two connexins because they are abundantly coexpressed in various cells (including atrial myocytes and some endothelial and smooth muscle cells). Double whole cell patch-clamp methods were used to compare the gating properties of channels formed between cells coexpressing Cx40 and Cx43 with those of homotypic Cx40 (Cx40-Cx40) and Cx43 (Cx43-Cx43) and heterotypic Cx43-Cx40 forms.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cells and culture conditions. Experiments were carried out using HeLa cells transfected with DNA coding for rat Cx40, rat Cx43, or mouse Cx45. Cx40 or Cx43 DNA was subcloned into the eukaryotic expression plasmid pcDNA3.1/hygro (Invitrogen) or pSFFV-neo (15). Constructs containing a (His)₆ tag appended to the Cx43 COOH-terminal were obtained using PCR methods. Cells were stably transfected with linearized DNA using lipofectin (Life Technologies) as described earlier (39). Cells that stably expressed both Cx43 and Cx40 were generated by sequential transfection with the two different expression plasmids. HeLa cells were grown in minimal essential medium supplemented with 10% fetal bovine serum, 0.1 mM nonessential amino acids, 2 mM L-glutamine, 100 U/ml penicillin, and 100 µg/ml streptomycin (Life Technologies). Stable clones were selected by culturing in medium containing 500 µg/ml G418 (Life Technologies) and/or 150 µg/ml hygromycin (Calbiochem). Clones were screened for Cx40 or Cx43 expression by immunofluorescence and immunoblotting. Selected clones were maintained in the medium described above but supplemented with G418 (250 µg/ml) and/or hygromycin (75 µg/ml). For later identification, some cultures were tagged with cell tracker green (5-chloromethyl-fluorescein diacetate; Molecular Probes) (37). Tagged cells expressing one type of connexin and nontagged cells expressing the other type of connexin were mixed and seeded onto sterile glass coverslips placed in multiwell culture dishes (~10⁴ cells/cm²). Electrophysiological measurements were carried out on cells cultured for 1-3 days.

Anti-connexin antibodies. Commercially available mouse monoclonal antibody (Chemicon) directed against amino acids 252-270 of rat Cx43 was used at a dilution of 1:2,000 for immunoblots. Rabbit anti-Cx43 antiserum directed against a bacterially expressed Cx43 fusion protein (29) was used at a dilution of 1:200 for immunofluorescence. Rabbit anti-Cx40 antiserum directed against a bacterially expressed Cx40 carboxyl tail fusion protein (27) was used at a dilution of 1:4,000 for immunoblots. Rabbit polyclonal antibodies directed against a synthetic peptide immunogen corresponding to residues 316-329 of dog Cx40 were affinity purified as described earlier (23) and were used at a dilution of 1:200 for immunofluorescence.

Immunoblot analysis. Protein extracts from cells were prepared as described by Laing and Beyer (28). Aliquots containing 10 µg of protein were separated by SDS-PAGE on 12% polyacrylamide gels and blotted onto Immobilon-P membranes (Millipore). Immunoblots were developed with ECL (or ECL Plus) chemiluminiscence reagents (Amersham Pharmacia Biotech) following the manufacturer's procedures. Peroxidase-conjugated goat anti-rabbit IgG (1:5,000 dilution) or donkey anti-mouse IgG (1:4,000 dilution; Jackson ImmunoResearch Laboratories) was used as the secondary antibody. Rainbow molecular weight marker standards (Amersham Pharmacia Biotech) were used to calibrate the gels.

Chemical cross-linking. Cross-linking of connexins in unfractionated HeLa cell lysates was performed as described by Musil and Goodenough (32). Cells were lysed in 1% Triton X-100 in PBS at 4°C and then centrifuged at 100,000 g for 30 min at 4°C. Supernatants containing connexons (hexamers) were reacted with 2 mM disuccinimidyl suberate (DSS; Pierce) or DMSO for 30 min. Samples were quenched with 50 mM Tris (pH 7.5) for 15 min at 4°C, run on 7% SDS-polyacrylamide gels, and then blotted as described above.

Purification of His₆-tagged proteins. His₆-tagged proteins were purified from Triton X-100-solubilized supernatants using nickel-nitrilotriacetic acid (Ni-NTA) resin (Qiagen) as described by the company in the protocol for batch purification under native conditions with the following few modifications. To reduce nonspecific binding, the Ni-NTA resin was blocked for 15 min with 2.5% bovine serum albumin in lysis buffer; the concentration of imidazole in the lysis buffer was reduced to 1 mM, and all buffers contained 1% Triton X-100. Triton-extracted protein samples were incubated with resin for 2 h at 4°C. For protein analysis, 4-15% gradient SDS-polyacrylamide gels were run and immunoblotted as described above.

Immunofluorescent labeling of cells. Cells cultured on multiwell slides were stained as described earlier (30). CY3-conjugated goat anti-mouse or anti-rabbit IgG antibodies (Jackson ImmunoResearch Laboratories) were used as secondary antibodies. In double-labeling experiments, cells were incubated simultaneously with both anti-Cx43 monoclonal antibody and anti-Cx40 polyclonal antibody and then with secondary antibodies consisting of fluorescein-conjugated goat anti-mouse IgG (1:50 dilution) and CY3-conjugated anti-rabbit IgG (1:800 dilution) as described earlier (30).

Solutions. During experiments, the cells were superfused with bath solution containing (in mmol/l) 110 CsCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, and 10 HEPES (pH 7.4). The patch pipettes were filled with saline containing (mmol/l) 110 CsCl, 0.1 MgCl₂, 0.1 CaCl₂, 3 EGTA, and 10 HEPES (pH 7.2).

Electrical measurements. Glass coverslips with adherent cells were transferred to an experimental chamber perfused with bath solution at room temperature (21-23°C). The chamber was mounted on the stage of an inverted microscope (Olympus IMT2). Patch pipettes were pulled from glass capillaries (code 7052, A-M Systems) with a horizontal puller (Sutter Instruments). The resistance of the filled pipettes measured 1-1.5 M. Experiments were carried out on mixed cell pairs. A dual voltage-clamp method and whole cell recording were used to control the membrane potential of both cells and to measure currents (6, 36). Each cell was attached to a patch pipette connected to a separate micromanipulator (WR-88, Narishige Scientific Instruments) and amplifier (Axopatch 200). Initially, the membrane potentials of cell 1 and cell 2 were clamped to the same value: the voltage of cell 1 (V₁) = the voltage of cell 2 (V₂). V₂ was then changed to establish a transjunctional voltage (V_j) = V₂  V₁. Currents recorded from cell 2 represent the sum of two components, the junctional current (I_j) and the membrane current of cell 2; the current obtained from cell 1 corresponds to I_j. To measure I_j, both cells were held at the same holding potential (V_h), i.e., V_h = 0 mV. The voltage of one of the cells was then stepped to different levels (37). A bipolar pulse protocol was used as described previously (5, 33). The amplitudes of I_j were determined at the beginning [instantaneous I_j (I_j,inst)] and at the end of each pulse [steady-state I_j (I_j,ss)] to estimate the instantaneous and steady-state conductances (g_j,inst and g_j,ss, respectively). Because the "apparent instantaneous" current rapidly decays with larger V_j, g_j,inst and g_j,ss were normalized with respect to a 10-mV prepulse conductance (constant with time). We used this as our standard for g_j,inst and g_j,ss normalization. We used I_j,ss or g_jss to indicate junctional current or conductance at the 400-ms or 10-s time points, where g_jss represents an approximation of the steady state.

Signal recording and analysis. Voltage and current signals were recorded on chart paper (Gould RS 2400, Gould Instruments) and videotape (Neurocorder DR-384, Neuro Data Instruments). For off-line analysis, the current signals were filtered at 1 kHz (low-pass filter), digitized with a 12-bit analog-to-digital converter (DT21EZ, Data Translation), and stored with a personal computer. Data acquisition and analysis were performed with custom-made software (6, 25). Curve fitting and statistical analysis were performed using SigmaPlot and SigmaStat (Jandel Scientific), respectively. Macroscopic conductances were compared using the Mann-Whitney rank sum test. Unless otherwise noted, data are presented as means ± SE. Single channel currents were studied in low-conductance cell pairs as previously described (5, 6, 12, 36, 37). The single channel data were fit using the analytical approach of Kullman (26).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Double expression of Cx40 and Cx43 in HeLa transfectants. The properties of channels formed by cloned connexins can be examined by their exogenous expression in transfected HeLa cells, because these cells are virtually communication deficient. To examine the properties of Cx40 and Cx43 channels alone and after potential mixing with each other, we generated HeLa cell lines that were stably transfected with Cx40 or Cx43 alone (HeLa Cx40 and HeLa Cx43) or with both Cx43 and Cx40 (HeLa Cx43/Cx40) with the use of two different selectable markers. To facilitate isolation of expressed connexins, we also prepared a Cx43 construct in which a His₆ tag was appended to the carboxyl terminus. We isolated HeLa clones expressing the His₆-tagged Cx43 alone [HeLa Cx43(His)] or with Cx40 [HeLa Cx43(His)/Cx40].

View larger version (182K): [in this window] [in a new window]   Fig. 1.   Immunofluorescent detection of connexin (Cx) expression in transfected HeLa cells. A: Hela cells expressing connexin Cx40 (HeLa Cx40); B: Hela cells expressing Cx43 (HeLa Cx43); C-F: HeLa cells expressing both Cx43 and Cx40 (HeLa Cx43/Cx40). A, C, and E: cells reacted with rabbit anti-Cx40 antibodies followed by CY3-labeled anti-rabbit secondary antibodies. B, D, and F: cells were reacted with rabbit (B and D) or mouse (F) anti-Cx43 antibodies followed by CY3-conjugated (B and D) or fluorescein-conjugated (F) secondary antibodies. E and F: double-label immunofluorescence studies of HeLa Cx43/Cx40 cells confirming the substantial colocalization of these connexins. Bars, 15 µm for A and B; 12 µm for C and D; and 8 µm for E and F.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Immunoblot analysis of Cx40 (A) and Cx43 (B) in HeLa cells transfected with connexins. Whole cell lysates from parental HeLa cells or HeLa cells transfected with Cx40, Cx43, or Cx43 tagged with His6 [Cx43(His)] were resolved by SDS-PAGE, transferred to membranes, and blotted with anti-Cx40 (A) or anti-Cx43 (B) antibodies. Migration of molecular weight markers is indicated to the left of the blot.

View larger version (15K): [in this window] [in a new window]   Fig. 3.   Absolute abundances of Cx40 and Cx43 in HeLa transfectants. Immunoblots of whole cell lysates were performed as in Fig. 2. Blots also contained serial dilutions of bacterially expressed fusion proteins containing COOH-terminal portions of Cx40 or Cx43. Reaction products were quantitated using a phosphorimager, and abundances of Cx40 and Cx43 were determined by comparison to these standard curves. Each bar represents the mean value ± SE based on 2-4 determinations.

View larger version (65K): [in this window] [in a new window]   Fig. 4.   Cross-linking of connexins in HeLa Cx43/Cx40 cells. Triton X-100-soluble extracts (containing connexons) from HeLa Cx43/Cx40 cells were reacted with disuccinimyl suberate (DSS) cross-linker dissolved in DMSO (+) or were incubated with DMSO solvent alone (). These samples were resolved by SDS-PAGE, and Cx43 and Cx40 were detected by immunoblot analysis using rabbit anti-connexin antibodies. Migration of molecular weight markers is indicated to the left of the blot.

View larger version (34K): [in this window] [in a new window]   Fig. 5.   Affinity purification of His6-tagged Cx43 and associated Cx40. A Triton X-100-soluble extract from HeLa Cx43(His)/Cx40 cells (containing connexons) was affinity purified using a Ni-NTA column (as described in MATERIALS AND METHODS). Cx43 and Cx40 in the extract applied to column and in the eluted fractions were detected by immunoblotting using anti-Cx43 and anti-Cx40 antibodies. Migration of molecular weight markers is indicated to the left of the blot.

Voltage dependence of Cx43/Cx40 gap junction currents. Homotypic Cx40 or Cx43 channels have been characterized in primary cell cultures (36) and in transfected cells (1, 6, 8, 37). The formation of heterotypic channels containing Cx40 and Cx43 has been more recently demonstrated and analyzed (37).

View larger version (36K): [in this window] [in a new window]   Fig. 6.   Dependence of intercellular coupling on transjunctional voltage (Vj) of Cx43/Cx40 gap junctions. A-D: gap junction currents (Ij) elicited from HeLa Cx43/Cx40 cells using a bipolar pulse protocol. A and C: symmetrical current inactivation (A, short pulse; C, long pulse); B and D: polarity-dependent Ij inactivation (B, short pulse; D, long pulse). E and F: instantaneous () and steady-state plots (short pulse, ; long pulse, ) of normalized juctional conductance (gj) versus Vj. E: quasisymmetrical relationship; continuous line (short pulse), Boltzmann fit: initial Vj (Vj,0) = 83/72 mV, minimum gj (gj,min) = 0.33/0.28, and z = 1.5/1.6, for negative/positive Vj, respectively; dashed line (long pulse), Boltzmann fit: Vj,0 = 72/66 mV, gj,min = 0.28/0.23, and z = 1.7/1.8, for negative/positive Vj, respectively. F: asymmetrical relationship; short pulse, Boltzmann fit for negative Vj: Vj,0= 104 mV, minimum gj (gj,min) = 0.29, maximum gj (gj,max) = 1.02, and z = 1.7; long pulse, Vj,0 = 76 mV, gj,min = 0.27, gj,max = 1.0, and z = 2.5.

Voltage dependence of Cx43/Cx40-Cx40 and Cx43/Cx40-Cx43 gap junction currents. We cocultured cotransfected cells (HeLa Cx43/Cx40) with singly transfected cells (HeLa Cx40 or HeLa Cx43). This strategy allowed us to observe channels in which one hemichannel potentially contained two connexin types but the other contained only a single connexin (was homomeric) to determine whether this configuration would distinguish heteromeric forms from homotypic or heterotypic channnels.

View larger version (32K): [in this window] [in a new window]   Fig. 7.   Macroscopic properties of junctions in cell pairs formed between a cotransfected HeLa Cx43/Cx40 cell and a HeLa cell expressing only Cx40 or Cx43. A and B: HeLa Cx43/Cx40-HeLa Cx40 cell pairs. Top, Ij elicited in response to a series of voltage steps (Vj). Currents were measured from the Cx43/Cx40 cell in A and from the Cx40 cell in B. Bottom, instantaneous () and steady-state plots () of normalized gj versus Vj. A: quasisymmetrical relationship with Boltzmann fit: Vj,0 = 71/92 mV, gj,min = 0.39/0.28, and z = 2.2/0.7, for negative/positive Vj, respectively. B: asymmetrical relationship with Boltzmann fit for negative Vj: Vj,o= 103 mV, gj,min= 0.33, gj,max = 1.03, and z = 2.2. C: Ij elicited in response to a series of voltage steps (Vj) in HeLa Cx40/Cx43-HeLa Cx43 cell pairs. Top, symmetrical current inactivation; bottom, asymmetrical currents voltage dependence. Currents were measured from the HeLa Cx43/Cx40 cell. D: instantaneous () and steady-state plots () of normalized gj versus Vj. Boltzmann fit: Vj,0 = 82/87 mV, gj,min = 0.26/0.42, gj,max = 1.04/1.05, and z = 1.3/1.0, for negative/positive Vj, respectively.

Time- and conductance-dependent transition from asymmetric to symmetric voltage dependence. The total gap junction conductance between cell pairs often declined with time. As already indicated, the degree of asymmetry of I_j was variable and dependent on total gap junction conductance. In some cases, the changes of total gap junction conductance between cell pairs led to transformation from asymmetrical currents to symmetrical ones. This was observed for all cell types investigated (i.e., HeLa Cx43/Cx40-HeLa Cx43/Cx40, HeLa Cx43/Cx40-HeLa Cx40, and HeLa Cx43/Cx40-HeLa Cx43). Figure 8, A and B, illustrates an example of a mixed HeLa Cx43/Cx40-HeLa Cx40 cell pair. At the beginning of the measurements, the total conductance of the cell pair was ~3 nS (Fig. 8A, top). Stepping V_j to make the inside of the HeLa Cx40 cell positive resulted in a large I_j,inst with pronounced inactivation. Conversely, stepping V_j to make the inside of the HeLa-Cx40 cell negative led to a smaller I_j,inst with marginal or no inactivation. When spontaneous total conductance dropped to ~2 nS, I_j became more symmetrical (Fig. 8A, bottom).

View larger version (28K): [in this window] [in a new window]   Fig. 8.   A and B: gap junction conductance versus voltage dependence. A: recordings from a HeLa Cx43/Cx40-HeLa Cx40 cell pair. Top, total gj = 3 nS; bottom, total gj = 2 nS. B: steady-state plots of normalized gj versus Vj when total conductance was ~3 nS () and ~2 nS (). C and D: multichannel properties of "heterotypic" junctions formed between HeLa Cx43/Cx40 and HeLa Cx45 cells. C: responses of Ij to Vj steps. Pulses were applied to the HeLa Cx45 cell. Currents were measured from the HeLa Cx43/Cx40 cell. D: instantaneous () and steady-state plots () of normalized gj versus Vj. The steady-state gj exhibited pronounced Vj dependence gating when the HeLa Cx45 cell was negative and virtually no gating and/or reopening when it was positive. Continuous curves represent fitted Boltzmann equations with the following parameters: Vj,0 = 59 mV, gj,min = 0.05, gj,max = 1.01, and z = 2.9.

Properties of Cx43/Cx40-Cx45 gap junction currents. The gating properties of heteromeric Cx43/Cx40 channels were difficult to distinguish from homotypic forms of Cx40 or Cx43 and heterotypic Cx40-Cx43. This prompted us to try to unmask heteromeric gating by pairing Cx45-expressing HeLa cells with HeLa Cx43/Cx40 cells. The bipolar pulse protocol was used to study multichannel currents from a HeLa Cx43/Cx40-HeLa Cx45 cell pair (Fig. 8C). Stepping V_j to make the inside of the HeLa Cx45 cell negative or the inside of the HeLa Cx43/Cx40 cell positive resulted in an I_j,inst with pronounced inactivation. Conversely, stepping V_j to make the inside of the HeLa Cx45 cell positive and the inside of the HeLa Cx43/Cx40 cell negative led to an I_j,inst with distinct activation.

Simulation of macroscopic gap junction conductance involving Cx40 and Cx43 channels. We modeled our data to examine whether the g_j = f(V_j) relationship for cells cotransfected with Cx40 and Cx43 could be explained under some conditions where only heterotypic and homotypic channels were present.

View larger version (27K): [in this window] [in a new window]   Fig. 9.   Simulation of the Vj dependence of macroscopic gj observed in HeLa Cx43/Cx40 cells based on homotypic and heterotypic combinations of HeLa Cx43 and HeLa Cx40 cells (which necessarily contain only homomeric connexons). A: observed and predicted steady-state plots of normalized gj versus Vj for HeLa Cx43/Cx40-HeLa Cx43/Cx40 cell pairs. The solid line represents experimental data shown in Fig. 6C. B: observed and predicted steady-state plots of normalized gj versus Vj for HeLa Cx43/Cx40-HeLa Cx40 cell pairs. The solid line represents experimental data shown in Fig. 7A. These are smooth curves through the data points (shown with error bars denoting standard deviations), not Boltzmann fits. The dashed or dotted lines represent the predicted normalized conductances calculated from various combinations of homotypic and heterotypic Cx40 and Cx43 gap junction channels based on the data from Valiunas et al. (37) and the assumption that there are no heteromeric channels. For the homotypic populations in HeLa Cx43/Cx40 cells, we assumed equal abundance of Cx40 and Cx43. For the heterotypic population, we assumed that the two possible polarities (Cx43-Cx40 and Cx40-Cx43) always existed in equal amounts.

Conductances of single channels. To study single channel currents, we selected cell pairs with one or two to three operational channels. The pulse protocol adopted involved an inversion of the V_j polarity. Figure 10 illustrates experiments in weakly coupled pairs or in normally coupled pairs after advanced spontaneous uncoupling. Figure 10, A-C, shows data from HeLa Cx43/Cx40-HeLa Cx40 cell pairs, and Fig. 10, D-F, shows data from HeLa Cx43/Cx40-HeLa Cx43/Cx40 cell pairs.

View larger version (28K): [in this window] [in a new window]   Fig. 10.   A-C: single channel properties of gap junction channels between HeLa Cx43/Cx40-HeLa Cx40 cells. A and B: bipolar pulse protocol [voltage from cell 1 (V1) and from cell 2 (V2)] and associated single channel currents; the current from cell 2 (I2) was recorded from the HeLa Cx43/Cx40 cell. C: operation of a channel during a maintained Vj of 60 mV. The Cx40 cell was negative inside. Solid line, zero current level; dashed line, residual current level; right, current amplitude histogram. D-F: single channel properties of gap junction channels between HeLa Cx43/Cx40-HeLa Cx43/Cx40 cell pairs. D: bipolar pulse protocol (V1 and V2) and associated single channel currents (I2). E and F: operation of channels during a maintained Vj. The current histograms (right) revealed several conductance levels. For specific data, see text.

View larger version (27K): [in this window] [in a new window]   Fig. 11.   Histograms of single channel conductances. The smooth curves represent the theoretical fits of data to Gaussian distributions. In the case of HeLa Cx43/Cx40-HeLa Cx43/Cx40 cell pairs (A), the sum of 5 Gaussians was used; in the case of heterotypic HeLa Cx43/Cx40-HeLa Cx40 gap junctions (B), the sum of 4 Gaussians was used. For specific data, see text.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The potential physiological consequences of formation of mixed channels by Cx40 and Cx43 may have importance for understanding cardiovascular function, because these two proteins are coexpressed in a number of cell types. Conflicting data have been published regarding the heterotypic and heteromeric interactions of these proteins. The initial electrophysiological observations suggested that Cx40 and Cx43 did not interact to form heterotypic channels in Xenopus oocytes (7, 41). Other studies performed in HeLa cell transfectants utilized dye injection (14) or the induction of channels in approximated cells (17). Unfortunately, these approaches may not have allowed detection of heterotypic channel formation, because later studies by Valiunas et al. (37) revealed that Cx40 and Cx43 do form functioning heterotypic gap junction channels in transfected HeLa and RIN cells.

Recently, He et al. (18) have argued for functional mixing of Cx40 and Cx43. They studied A7r5 rat vascular smooth muscle cells, which normally coexpress Cx40 and Cx43. The macroscopic voltage dependence that they detected in these cells was weaker than that produced by either homotypic form, and some of the observed single channel conductances could not be readily explained as homotypic Cx43 or Cx40 forms. These authors concluded that their data provided evidence for heteromeric Cx40/Cx43 gap junction channels. However, they did not consider the possible existence of heterotypic Cx40-Cx43 channels, which we (37) recently detected and studied.

In the present study, we extensively analyzed the biochemical properties of HeLa cells coexpressing Cx40 and Cx43. Immunoblotting experiments showed that cells produced both Cx40 and Cx43 proteins, and quantitation of immunoblots showed that the cotransfected cells produced very similar amounts of the two proteins. Immunofluorescence microscopy showed that both Cx40 and Cx43 localized to virtually identical cell surface locations, consistent with their participation in the same gap junctions. Cotransfected cells were disrupted with Triton X-100 under conditions that solubilize connexons (hexameric hemichannels); chemical cross-linking of this material showed that both Cx40 and Cx43 formed oligomers in these cells. Finally, when the Triton X-100 extracts were affinity purified using Ni-NTA-Sepharose, Cx40 copurified with Cx43(His); this result implied that Cx43(His)-containing connexons also contained Cx40. Thus our cotransfected cells abundantly expressed both Cx40 and Cx43 and contained biochemically detectable heteromeric connexons. Our data extend the data of He et al. (18), who showed coimmunoprecipitation of Cx40 with Cx43 from A7r5 cells, and support their conclusion of heteromeric mixing. One caveat of both studies is that they are based on Triton X-100 solubilization, which has become a standard procedure to examine connexon assembly (3, 24, 32). However, this is not a quantitative procedure; moreover, as noted in earlier studies by Musil and Goodenough (31), some cellular connexin (including some material in gap junction plaques) is not solubilized by this procedure. Much of the Triton X-100-solubilized material may derive from connexons within the assembly/export pathway and from hemichannels and channels that have not been incorporated into plaques. An assumption is that the mixing of connexins to form heteromers is similar in the synthetic pathway and in the plasma membrane.

We also extensively analyzed the electrophysiological properties of HeLa cells coexpressing Cx40 and Cx43. Our double whole cell patch-clamp data are less striking than might have been anticipated in suggesting unique electrophysiological properties conferred by heteromeric mixing.

One of our findings was that in HeLa cell pairs with one or more member coexpressing Cx40 and Cx43, the junctional conductance was substantially less than in pairs where each cell was transfected with only a single connexin (Table 1). For example, the junctional conductance of the cotransfected HeLa Cx43/Cx40 cell pairs (14.3 nS) was only 56-66% of the conductance obtained in homotypic Cx40 or Cx43 cell pairs. These results are somewhat surprising compared with our connexin protein data. The HeLa Cx43/Cx40 cell pairs were prepared by transfecting Cx40 into the same HeLa Cx43 cell clone used for the homotypic studies. Thus the HeLa Cx43/Cx40 cells contain indistinguishable amounts of Cx43 compared with the HeLaCx43 cells plus the introduction of a similar amount of Cx40 (Fig. 3). We might have expected an increased (approximately doubled) conductance in the HeLa Cx43/Cx40 pairs. The observed reduction cannot be due to impaired trafficking of Cx43 produced by introduction of Cx40, because both connexins were detected in similar plasma membrane locations (Fig. 1). Therefore, it must be that some of the heteromeric Cx43/Cx40 connexons are nonfunctional as hemichannels or are unable to pair with other connexons to make functionally complete gap junctional channels.

We observed even further reduced conductances in the partial heterotypic cell pairs formed when HeLa Cx43/Cx40 cells were paired with cells expressing only Cx40 or Cx43 (14.3 vs. 7.6-8.0 nS; Table 1). These data suggest a reduced efficiency of channel formation/function for the heterotypic channel.

The single channel data are equivocal in terms of distinguishing between homotypic, heterotypic, and heteromeric forms. Many of the single channel events that we observed in pairs of HeLa Cx43/Cx40 cells were similar to ones that we detected in homotypic or heterotypic pairs of HeLa Cx40 and/or HeLa Cx43 cells; a relatively small number of events were unique. The conductances that are dissimilar from homotypic or heterotypic forms provide the strongest argument for heteromeric channel types, but interpretation of these data is unclear, because these events may also represent substates of any of the gap junction channels present or involvement of an endogenous connexin (such as Cx45). [In studies of cardiac myocytes, Elenes et al. (13) suggested that Cx45 might facilitate interactions of Cx40 and Cx43.] Thus the data do not unequivocally demonstrate unique conductances that can be defined as heteromeric Cx43/Cx40. Moreover, it is quite possible that heteromeric Cx43/Cx40 channels have similar or identical conductances to the homomeric/homotypic or homomeric/heterotypic forms, as speculated by Hopperstad et al. (19). These data differ from our previous observations of many different single channel events when Cx43 and Cx37 were coexpressed; this difference suggests the possibility that less functional heteromeric channels are formed between Cx40 and Cx43 than between Cx37 and Cx43.

Brink et al. (5) calculated that 196 different gap junction channel forms are potentially formed when two connexins are equally coexpressed and have similar affinities for each other. In contrast, the maximum number of conductive forms would be only 14 for a pair between a cell coexpressing two connexins and a partner cell expressing only a single connexin (and therefore containing only homomeric connexons). Therefore, one expectation might be that the event histogram of channel conductances for the cotransfectant cell pairs (HeLa Cx43/Cx40 paired with HeLa Cx43/Cx40) would show more conductive states than are detected between HeLa Cx43/Cx40 and HeLa Cx40 cells. The data shown in Fig. 11 qualitatively meet this criterion; however, in fact, there is only one additional conductive state demonstrable for the cotransfectant pairs relative to the HeLa Cx43/Cx40-HeLa Cx40 cell pairs. The conductive state missing in the HeLa Cx43/Cx40-HeLa Cx40 cell pairs is in the range of 75 pS, which is similar to homotypic Cx43 observed under our conditions (37). There are two alternative possible explanations for these data. First, it is possible that virtually all of the channels in the cotransfectants are heteromeric, but they function similarly to homotypic and heterotypic Cx43 and Cx40 channels. The conductances of the heteromers might not be distinguishable. Any of the peaks in the channel event histograms could represent multiple channel populations with similar or identical conductances but with very different connexin composition. Second, alternatively, it is possible that there is only a small population of heteromeric forms, which are not easily detected with our analysis; functional heteromer formation might be rare.

Interestingly, most, if not all, of the conductances in both histograms corresponded to conductances that are similar to the homotypic and heterotypic forms of Cx43 and Cx40. The least conductive peak in both histograms (33 and 37 pS) might represent channel activity of heteromeric forms. However, we cannot exclude involvement of subconductive states of either Cx40 or Cx43 channels or endogenous Cx45 channels.

Our new data regarding V_j dependence are consistent with previous reports (5, 18) utilizing cotransfected cell pairs in which the voltage dependence of coexpressing cells was shown to be broadened or less sensitive than the equivalent homotypic forms. We found that the junctional conductances observed in cells cotransfected with Cx40 and Cx43 showed weaker voltage dependence than their homotypic counterparts. A subset of cell pairs showed asymmetric voltage dependence similar to heterotypic Cx40-Cx43 channels. Similar observations were made in cell pairs where one hemichannel of the gap junction was homomeric (Cx40 or Cx43) and the other hemichannel was a cotransfected cell (HeLa Cx43/Cx40). Such cases with asymmetrical V_j dependence (as in Fig. 7, B and C) exhibited gating polarities corresponding to heterotypic Cx40-Cx43 channel gating polarities (37).

We tried to explain our data by fitting them with a combination of homotypic and heterotypic forms. The summarized data and fits shown in Fig. 9 do not distinguish between two possible situations where 1) the majority of the channels are heteromeric and behave like homotypic and/or heterotypic channels, or 2) there is only a small heteromeric channel population and the total junctional conductance is dominated by homotypic and heterotypic channel types. Such conclusions are also supported by the data from Cx45-Cx43/Cx40 gap junctions (Fig. 8, C and D). Interestingly, the macroscopic properties observed were indistinguishable from heterotypic/homomeric gap junction channels, i.e., the recorded currents, their V_j dependence, and gating polarities (Fig. 8, C and D) corresponded closely with those of heterotypic Cx40-Cx45 or Cx43-Cx45 gap junctions (37).

In terms of voltage gating, either the heteromeric hemichannel forms follow the behavior of their homomeric hemichannel counterparts or the heteromeric forms are functionally nonexistent.

It is possible that heteromeric mixing has a more profound influence on other properties of gap junction channels. For example, Gu et al. (16) presented data suggesting that the carboxyl tail of Cx43 can modulate the pH-dependent gating of Cx43 when the two connexins are coexpressed in Xenopus oocytes. Thus chemical gating is a potentially important physiological parameter with regard to heteromers. It is also possible that heteromeric Cx43/Cx40 channels might have altered permeability properties, because the two connexins individually form channels with some differences in permeability and selectivity (1, 2, 11, 38, 40), but a detailed characterization of these permeability/selectivity properties requires a full study of its own.

In summary, our data from expression of two major cardiac connexins (Cx40 and Cx43) in HeLa cells show biochemically that the two connexins can mix to form heteromeric connexons. However, reduced total conductances between pairs of coexpressing cells suggest that some heteromeric channels may be nonfunctional. Other electrophysiological analyses suggest that most properties of these cells can be understood based on the properties of homomeric and heterotypic channels. Single channel data revealed conductances that were consistent with the dominant channel forms being homomeric/homotypic or heterotypic-like. Moreover, studies of the cotransfected HeLa Cx43/Cx40 cells yielded g_j,ss = f(V_j) relationships that could be explained on the basis of homotypic and heterotypic channels alone, although the qualifier is that a high percentage of heterotypic channel types (between 30% and 70%) would be necessary to approximate the data. A related point is our observation that junctional conductances were initially asymmetrically voltage dependent but became symmetrically dependent as junctional conductance was reduced (Fig. 8, A and B). This is consistent with the presence of at least two distinct populations of voltage-dependent channels, one symmetric and other asymmetric. Thus either 1) heteromeric forms are the most prevalent channel type and have gating properties similar to heterotypic and homotypic channels, or 2) the population of functional heteromeric channels is so low that their gating behavior cannot be detected apart from the combination of homotypic and heterotypic channels. In either case, the heteromeric channels do not make a distinguishable contribution to voltage-dependent gating.