Connexin hemichannel and pannexin channel electrophysiology: How do they differ?

Abstract

Connexin hemichannels are postulated to form a cell permeabilization pore for the uptake of fluorescent dyes and release of cellular ATP. Connexin hemichannel activity is enhanced by low external [Ca²⁺]_o, membrane depolarization, metabolic inhibition, and some disease-causing gain-of-function connexin mutations. This manuscript briefly reviews the electrophysiological channel conductance, permeability, and pharmacology properties of connexin hemichannels, pannexin 1 channels, and purinergic P2X₇ receptor channels as studied in exogenous expression systems including Xenopus oocytes and mammalian cell lines such as HEK293 cells. Overlapping pharmacological inhibitory and channel conductance and permeability profiles makes distinguishing between these channel types sometimes difficult. Selective pharmacology for Cx43 hemichannels (Gap19 peptide), probenecid or FD&C Blue #1 (Brilliant Blue FCF, BB FCF) for Panx1, and A740003, A438079, or oxidized ATP (oATP) for P2X7 channels may be the best way to distinguish between these three cell permeabilizing channel types. Endogenous connexin, pannexin, and P2X₇ expression should be considered when performing exogenous cellular expression channel studies. Cell pair electrophysiological assays permit the relative assessment of the connexin hemichannel/gap junction channel ratio not often considered when performing isolated cell hemichannel studies.

Connexins as Hemichannels

Once the second mammalian gap junction protein was cloned from a rat heart cDNA library, the family of gap junction proteins was given the name “connexins” [1,2]. The notion that the undocked hexamer of connexin subunits, historically called a “connexon”, could form a functional plasmalemma ion channel, i.e. a “hemichannel”, was first reported when attempts to exogenously express the newly cloned lens connexin46 (Cx46) in Xenopus oocytes caused them to lyse [3]. Since the discovery of Cx46 hemichannels, essentially every connexin tested has been induced to form open hemichannels when presented with favorable conditions of low extracellular calcium ([Ca²⁺]_o) and positive membrane potentials (V_m) [4]. Whether connexin hemichannels open under physiological conditions, i.e. negative V_m and 1–2 mM [Ca²⁺]_o, is less certain. A physiological role for lens Cx46 and Cx50 hemichannels has been proposed since the lens is an avascular tissue and the microcirculatory homeostasis of the lens depends on nutrient, electrolyte, and water flow through plasmalemmal channels [5–7]. No connexin hemichannel has received more attention than connexin43 (Cx43) hemichannels. Cx43 is the most abundantly expressed connexin in the human body, being expressed in essentially every tissue with the exception of certain cell types like erythrocytes, skeletal muscle fibers, and spermatozoids [1,8]. The positive correlation between Cx43 expression in macrophage cell lines and the ability of 100 μM ATP^4- to permeabilize their cell membranes led to the initial speculation that Cx43 hemichannels form the ATP-release channel in macrophages [9]. However, recent evidence of ATP permeabilization of Cx43 knockout macrophages and discovery of the pro-inflammatory role of purinergic P2X₇ and pannexin 1 (Panx1) channels in immune cells provide sufficient reason to question this interpretation [10–12]. Fluorescent dye uptake into cultured Novikoff, normal rat kidney (NRK), human embryonic kidney (HEK) cells, and cortical astrocytes was later proposed to occur via putative Cx43 hemichannels [13–15]. Of paramount importance were the observations that metabolic inhibition activated cardiomyocyte and astrocyte Cx43 hemichannel-mediated dye uptake with eventual loss of membrane integrity and cell death [14,15]. Astrocyte permeabilization, dye uptake, and lysis apparently occurred via Cx43 hemichannels since astrocytes cultured from conditional or germ-line Cx43 knockout mice remained impermeable to Lucifer Yellow (LY) dye after 6 hrs of metabolic inhibition. Metabolic inhibition presumably increased Cx43 hemichannel activity by recruitment of hemichannels to the cell surface, a process that is modulated by Cx43 dephosphorylation and S-nitrosylation [16]. Protein Kinase C (PKC) activation reduces dye uptake by phosphorylation-dependent changes in Cx43 channel conductance and permeability [17]. Thus, there is supportive evidence for a pathophysiological role of Cx43 hemichannel activation during metabolic stress, but physiological activation of connexin hemichannels remains controversial.

The pathophysiological role of connexin hemichannel induced cell membrane permeabilization and death has taken on new meaning with the association of human disease-linked connexin mutations to the genesis of abberant hemichannel activity [18]. Of the 21 human connexin genes, 10 have been linked to inherited human diseases including neuropathies, deafness, skin diseases, oculodentodigital dysplasia (ODDD), and cardiac arrhythmias [19,20]. Many connexin mutations result in loss of gap junction function owing to trafficking deficiencies (i.e. gap junction plaque formation) or channel malformation (non-functional or communication-deficient channels) [21], whereas abnormal connexin hemichannel activity should be considered a gain-of-function mutation. Thus, when considering the functional consequences of a disease-linked connexin mutation, hemichannel activity should be considered as a possible mechanism in addition to the dominant or recessive inhibitory effects of mutant connexin proteins on the formation of homologous and heterologous wild-type connexin gap junction channels.

Pannexin and Purineric P2X₇ Receptor channels

Pannexin-1, 2, and 3 (Panx1, Panx2, and Panx3) are vertebrate homologues of the invertebrate innexin gap junction proteins and are abundantly expressed in the central nervous system [22]. They possess a similar membrane topology to the connexins consisting of cytoplasmic N- and C-termini, four transmembrane domains, two cysteine-containing extracellular loops, and one cytoplasmic loop [23]. Panx1 is ubiquitously expressed throughout the body, while Panx2 expression is restricted to the CNS. Despite the similar membrane topology, the pannexins share little sequence homology with the connexins and possess only two extracellular loop cysteines instead of three (so does Cx23). Only Panx1 induced large whole cell membrane currents when expressed in Xenopus oocytes and was initially reported to mediate intercellular communication [22]. Hence pannexin channels were called “hemichannels”, denoting their similarity to the connexon hemi-gap junction channel, i.e. connexin hemichannel. Recent evidence disputes the claim that Panx1 forms functional gap junctions, presumably due to the presence of an E2 N-linked glycosylation site that is essential to surface expression [24]. The present day consensus is that the pannexins do not form functional gap junctions and the hexameric Panx1 membrane channel should, therefore, not be referred to as a hemichannel [25].

The P2X(1-7) receptor family form ligand-gated ion channels that are activated specifically by ATP (unlike the P2Y G-protein coupled receptors that are also activated by ADP and other nucleotides) [26]. The P2X receptors form nonselective cationic ion channels that are permeable to Ca²⁺ and thus contribute to intracellular Ca²⁺ signaling in response to ATP release. The P2X₇ receptor channel is unique among the P2XRs in that it possesses a long cytoplasmic C-terminal domain that contributes to the formation of an additional open “dilated” pore state that results in cell permeabilization and influx of fluorescent dyes with molecular weights of ≤ 800 daltons, e.g YO-PRO-1 [27,28]. Low external divalent cation (e.g. Ca²⁺ and Mg²⁺) concentrations potentiate the ATP activation of P2X₇ receptors, but will not activate the channel in the absence of the agonist [29]. P2X₇ receptor activation first induces a cationic permeability that excludes large organic cations like N-methyl-D-glucamine (NMDG) and YO-PRO and subsequently induces the dilated pore state that dramatically increases large cation permeability [30]. The trimeric P2X₇ receptor subunit possesses cytoplasmic N- and C-termini, two transmembrane domains, and one extracellular loop with numerous disulfide bonded cysteine residues. There is speculation that Panx1 mediates the formation of the large pore state of the P2X₇ receptor since Panx1 co-expression, knockdown, or pharmacological blockade correlated with dye uptake in P2X₇ expressing cells [31,32]. Panx1 also co-immunoprecipitates with P2X₇ when both proteins are expressed in HEK cells, although this association was observed with transient overexpression of an epitope-tagged Panx1 protein [31]. In contrast, one recent study failed to detect a decrease in cellular dye uptake with Panx1 RNAi silencing or pharmacological inhibition of prospective connexin hemichannel or Panx1 channels [33]. Furthermore, structure-function mutagenesis of the P2X₇ pore-forming second transmembrane domain increased the permeability to Cl⁻ ions and the anionic dye fluorescein-5-isothiocyanate (FITC) while decreasing the permeability of the cationic ethidium bromide (EtBr) dye of open P2X7[G345C] mutant receptors, consistent with the intrinsic large pore-forming ability of the P2X7 receptor [34].

Thus, activation of P2X₇ and/or Panx1 channels will result in cell permeabilization, intracellular Ca²⁺ transients, and uptake of fluorescent dyes such as ethidium bromide (EtBr), propidium iodide (PI), Lucifer Yellow (LY), and YO-PRO, as will connexin hemichannels (LY permeability will be limited for the weakly cation-selective connexin hemichannels like Cx40, Cx46, and Cx50 since LY is anionic). Connexin hemichannel conductances range from 17 to 350 pS depending on the connexin type (e.g. 220 pS for Cx43) [4], and reports of the Panx1 channel conductance vary from 300 – 475 pS to recent observations of a ≈70 pS channel [35–38]. A 400–420 pS channel has been described that fits the pharmacological profile of the P2X₇ receptor permeabilization pore in macrophages and transfected HEK293 cells [39,40]. Thus, ambiguities about the channel conductances and permeabilities of these three channel types makes distinguishing between them difficult, if not impossible.

External Ca²⁺ Gating of Connexin Hemichannels

The typical method for demonstrating the presence of connexin hemichannels is to voltage clamp connexin-expressing isolated cells bathed in a divalent-free saline and vary the membrane potential from negative to positive potentials since low [Ca²⁺]_o (< 1 mM) and positive V_m conditions increases the hemichannel open probability (P_o) [3–5,8,13,14,41]. Native connexin hemichannel P_o is significantly reduced when [Ca²⁺]_o is elevated to physiological levels, e.g. 2 mM. Macroscopic current – voltage relationships (e.g. −80 to +80 mV) and differential high – low [Ca²⁺]_o whole cell current measurements are routine ways to quantify putative connexin hemichannel currents. The 50% inhibitory concentration constant (IC₅₀) for [Ca²⁺]_o closure of Cx46 hemichannels expressed in Xenopus oocytes is 380 μM, 5.3 mM for [Mg²⁺]_o, and may be as low as 100 μM [Ca²⁺]_o for Cx50 hemichannels [5,41]. Intracellular acidification also closes Cx46 hemichannels [4,42]. An external Ca²⁺-gate is thought to minimize connexin hemichannel P_o under physiological conditions, though the molecular identity of the connexin hemichannel calcium gate is still the subject of intense investigation (e.g. Cx26, Cx46). The external Ca²⁺-gate of Cx26 and Cx32 hemichannels is commonly proposed to be formed from a ring of acidic amino acid residues located on the connexin extracellular loops (E1 and/or E2) [43,44]. The extracellular location of a connexin hemichannel gate was confirmed by a substituted cysteine accessibility method (SCAM) experiment using the thiol-reactive maleimidobutyryl-biocytin (MBB) compound on an L35C mutant Cx46 hemichannel in the presence or absence of 5 mM [Ca²⁺]_o [45]. Experimental examination of a Charcot Marie Tooth disease (CMTX) D178Y Cx32 mutation led to the proposition that the hexameric arrangement of E2 D169 and D178 residues form an extracellular binding pocket for six Ca²⁺ ions that accounts for the [Ca²⁺]_o sensitivity of Cx32 hemichannels, albeit with a relatively low IC₅₀ of 1.3 mM [44]. Structure-based molecular dynamic simulations and experimental site-directed mutagenesis of Cx26, particularly of disease-associated mutations, approaches to connexin hemichannel gating and permeation are helping to delineate the molecular bases for the differential [Ca²⁺]_o sensitivities of specific connexin hemichannels [43,46].

Pharmacological blockade of connexin, pannexin, and P2X₇ channels

The application of known chemical gap junction channel blockers, fluorescent dye uptake assays, and exogenous expression systems are commonly used to associate a particular connexin with the presence of an apparent hemichannel current or molecular (dye/ATP) uptake/release pathway. Chemical classes of gap junction channel blockers include long-chain alkanols (e.g. heptanol, octanol), volatile anesthetics (e.g. halothane), long chain cis-unsaturated fatty acids and related amides (e.g. oleic acid, oleamide), glycyrrhizic acid derivatives (18-α/β glycyrrhetinic acid, carbenoxolone (CBX)), fenamates (e.g. flufenamic acid), and 2-aminophenoxyborate [47–56]. Essentially all of these lipophilic compounds affect the activity of other membrane ion channels or have other physiological effects which limit their utility as selective gap junction channel blockers. The antagonist properties of a variety of large pore channel blockers is summarized by Spray et al. (2006) [57], who also discusses the complexity of distinctly identifying the molecular origin of plasmalemmal permeabilization pores. CBX is often considered the best chemical uncoupling agent for cardiac gap junctions owing to a relative lack of pharmacological effects on the cardiac action potential [58]. However, CBX also blocks the recently identified pannexin1 channels at lower concentrations (5 μM) than Cx43 gap junctions, complicating the use of this agent as a pharmacological identifier of connexin hemichannel activity [56,59]. Selective blockade of Panx1 channels is now possible, however, with probenecid or food dye FD&C Blue #1 (Brilliant Blue FCF, BB FCF) with IC₅₀s of 150 and 0.27 μM, respectively [60,61]. BB FCF may be especially useful since 100 μM BB FCF had no effect on connexin (Cx32E₁43, Cx46) hemichannels or purinergic P2X₇ receptor channels, which are readily activated by benzoylbenzoyl-ATP (BzATP) and inhibited by oxidized ATP (oATP) [60]. P2X₇ receptors are relatively resistant to the P2 antagonist suramin, are slowly and irreversibly inhibited by oATP, and are reversibly inhibited by BBG, A740003, and A438079 [62,63]. Selective pharmacological blockade to distinguish connexin hemichannels from Panx1 and P2X₇ channels is not readily achievable with known chemical gap junction uncoupling agents. Perhaps the best way to distinguish a P2X₇ channel from a Panx1 channel or connexin hemichannel is by selective pharmacological blockade with oATP, A740003, or A438079 for P2X₇ channels and probenecid or BB FCF for Panx1 channels.

Another approach to the “selective” blockade of connexin channels was the development of “gap” peptides. Gap26 and Gap27 peptides were designed to mimic the extracellular loop 1 and 2 (E1 and E2) connexin domains and prevent gap junction formation by binding to the undocked connexin hemichannel prior to gap junction formation [64,65]. These connexin mimetic peptides are most effective at inhibiting neo-gap junction formation and fail to inhibit the chimeric Cx32E₁43 hemichannel activity despite the presence of the specific Cx43 E1 and E2 peptide sequences [66]. The Gap26 and Gap27 peptides partially inhibit Cx43 hemichannels, but also pannexin 1 channels [66,67]. Complicating matters further, a Panx1 mimetic peptide, ¹⁰panx1, partially inhibits Cx46 hemichannels despite the lack of decameric peptide sequence homology [66]. These Gap26, Gap27, and ¹⁰panx peptides are marketed by commercial vendors. The bottom line is that sequence homology, or the lack thereof, is not a reliable predictor of connexin hemi-, gap junction, or Panx1 channel blocking activity. However, one recently developed Gap 19 connexin cytoplasmic loop mimetic peptide has the unique property of inhibiting Cx43 hemichannels without affecting gap junction communication or Panx1channel activity [68]. The effect of Gap 19 on P2X₇ channels has not been examined.

Exogenous expression systems for the study of connexin hemichannels

The presence of three molecularly distinct channels capable of facilitating ATP release, fluorescent dye uptake, and plasmalemmal Ca²⁺ permeabilization, along with the knowledge that pharmacological gap junction blockade, with the possible exception of Gap19, is poorly selective among P2X₇, Panx1, and connexin hemichannels creates a dilemma for studying endogenous or exogenously expressed hemichannels. The two most abundant cellular systems for the exogenous expression of connexins are Xenopus oocytes and communication-deficient cell lines like HeLa or murine neuro2a (N2a) cells [69–71]. These preparations were originally developed for the study of newly cloned connexin gap junction proteins, but have been adapted to the study of connexin hemichannels since the concept of their existence emerged [3]. Xenopus oocytes express an endogenous connexin38, similar to mammalian Cx37, that is capable of heterogeneous coupling with Cx43 and other connexins and must, therefore, be knocked down with Cx38 antisense RNA injections prior to injection of the connexin mRNA to be studied [72]. HeLa cells exhibit a low level of endogenous gap junction coupling, presumably mediated by Cx45, which may complicate interpretation of connexin-specific functional gap junction interactions [73,74]. N2a cells are essentially devoid of endogenous connexin expression [70], but reportedly express Panx1 and P2X₇ [75,76].

Another mammalian cell line that is gaining in popularity, particularly for the biophysical study of Panx1 channels, is human embryonic kidney 293 (HEK293), and their SV40 large T antigen transfected variant, HEK293T, cells. The HEK293 cell line is commonly used for functional studies of exogenously expressed ion channels and receptors and only infrequently utilized for gap junction studies, e.g. hemichannel function of Cx26 and Cx30 deafness mutations [77–79]. Biophysical studies in HEK293T cells have demonstrated that Panx1 channels are activated by C-terminal caspase cleavage of an auto-inhibitory domain, even in the presence of 2 mM [Ca²⁺]_o and [Mg²⁺]_o, or inhibited by S-nitrosylation [80,81]. Although the original source of these cells is the American Type Culture Collection (ATCC) accession code CRL-1573, there are conflicting reports about the immunoreactive expression of Cx43 in HEK293 cells [82,83]. Despite the inability to detect Cx43 gap junctions by Western blot or immunohistochemical staining in HEK293 cells, low levels of endogenous electrical coupling have been recorded [83–85]. Stong et al. [78] reported that the HEK293 cell line lacks endogenous connexin expression, citing a methods chapter on exogenous connexin expression in mammalian cells [86], although the cited work made reference to HeLa cells, not HEK293 cells, and no connexin immunoblots were performed on HEK293 cells until later [82,83]. Since electrical coupling and Cx43 immunolabeling has been detected in HEK293 cells, endogenous connexin expression in HEK293 cells should not be considered null. HEK293 cells do not endogenously express P2X₇ receptor channels although they reportedly express Panx1 [87,88].

This laboratory tested a population of HEK293T cells in the 1990s and recently tested HEK293 cells obtained from two different laboratories (Fig. 1A) for endogenous gap junction coupling and measurable levels of electrical coupling were always detected. The level of Cx43 expression may vary dependent on the source of the HEK293 cells after decades of culturing, yet high levels of gap junction conductance (g_j) were readily measured even in the lower Cx43 expressing variety of HEK293 cells we have assayed (HEK-2). We attempted to knockdown the endogenous Cx43 expression by stable transfection with a shRNA construct directed against human Cx43 (Origene HuSH-29 #SKU-TR312771; human GJA1 29mer shRNA). Four different Cx43 shRNA constructs and one scrambled shRNA control were transfected into HEK293-2 cells and fourteen Cx43 and two control shRNA stable clones were selected with puromycin (2 μg/ml). Clones were immunoblotted for Cx43 protein with tubulin as a loading control as previously described [89]. Representative Western blots of WT HEK293-2, the best Cx43 shRNA clone (#80-1), and one scrambled control are presented in Fig. 1B. These Western blot experiments were performed in triplicate and Cx43 protein levels were quantified by densitometry scans and the Cx43 shRNA construct reduced Cx43 protein expression by 76% relative to wild-type HEK293-2 cells (Fig. 1C). ANOVA analysis indicated that mean Cx43 relative expression levels were not significantly different (F-value = 0.08). We measured the g_j values in HEK-2 and the selected Cx43 shRNA HEK293 #80-1 cell pairs by conventional double whole cell patch clamp techniques, corrected for junctional series resistance errors as previously described [90], and the average g_j values measured 33.9 ± 10.8 nS, n = 6 for the untransfected HEK-2 cells and 3.6 ± 0.7 nS, n = 7 for the Cx43 shRNA knockdown #80-1 cells. Thus, despite a ¾ decrease in Cx43 expression and a 90% reduction in g_j, we were not able to generate a HEK293 cell clone that was devoid of endogenous Cx43 expression and electrical coupling. Gap junction plaques could be detected as small puncta by immunocytochemical methods between cultured wild-type HEK293-2 cells (Fig. 1D).

Figure 1.

A, An immunoblot for Cx43 showing a variable amount of endogenous Cx43 expression in HEK293 cells obtained from two different research laboratories. B, An immunoblot for Cx43 from the low Cx43 expressing HEK293-2 cells, a stable clone (80-1) of the same cell line after knockdown of Cx43 with an anti-Cx43 shRNA, and a scrambled control shRNA. Tubulin was used as a loading control. C, Quantification of the relative Cx43 expression levels from the HEK-2 cells, the Cx43 shRNA HEK-2 cell clone, and the scrambled shRNA control from three immunoblots like the one displayed in panel B. D, Anti-Cx43 antibody (Millipore #AB1728) was immunolocalized in HEK-2 cells with an Alexa Fluor555 conjugated secondary antibody. Gap junctions appeared as small puncta (arrows) located between adjacent HEK293 cells.

Relative preference of a connexin to form hemichannels or gap junctions

Typically connexin hemichannels are studied in isolated cells, which biases the system in favor of hemichannel formation since the lack of cell-cell contact prevents existing hemichannels from docking to form functional gap junctions. The ability of a particular connexin to form gap junctions instead of hemichannels is rarely considered in the same preparation. John et al. [14] stated that “we cannot absolutely exclude that hemichannel density in isolated myocytes is increased by the isolation procedure”. A structural study of lens Cx50 suggests that there are at least 10-fold more gap junction channels than hemichannels in lens fiber cells [7,91]. Since connexin hemichannels are generally sensitive to [Ca²⁺]_o, we measured the Ca^2+-sensitive nonjunctional membrane currents and gap junction currents (I_j) in paired HEK293-2 cells and the Cx43 shRNA knockdown #80-1 clone. At least one presumed Cx43 hemichannel was evident during whole cell current recordings from an HEK293-2 cell in 0 mM [Ca²⁺]_o saline during the activating voltage step pulse from −40 to +30 mV (upward arrow) that sometimes remained open after stepping the membrane potential back to −40 mV (downward arrow) (Fig. 2A). After addition of 2 mM CaCl₂ to the bath, the observed hemichannel activity decreased (Fig. 2B) and the ensemble averaged currents revealed a decrease in whole cell current of 15.8 pA during the −40/+30/−40 mV pulse protocol (Fig. 2C). A similar 14.9 pA current decrease was seen with the addition of 2 mM [Ca²⁺]_o in the partner cell (data not shown). By comparison, the slope gap junctional conductance (g_j) of this wild-type HEK293-2 cell pair during a ±120 mV V_j ramp measured 5.9 nS, > 25 times higher than the Ca²⁺-sensitive membrane conductance of 225 pS. Assuming a Cx43 hemichannel conductance (γ_HC) of 220 pS and a Cx43 gap junction channel conductance (γj) of 110 pS, these values correspond to a single open Cx43 hemichannel and 54 gap junction channels in this HEK293 cell pair. The 2.0 mM [Ca²⁺]_o-sensitive current averaged 0.85 ± 0.49 nS (mean ± SEM, n = 12) in the HEK293-2 cells and 0.38 ± 0.16 nS in the #80-1 cells (n = 10). By comparison, the g_j values were 40 and 10 times higher than the endogenous [Ca²⁺]_o-sensitive whole cell current in these HEK and 80-1 cells. If one factors in the respective Cx43 gap junction channel and hemichannel conductances (γj and γ_HC) of 110 and 220 pS [4], then the approximate gap junction channel to hemichannel ratio is between 20 to 80. This means that only 1.25–5% of the Cx43 channels exist as hemichannels. A 1% occurrence of Cx43 hemichannels in an adult ventricular cardiomyocyte containing approximately 10,000 open gap junction channels (g_j ≈ 1 μS, [92]) suggests that a cardiomyocyte may contain a hundred hemichannels, more than enough to cause Na⁺ and Ca²⁺ overloading and membrane depolarization if activated [14].

Figure 2.

A, A whole cell patch clamp recordings from an HEK293-2 cell during repeated runs of a −40 to +30 mV voltage step under low (nominally zero) [Ca²⁺]_o conditions. Activation of a membrane channel that remained open upon return to −40 mV was frequently evident. B, A whole cell recording from the same HEK-2 cell after the addition of 2 mM CaCl₂ to the bath saline illustrates the reduction in observed endogenous channel activity. C, Ensemble averages of the five current traces displayed in panels A and B illustrating the difference in whole cell currents between 0 and 2 mM [Ca²⁺]_o. D, The junctional current – voltage relationship acquired from this HEK-2 cell coupled to a partner HEK-2 cell via Cx43 gap junctions. The ratio of [Ca²⁺]_o-sensitive nonjunctional membrane current to gap junction current was 0.04 in this HEK-2 cell pair.

Perhaps the best evidence for abberant mutant connexin hemichannel activity being the primary cause of a disease pathology is the Cx26 G45E mutation. The Cx26 G45E mutation was identified in patients with hereditary hearing loss and keratitis-ichthyosis-deafness syndrome (KIDS) and, when expressed Xenopus oocytes or HEK cells, induced hemichannel currents and cell death [18,78,93,94]. The mechanistic basis for the cellular death induced by the Cx26 G45E mutation appears to be the increased Ca²⁺ permeability of the mutant hemichannel, not a decrease in the [Ca²⁺]_o-sensitive gating of the hemichannel as observed with the A40V mutation [95]. A mouse model of the Cx26 G45E mutation was generated that recapitulated numerous symptoms of KIDS, thus providing a pathophysiological link for this abberant Cx26 hemichannel mutation to the human disease condition [96]. The functional data acquired using this molecule-to-mouse approach provides the strongest evidence that the aberrant hemichannel activity and Ca²⁺ permeability of the Cx26 G45E mutation is the basis for the disease pathologies associated with KIDS.

Concluding Remarks

Connexins are capable of forming hemichannels that are characterized by possessing a γ_HC twice that of the measured γ_j of their respective homomeric gap junction channels, are permeable to monovalent cations, Ca²⁺, ATP^4-, metabolites, fluorescent dyes and molecules up to 500–1,000 Da in molecular weight, are sensitive to the usual lipophilic gap junction uncouplers and cytoplasmic acidification, and increase their open probability with submillimolar [Ca²⁺]_o, positive membrane potentials, and metabolic inhibition. Panx1 channels and P2X₇ receptor channels exist in many cell types and distinguishing between them and connexin hemichannels is difficult since they may possess similar nonselective cation conductances and molecular permeabilities to Ca²⁺, ATP^4-, and fluorescent dyes like YO-PRO. Panx1 and connexin hemichannels possess overlapping sensitivities to CBX and mimetic peptides despite the lack of sequence homology, with the possible exception of Gap19 which blocks Cx43 hemichannels but not gap junctions. Panx1 channels are not gated by [Ca²⁺]_o, are activated by caspase-dependent proteolytic cleavage, inhibited by S-nitrosylation, and apparently do not form gap junctions owing to glycosylation of their second extracellular loop. P2X₇ channels are also not activated by low [Ca²⁺]_o, but their sensitivity to the natural ATP agonist is enhanced by low external divalent cation concentrations. Perhaps the best way to distinguish between connexin hemichannels and Panx1 or P2X₇ channels is pharmacologically since Panx1 channels are specifically inhibited by probenecid and the FD&C Blue #1 (Brilliant Blue FCF, BB FCF) and Brilliant Blue G (BBG) dyes with micromolar and nanomolar sensitivities, respectively. P2X₇ channels may also be inhibited by BBG, but are selectively activated by (benzoly)benzoyl ATP (BzATP), reversibly inhibited by A740003 and A438079, and slowly and irreversibly inhibited oxidized ATP (oATP). Many immune and neuronal cell types express connexins, pannexins, and purinergic receptors. So mammalian cell lines used to exogenously express and electrophysiologically study any one of these channels should be screened for endogenous connexin (e.g. Cx43), Panx1, and P2X₇ protein expression by real-time PCR and immunoblot assays. Immunohistochemical labeling of cells to detect the presence of gap junctions is the least sensitive assay and may not detect the presence of Cx43 despite an abundance of electrical coupling in HEK293 cells as an example. Electrophysiological assays like patch clamp channel or whole cell voltage clamp recordings are the most sensitive assays since these approaches are able to detect a single open channel. Finally, we estimate that only 1–5% of the wild-type hexameric connexin oligomers exists as a function hemichannel, relative to the abundance of gap junction channels found in those same cells, and that studying connexin hemichannels in isolated cells biases the results in favor of hemichannel formation and may overestimate the functional relevance of their activity in an artificial system where gap junction formation is not possible.