Nature of plasmalemmal functional “hemichannels”

Abstract

The molecular identity of the protein forming “hemichannels” at non-junctional membranes is disputed. The family of gap junction proteins, innexins, connexins, and pannexins share several common features, including permeability characteristics and sensitivity to blocking agents. Such overlap in properties renders the identification of which of these protein species actually establishes the non-junctional membrane conductance and permeability quite complicated, especially because in vertebrates pannexins and connexins have largely overlapping distributions in tissues. Recently, attempts to establish criteria to identify events that are “hemichannel” mediated and those to allow the distinction between connexin- from pannexin-mediated events have been proposed.

Here, I present an update on that topic and discuss the most recent findings related to the nature of functional “hemichannels” focusing on Connexin43 and Pannexin1.

Introduction

The molecular identity of the protein that performs the function of the so-called gap junction “hemichannel” in mammalian tissues has been debated. This is because the assigned properties of connexin hemichannels overlap with those of other proteins, especially those of pannexin1 channels [1-3]

In this review, an update on this topic is presented. The evidences and controversies related to the assignment of connexin43 hemichannels and pannexin1 channels as being the functional “hemichannels” are highlighted.

1. Gap junction family of proteins: connexins and pannexins

a. Connexins and connexons

In mammals, the gap junction proteins, connexins, consists of about 20 members. Connexins are generally believed to oligomerize in the Golgi-ER to form hexameric structures called connexons. Connexons are inserted into the plasma membrane where they join across the extracellular gap to a connexon from an adjacent cell. The pairing of connexons across extracellular space involves high affinity protein-protein interaction such that when the gap junction channel opens, there is no ionic leakage between the inside of the cell and extracellular space (see [4]). This results in an irreversible union of the connexons, as indicated by the requirement for hypertonic urea and EGTA treatment for any appreciable dissociation in biochemical experiments [5], the high shearing forces required for junction splitting using atomic force microscopy [6] and high forces necessary for hemichannel opening by stretching the carboxyl terminus of Cx43 [7]. As a consequence, gap junction retrieval from the surface membrane involves internalization of both the cell's own connexons and those contributed by its neighbor [8, 9].

The formation of gap junction channels differs from that of other channels in that the connexon subunits must travel, most likely by lateral diffusion, to the site of the gap junction ([10-12] but see [13, 14]). Thus, the connexons contributed by each cell must exist for some time as unpaired connexons or hemichannels in the nonjunctional membrane. In fact, the cell surface expression of unapposed connexons has been demonstrated by electron microscopic, biochemical, immunocytochemical, and/or electrophysiological methods [15-21]. Direct measurement of the initial gap junction channel forming between a pair of cells indicated that nonjunctional conductance did not change during the formation period, thus indicating that connexin hemichannels remained closed until secure docking with a connexon in an apposed membrane [21, 22].

The finding of nonjunctional permeability induced by expression of connexins, together with the biochemical, immunological, and functional studies showing that at least some connexon types may also be inserted in nonjunctional regions of the cell membrane [15, 23, 24] challenged the classical views about junctional proteins, by indicating that, at least under certain conditions, some of these proteins may also form functional channels in domains of the cell membrane that are not involved in cell contact. There is now substantial evidence that Cx46 and Cx50 form such hemi-channels when expressed in Xenopus oocytes [25] and possibly also in the lens [26], and it is widely believed that Cx43 can also form such structures in a variety of cell types [1, 27-31]. Other reports have proposed that this property is shared by several other connexins, including Cx23, Cx26, Cx30, Cx30.2, Cx31.9,Cx32, Cx35, Cx36, Cx37, Cx41.8, Cx45, Cx45.6, Cx55.5, and Cx56 [30, 32-40].

b. Pannexins and pannexons

Pannexins (Panx 1, 2, and 3), a newly discovered group of proteins, are classified as gap junction proteins due to their significant but low (∼20%) homology to the innexins, the gap junction proteins of invertebrates, although they bear no sequence homology with chordates gap junctions, the connexins [41-43]. Structurally, pannexins bear membrane topology similar to connexin, consisting of 4 transmembrane domains, two extracellular loops, a cytoplasmic loop, and cytosolic N- and C-termini. Differently from connexins, pannexins have 4 conserved cysteins, two in each extracellular loop and, at least for Panx1 and Panx3, one glycosylation site in the second and first extracellular loop, respectively [44, 45]. In terms of oligomers (pannexons), Panx1 has been shown to form hexameric structures, while Panx2 oligomerize as octamers [46]. For Panx3, however, this has not yet been determined.

Differently from connexin gap junctions that are localized at specific areas of appositional membranes, fluorescence and electron microscope labeling of pannexins revealed localization throughout the plasma membrane without the formation of canonical gap junction structures. Initial studies [47], however, indicated that exogenous expression of Panx1 (but not Panx2) could form gap junction channels with low current levels and weak voltage sensitivity after pairing Xenopus oocytes for a long period of time. Nevertheless, Bruzzone et al. [47, 48] also showed robust non-junctional currents in oocytes expressing Panx1, emphasizing that Panx1 promptly formed plasma membrane channels. These pannexon currents were shown to be insensitive to extracellular calcium, blocked by carbenoxolone (IC₅₀ 5 μM) and only moderately sensitive to flufenamic acid (FFA: 300 μM reducing currents by 30–40%) [48]. When exogenously expressed, Panx1 channels have been shown to have unitary conductance of 450-500 pS [49, 50], to be more sensitive to mefloquine than are connexin gap junction channels [50] and be blocked by probenecid, a compound that does not affect connexin channels [51].

It is now well accepted that Panx1 forms plasma membrane channels and does not form intercellular gap junction channels. Thus the term “hemichannel” as first used by Harris et al. [52] to describe the voltage gates of a gap junction channel and later used as a synonym for connexons, cannot be applied to Panx1 channels (see [3]).

2. Who performs the function of “hemichannel”: Connexin43 × Pannexin1?

There is considerable controversy regarding whether or not Cx43 forms functional hemichannels. This is probably because there has been no electrophysiological demonstration of functional expression of Cx43 hemichannels in Xenopus oocytes [53, 54], and thus published reports of presumed Cx43 hemichannel-mediated events appear to be at odds with the original publications. In addition, another complicating factor that generates a “hemichannel identity” problem is that Cx43 tissue distribution extensively overlaps with that of Panx1 which has also been proposed to account for the release of ATP as well as the uptake of fluorescent probes. Nevertheless, there are several ways to distinguish connexons from pannexons. One of them relies on the distinct intrinsic properties that pannexons and connexons display. Pannexons have been shown to open under physiological conditions, while most connexons have not. Notably, whereas pannexin channels can be opened at normal resting potential and in normal extracellular Ca²⁺ solutions, by mechanical stretch, elevated extracellular K⁺ concentration and following purinergic P₂ receptor stimulation [49, 55-58], connexons have only been demonstrated to open under supra- or patho-physiological conditions (no extracellular Ca²⁺ or Mg²⁺, depolarization exceeding +40 mV).

a. Cx43 hemichannel conundrum

The most compelling evidence for functional unpaired connexons made by Cx43 has come from HeLa cells overexpressing Cx43, in which currents were evoked by membrane depolarization above +20 mV, single channel conductance was about twice that of Cx43 gap junction channels [59]. In that study, the authors used HeLa cells expressing untagged and eGFP tagged Cx43, which displayed distinct gap junction channel properties, to show that hemichannels formed by these connexins displayed the predicted biophysical properties of half channels. Tagging eGFP to the C-terminus of Cx43 (Cx43-EGFP) resulted in gap junction channels that had the same fully open unitary conductance (about 110 pS) as wild-type Cx43 cell-cell channels, but did not show the 30 pS substate, consistent with reports that this mutant and truncations abolished the substate [60]. The fully open conductance of the Cx43 hemichannel was shown to be approximately twice that of the cell–cell channel, as predicted from series arrangement. Similar to their cognate gap junction channels, Cx43-eGFP hemichannels exhibited transitions between the fully open and the closed states with unitary conductances of about 220 pS. No transitions to the substate were recorded from Cx43-eGFP hemichannels. Furthermore, when eGFP was tagged to Cx43 N-terminal (eGFP-Cx43) no gap junction channel or hemichannel currents were recorded [59].

In native tissues, however, the identification of Cx43 hemichannels is not so clear. A recent study performed on acute brain slices proposed that under physiological conditions, Cx43 hemichannel-like openings in astrocytes corresponded to conductances of 165pS and 26pS [61]. Moreover, these events were found to be very rare, such that from a total of 700 astrocytes analyzed, only in four cell-attached and 14 inside-out patches were these Cx43-like channel openings observed [61]. Thus, the major concern regarding the assignment of functional connexons is that the lack and/or rarity of opening of connexons, at least those made by Cx43, is not consistent with the substantial dye uptake observed in many cell types [1, 62, 63].

Several groups provided evidence that under pathological conditions, “hemichannels” contribute to membrane permeabilization. Under inflammatory conditions, the high activity of “hemichannels” in astrocytes, as measured by the influx of fluorescent probes and by electrophysiological recordings, was attributed to openings of unnapposed Cx43 channels [64]. Twenty-four hours after exposure of cultured astrocytes to microglia conditioned media pre-stimulated with pro-inflammatory cytokines, intense La³⁺-sensitive ethidium bromide uptake and appearance of 220 pS channel activity were reported; surprisingly, under this same condition, total and surface Cx43 expression levels were greatly reduced, as was coupling between astrocytes. [64]. Despite this counter-intuitive finding, the authors proposed that Cx43 gap junction channels and hemichannels are oppositely modulated by pro-inflammatory cytokines through a p38-MAP kinase pathway.

Opposite regulation of Cx43 gap junction and its cognate connexons have also been proposed to occur in astrocytes following metabolic inhibition [59, 64] and in C6 glioma cells exposed to fibroblast growth factor (FGF)-2 and to LPS [65]. Interestingly, the opposite effects of LPS and FGF-2 in blocking Cx43 gap junction channels and in opening Cx43 hemichannels seems to be cell-type dependent. De Vuyst et al. [65] reported that in Cx43-transfected HeLa cells LPS and FGF-2 inhibited both types of Cx43 channels through a kinase pathway, while in C6 cells, these agents inhibit Cx43 gap junctions and promote the openings of Cx43 hemichannels through arachdonic acid generation [65]. The idea that cells respond differently to FGF-2 has been previously reported to occur in the CNS, where FGF-2 down-regulates Cx43 protein and mRNA levels in astrocytes from cortex and striatum but not from mesencephalon [66].

b. Pannexons can perform “hemichannel” function

The first evidence that native Panx1 channels have functions previously attributed to connexin hemichannels, came from studies performed in erythrocytes [55]. Since red blood cells lack connexins they are most suitable to test for the presence of “non-connexin hemichannel”-like pathway. Using RT-PCR, immunocytochemistry and western blot analyses, erythrocytes were shown to express Panx1; upon K⁺-induced depolarization, a carbenoxolone-sensitive pathway for the uptake of fluorescent dye and for the release of ATP were recorded [55]. Electrophysiological recording evidenced erythrocyte plasma membrane channels with properties (unitary conductance, reversal potential, activation voltage, permeability to ATP) similar to those recorded from oocytes expressing exogenous Panx1 [49].

In cells co-expressing connexins and pannexins, strong evidence that Panx1 but not Cx43 is the molecular substrate of “hemichannels” was recently provided using transgenic mice lacking Cx43 [67]. Electrophysiological recordings performed in wild-type and Cx43-null solitary astrocytes indicated that the amplitudes of voltage-activated currents measured in Cx43-null astrocytes were identical to those recorded from wild-type cells. Also, no difference in terms of membrane permeability to fluorescent dyes and to ATP was recorded in astrocytes from the two genotypes. However, the voltage activated currents, dye uptake, and ATP release were mostly eliminated by treating Cx43-null astrocytes with Panx1-short interfering RNA [67].

c. When pannexons and connexons co-exist: who opens first?

Using pharmacological agents to discriminate between pannexons and connexons and cells with selective deletion of Cx43 (Cx43-null mice) or Panx1 (Panx1 shRNA), Garre et al. [68] provided evidence that 2 hrs after FGF-1 stimulation, increased membrane permeability to Lucifer yellow and to ethidium bromide was entirely mediated by Panx1 channels. At a later time point (7 hours post FGF-1), however, increased permeability was attributed to both Panx1 channels and Cx43 hemichannels [68]. Similarly to the reported effects of FGF-2 (see above), FGF-1 treatment was proposed to cause an increase in Cx43 hemichannel activity, as measured by membrane permeability to fluorescent dyes, which was paralleled by a 40% decrease in total Cx43 protein level and 20% reduction in surface (biotinylated) Cx43 protein expression level [68].

Based on the effects of pharmacological agents and measurement of the amount of ATP released during FGF-1 exposure, Garre et al. [68] proposed that FGF-1-induced vesicular release of ATP would activate P2X₇ receptors leading to opening of Panx1 channels. Pannexons would then provide an additional route for the release of this nucleotide, thus contributing to a positive feedback loop [68]. Although the signaling pathway that mediates the effects of FGF-1 on Cx43 gap junctions and hemichannels was not investigated in that study, it seems to involve the carboxyl-terminus domain. HeLa cells expressing full-length Cx43 but not cells expressing tail-less Cx43 were reported to display increased Cx43 hemichannel activity following 7 hour FGF-1 stimulation [31]. However, contrary to the results obtained in astrocytes [68], increased Cx43 surface expression with no changes in total protein expression levels was reported for HeLa cells treated with FGF-1 [31].

Final Considerations

In this review, by focusing on connexin43 and pannexin1, I have summarized the evidence that functional “hemichannels”, defined as large non-junctional channels, exist and have also indicated some of the inconsistencies related to the nature of these so-called functional “hemichannels”.

Undoubtedly, a small fraction of connexons are located in non-junctional areas and they may under certain circumstances operate as “hemichannels”, releasing biologically active molecules. However, one has to keep in mind that there are other non-junctional channels that bear properties consistent with data supporting functional “hemichannel” involvement, as is the case for pannexons. We are in a time where the identification of which protein mediates the flux of signaling molecules through the plasma membrane has important implications for the understanding of physiological processes and the prevention of pathological events involved in cell-cell communication.

Highlights.

• Recent findings related to the nature of functional “hemichannels” are presented.

• Distinction between pannexons and connexons

• “Hemichannels” assignment: Evidences and controversies.