Emerging concepts regarding pannexin 1 in the vasculature

Abstract

Pannexin channels are newly discovered ATP release channels expressed throughout the body. Pannexin 1 (Panx1) channels have become of great interest as they appear to participate in a multitude of signalling cascades, including regulation of vascular function. Although numerous Panx1 pharmacological inhibitors have been discovered, these inhibitors are not specific for Panx1 and have additional effects on other proteins. Therefore, molecular tools, such as RNA interference and knockout animals, are needed to demonstrate the role of pannexins in various cellular functions. This review focuses on the known roles of Panx1 related to purinergic signalling in the vasculature focusing on post-translational modifications and channel gating mechanisms that may participate in the regulated release of ATP.

Introduction

A key role for purinergic signalling in vascular function has been understood for well-over half of a century. What is not clear is the “how” related to release of nucleotides to induce the multiple purinergic cascades. This is especially important for drug-targeting of the purinergic signalling cascade. It had been postulated for some time that release of nucleotides is almost exclusively an exocytosis event. However, recent work has begun to demonstrate pannexin channels as another mean of ATP release from cells that is both highly regulatable and coupled to activation of receptors. This concept is important as it potentially allows for a concentration specific release of nucleotides both spatially and temporally. Although there are three pannexins (Panx1, 2, 3), Panx1 is the most ubiquitous. This review focuses on what is currently known about Panx1 related to purinergic signalling in the vasculature, especially the resistance arteries.

On the importance of methodology and pharmacology in the study of pannexins

Since the initial discovery of Panx1 channels in 2003, great strides have been made in an effort to develop appropriate tools to study the role of these ATP-releasing channels in various cell types, tissue and organisms. From a pharmacological standpoint, Dahl et al. [1] defined Panx1 channels as a “sewer for drugs”, listing 37 drugs that have been reported to inhibit Panx1 channels in the past decade alone. The majority of these drugs are well-known blockers of other targets, such as organic anion transporters (probenecid), potassium channels (glibenclamide), chloride channel inhibitors (NPPB, DIDS), P2X₇ purinergic receptors (e.g. Brilliant Blue G, BzATP, suramin), gap junctions (flufenamic acid, 18α-glycyrrhetinic acid, carbenoxolone (CBX), connexin mimetic peptides, mefloquine) or inhibitors of the mitochondrial ATP efflux (bongkrekic acid, atractyloside) [1]. In an effort to uncover the role of Panx1 channels in vitro and in vivo, the scientific community mostly relied on CBX, mefloquine and probenecid. CBX has been used for over 25 years as a connexin-based channel blocker, but in 2005 it was one of the first agents shown to directly inhibit Panx1 currents [2]. In that study, the authors demonstrated a higher sensitivity of Panx1 channels to CBX compared with connexin-based channels, and this characteristic has been further used to help in differentiating between Panx1 and connexin-based channels. Similarly, mefloquine has shown to be more effective on Panx1 channels at low doses (~50 nM) while concentrations greater than 10 μM are required to inhibit connexin-based channels [3,4]. Nevertheless, these pharmacological properties have been assessed in overexpressing in vitro models, making it difficult to extrapolate the efficacy and specificity of these drugs in native cells or tissue, especially for tissues co-expressing connexins and pannexins such as the vasculature. On the other hand, probenecid does not affect connexin-based channels even at doses as high as 5 mM, while it effectively inhibits Panx1 channels with an IC50 of ~150 μM [5]. Probenecid was thus considered promising to use in the vasculature, but in reality, concentrations as high as 1–2 mM are required to observe an inhibitory effect in native cells or tissue, making the specificity of the drug questionable [5–8]. Of note, the efficacy of probenecid and CBX on Panx1 channel activity is severely impaired when other proteins are co-expressed, as shown with the potassium channel subunit Kvbeta3 [9]. This led to further question the influence of other proteins on the efficacy of those drugs in inhibiting Panx1 channels in native cells or tissue. With the growing need for more specific Panx1 inhibitors, and following the model of connexin mimetic peptides, the ¹⁰Panx1 peptide was developed in 2006 by the Surprenant laboratory with the hope to specifically target Panx1 channels [10]. Unfortunately, it was later evidenced that the ¹⁰Panx1 peptide also affects other targets such as connexin-based channels, and its effect can be replicated by other peptides of equivalent size, or by 1.5 kDa polyethylene glycol [11].

It has now become evident that the most commonly used “Panx1 pharmacological inhibitors”, mefloquine, probenecid and ¹⁰Panx1, have a wide range of effects in native cells and tissue [12]. The interpretation of functional studies only using these drugs is thus rather difficult and a substantial number of publications solely based on pharmacological inhibitors targeting Panx1 have led to misleading conclusions with regards to the involvement of Panx1. To alleviate this, molecular tools such as RNA interference and knockout animal models have been used to greatly benefit the field in the understanding of the physiological and pathological relevance of the ubiquitous protein that is Panx1 [6,8,10,13–16]. However, it should be noted that cell-specific knockout animal models should be utilized when possible because a compensatory increase in Panx3 protein has been reported in the global Panx1^−/− mouse ([17,18]). Given these considerations, the scientific community should be aware of the considerable challenges involved in the study of Panx1 channels with regards to the experimental conditions as well as the pharmacological tools. One should be cautious when drawing conclusions on the involvement of Panx1 channels by relying solely on “Panx1 pharmacological inhibitors” and complementary experiments using molecular tools such as siRNA targeting Panx1 (e.g. [6]) or cell-type specific knockout animals (e.g. [16]) should be implemented.

Post-translational modification of pannexins

The gating of pannexin channels by post-translational modification is an area of great biological interest. To date, much information has been gleaned about the numerous types of post-translational modifications that are found on pannexin channels, which include N-glycosylation [19–21] caspase cleavage [14,22], S-nitrosylation [23,24], and phosphorylation [16,19,25–27]. However, the mechanisms and stimuli that directly regulate pannexin channel activation and gating remain under intense investigation. Of all known pannexin post-translational modifications, S-nitrosylation and phosphorylation are indicated to play significant roles in the vasculature, largely since nitric oxide (NO) and kinase mediated signalling events strongly predominate constrictor/dilatator responses in veins and arteries [28,29].

The reversible post-translational attachment of NO moieties to cysteine residues, termed S-nitrosylation, has profound effects on protein function [12]. Previous reports suggest that ischemic conditions activate pannexins [15,30] and inhibitors of nitric oxide synthase (NOS) and antioxidants, such as DTT and GSH, can ameliorate channel activation [31]. In veins and arteries, NO acts as both a second messenger and a protein-modifying molecule [32]. Building on this concept, Panx1 was shown to be nitrosylated at cysteine 40 and 346 using a biotin switch assay in HEK (over-expressing) and endothelial (endogenous) cells [23]. Nitrosylation by the NO donor GSNO attenuated Panx1 channel activity and prevented release of ATP from cells [23]. Furthermore, site-specific mutagenesis of cysteine 40 and 346 to alanine residues conferred a constitutively open Panx1 channel [23], suggesting that S-nitrosylation of Panx1 may act as a negative regulatory mechanism in the vasculature. While it is still unknown which physiological stimuli directly regulate vascular S-nitrosylation events, it is alluring to think that a NO-dependent inactivation event exists for Panx1 channels to balance the Panx1 channel-dependent vasoconstriction in resistance arteries [6,16,33].

In addition to S-nitrosylation, the covalent addition of phosphate groups to proteins is a major regulator of protein function. While several phosphorylation sites for pannexins have been predicted, only indirect evidence exists to establish a possible mechanism for pannexin gating by specific kinases [25,26]. Current research implicates members of the Src family kinases (SFKs) as regulators of Panx1 channels. Panx1 channel activation in J778 macrophages was sensitive to both the SFK inhibitor PP2 and a small interfering peptide that targets a proline rich region on the P2X₇ carboxyl terminus (C-terminus) suggesting that Panx1 channel activation is mediated through P2X₇ receptors and SFKs [25]. Using a similar approach, Weilinger et al. [26] demonstrated that treatment with PP2 or a small interfering peptide against Panx1 tyrosine 308 prevents anoxia-induced Panx1 channel opening in rodent hippocampal brain slices. More recently, Billaud et al. demonstrated that pharmacological and genetic inhibition of an intracellular motif containing tyrosine 198 (Y198) prevents adrenergic-stimulated ATP release and vasoconstriction in resistance arteries [16]. These responses could be rescued in arteries from mice with Panx1 deleted specifically from the smooth muscle cells (SMCs) after transfection with wild-type Panx1, but not the tyrosine-containing motif mutant [16]. While the direct phosphorylation of Panx1 by SFK members has yet to be described, the reports by Weilinger et al. [26] and Billaud et al. [16] suggest that kinase phosphorylation events at Panx1 Y308 and Panx1 Y198, located within known SFK homology recognition sites (SH2 and SH3 respectively), could regulate Panx1 channel activity. Currently, the only indirect evidence for Panx1 phosphorylation is by Riquelme et al. [27] who demonstrated an increase in tyrosine and serine phosphorylation in electrically stimulated skeletal muscle. These and future investigations will help define post-translational mechanisms that regulate pannexin channel activity, as well as help identify new pathways that may contribute to the pathologies associated with vascular dysfunction.

Pannexin channel gating

Though noted for its permeability to ATP [34], Panx1 channels exhibit low ion selectivity and thus offers a broad, if transitory, window from the cytoplasm to the extracellular space [35,36]. A discrepancy has arisen regarding Panx1 channel conductance [37], which was originally reported at ~450–500 pS [34,35,38] but more recently reported on the order of 70–75 pS [39–41]. Panx1 channels are reported to be outwardly rectifying [34] with a pore size estimated at 17–21 Å (1 Å=0.1 nm) [42] (although the structure is not yet resolved) suggesting that Panx1 channels may have the potential to undermine the selective permeability of the cell membrane [36]. This large permeability is likely why Panx1 channels predominantly remain closed at physiologic voltages [34] and undergo extensive post-translational modification.

Panx1 channels open to varying extents in response to multiple stimuli [17,43]. However, the cytoplasmic C-terminus of Panx1 is critical to the maintenance of the channels closed state and may inhibit the Panx1 channel via a “ball-and-chain” mechanism, similar to that observed in other ion channels, whereby the C-terminus may enter the pore from the cytoplasmic side [22,44]. The well-described caspase cleavage of the C-terminus between amino acids 376 and 379 [14,22] renders Panx1 channels constitutively truncated and permanently active [14]. Among channels with similarly hypothesized ball-and-chain mechanisms, Panx1 is the first known to undergo cleavage of the C-terminus as a method to permanently remove this channel inhibition mechanism [22].

Potentially important in the vasculature (e.g. in conduction pathways), a positive feedback loop of ATP-induced ATP release has been hypothesized, stemming from Panx1 channel release of ATP in response to mechanical stress [45]. Subsequent ATP activation of P2Y receptors on the same and/or adjacent cells could initiate a signalling cascade that releases intracellular Ca²⁺, opening further Panx1 channels to propagate an extracellular ATP signal alongside an intracellular wave of Ca²⁺ and inositol triphosphate [45]. This possibility is reliant on Panx1 channels being sensitive to intracellular Ca²⁺, for which there appears to be common knowledge but sparse physiological evidence.

Additionally, ATP-induced ATP release is also thought to be mediated through Panx1 channels and P2X receptors, which have been found to co-localize, where Panx1 channels open in response to ATP-gated P2X₇ receptor activation [10,46,47]. This positive feedback loop of ATP-induced ATP release may in turn be negatively regulated by extracellular ATP. Substituted cysteine accessibility data suggest that the first transmembrane domain and first extracellular loop (EL1) line the Panx1 channel pore [48]. At high extracellular concentrations, ATP is believed to inhibit Panx1 by interacting with arginine 75 on EL1, perhaps providing a steric obstruction at the extracellular end of the pore to block ATP passage through the open channel [36]. This could have critical importance in providing negative regulation to the purinergic receptor-activated positive feedback loop of ATP-induced Panx1 ATP release [49]. However, outside of evidence provided from non-vascular cell culture, it is currently not clear how this observation might relate to the intact vasculature.

Other important mechanisms controlling pannexin channel activation include membrane depolarization, decreased circulating oxygen levels and mechanical stimulation. Although there are reports of Panx1 channels being responsive to mechanical stimulation, such as cell swelling or stretching, it remains an active area of investigation whether this mechanism occurs in vivo [34,47,50]. It has also been shown that lowered blood pO₂ levels could open Panx1 channels on erythrocytes, releasing ATP to induce upstream vasodilation [51]. In addition, a key regulator in maintaining blood pO₂ levels, the carotid body, also releases ATP through Panx1 channels on type II cells, potentially further amplifying activation of the afferent nerves [52]. Oxygen-dependent activation of Panx1 has been demonstrated in erythrocytes and carotid bodies, both of which regulate peripheral resistance, but the impact of oxygen content on endothelial cells (ECs) or vascular smooth muscle cells (VSMCs) Panx1 channel activation is unknown. Lastly, membrane depolarization by activation of TRPM5 channels, with the addition of extracellular KCl, or application of a depolarizing voltage step has been shown to activate Panx1 channels in vitro [2,53,54]. However, KCl-induced vasoconstriction in isolated arteries was found to be independent of Panx1 channel activity indicating that the mechanisms for Panx1 channel activation are likely either cell type specific or altered in vitro [6].

Of note, the stimuli suggested to activate Panx1 channels (mechanical stimulation [34,47,50], increasing extracellular potassium concentration [54], rise in intracellular calcium [45], membrane depolarization [2], increased intracellular redox potential [55], S-nitrosylation [23], oxygen deprivation [51,52]) could all be physiologically relevant to vascular physiology/pathophysiology. Hence, caution should be taken in the experimental design to study the role of Panx1 in the vascular wall, where vessel tension, potassium and calcium concentrations as well as oxygen tension should be monitored as close as possible to in vivo conditions. Such control over these parameters is crucial to accurately assess the role of Panx1 in the vasculature.

Pannexins in endothelial cells

Little is known about the role of Panx1 in ECs in health and disease. Most studies have focused on defining the expression patterns of pannexin isoforms in murine and rodent endothelium [56,57], while few studies have explored the signalling cascades that result in channel activation and subsequent ATP release [13,14]. Various stimuli like mechanical strain and EC agonists have been shown to trigger EC ATP release [6,13,58–59]. However, the role of pannexins, Panx1 in particular, in ECs remains understudied.

In ECs, pannexins are suggested to be involved in vasodilation and inflammatory cell adhesion [13,23]. Pannexin channels in inflammatory cells could play an important role during pathophysiological inflammatory conditions, such as participation in the secretion pathway of IL-1β [60] as well as transmission of a “find me” signal in early apoptotic cells [14]. Moreover, nucleotide release from ECs is demonstrated to have a vital role in regulating leucocyte adhesion to the vascular wall [61]. In human umbilical vein endothelial cells (HUVECs), pannexin channel-mediated ATP release is evoked by application of thrombin [13]. Prothrombin is cleaved to thrombin in injured vessels and then binds to the protease-activated receptor PAR-1 on ECs resulting in subsequent ATP release from these ECs reportedly due to increased intracellular Ca²⁺ concentration ([Ca²⁺]_i) [62,63]. These results are in agreement with observations in Xenopus oocytes of Panx1 activation through increased [Ca²⁺]_i [45] suggesting a potential role for Panx1 in regulating thrombin-dependent vascular response.

Mechanical injury of ECs during surgical harvest of saphenous vein grafts has been shown to result in impaired endothelial and smooth muscle functions and intimal hyperplasia ex vivo [64]. Preparation-induced injury was shown to cause ATP release and activation of P2X₇ receptors and Panx1 channels, with downstream signalling of the P2X₇ receptor implicated in the development of intimal hyperplasia. However, inhibition of purinergic receptors or pannexin channels with periodate-oxidized adenosine triphosphate (oATP) and probenecid, respectively, restored EC function and prevented development of intimal hyperplasia [65]. These results further suggest that the pannexin-mediated ATP release within ECs could be a key regulator for physiologic and pathologic EC function.

EC function can also be controlled by perivascular nerves releasing neurotransmitters that bind to specific receptors on the EC surface, such as acetylcholine and calcitonin gene-related peptide (CGRP) [66–68]. Recently, Gaete et al. [69] suggested that the neuropeptide and strong vasodilator CGRP activates Panx1 channels in mesenteric resistance vessels and primary cultures of mesenteric ECs. CGRP_8–37, a CGRP receptor antagonist, and probenecid, a Panx1 channel blocker, inhibited pannexin channel activation as measured by ethidium uptake [69]. Furthermore, CGRP treatment resulted in decreased Panx1 and eNOS expression [69]. This observation could explain the endothelial dysfunction observed during pathological conditions associated with inflammation through the reduction in eNOS levels [70–75].

Pannexins in vascular smooth muscle cells

Pannexin expression in VSMCs is limited to arteries less than 300 μm in diameter in mice [57]. These small arteries contribute significantly to the regulation of peripheral resistance; however, the mechanisms controlling peripheral resistance are not fully understood. ATP release in the cardiovascular system is known to participate in physiological and pathological conditions such as hypoxia-induced vasodilation, α1D-adrenergic receptor (α1D-AR) mediated vasoconstriction and hypertension [32]. Multiple non-vascular SMCs release ATP, such as SMCs from the colon and bladder, both of which express Panx1 [76–79]. Only recently have ATP-releasing pannexin channels emerged as regulators of vascular function, with recent evidence revealing an important and vital role for Panx1 expression in VSMCs of small systemic arteries [6,16]. Systemic arteries are highly innervated by sympathetic nerves that release norepinephrine (NE), which then activates α1D-ARs, expressed on the adjacent VSMCs, and results in vasoconstriction [80,81]. Interestingly, this vasoconstriction is dependent upon activation of the purinergic receptors, where blockade of purinergic receptors with either a non-specific antagonist, suramin, or a P2Y subtype specific inhibitor, reactive blue-2, significantly blunted adrenergic vasoconstriction [6]. In addition, degradation of extracellular ATP, the purinergic receptor agonist, with an ectonucleotidase, apyrase, profoundly reduced adrenergic mediated vasoconstriction [6]. Together these data implicate the need for regulated ATP release from the VSMCs. Indeed, inhibition of Panx1 with non-specific and specific inhibitors or genetic deletion of VSMC Panx1 resulted in significant reduction in vasoconstriction following stimulation with phenylephrine (PE), an α1-AR agonist, revealing a clear function for Panx1 in VSMC vasoconstriction [6].

The functional link between α1D-ARs and Panx1 channels was further explored using mimetic peptides against Panx1, two peptides against the cytoplasmic tail region and two peptides against the intracellular loop region [16]. Application of these peptides and further experiments using point mutations in a cell culture model resulted in the discovery of a necessary amino acid motif (YLK at amino acids 198–200) in the intracellular loop region of Panx1 for Panx1 channel-dependent ATP release following α1D-AR stimulation [16]. Interestingly, vasoconstriction via alternative mechanisms, such as endothelin-1 or serotonin, is not dependent upon Panx1 mediated ATP release [16]. Together these data strongly implicate a role for Panx1-dependent regulation in adrenergic driven vasoconstriction.

Due to the role of Panx1 channels during vasoconstriction of isolated resistance arteries, blood pressure was examined to determine if Panx1-dependent vasoconstriction affected peripheral resistance and, therefore, overall mean arterial pressure (MAP). Mice lacking VSMC Panx1 are indeed hypotensive suggesting the regulation of peripheral resistance is dependent on Panx1 expression in VSMCs [16]. In agreement with an adrenergic-dependent mechanism, the drop in blood pressure is more profound during the night when mice are more active and thus during greater sympathetic activity [16]. Injection of the intracellular loop mimetic peptide, which blocked PE-stimulated vasoconstriction in isolated arteries by possibly targeting the YLK motif, also caused a significant drop in MAP in C57Bl6 mice [16]. In congruence with this observation, hypotension is observed when guinea pigs are treated with mefloquine, a potent Panx1 channel inhibitor [3,82]. Additionally, hypotension is listed as a side effect for mefloquine, an FDA approved drug used to prevent and treat malaria. These data indicate that Panx1 expression in VSMCs of resistance arteries is vital for the development of adrenergic vasoconstriction and overall peripheral resistance.

Future directions

Pannexin channels represent an exciting and new area of investigation in the vasculature, with cell biology and biochemistry providing important insight into function. Although the pharmacology remains problematic, molecular tools such as siRNA, shRNA and KO animal models have led to stronger conclusions about the function of Panx1.

Abbreviations

α1D-AR

α1D-adrenergic receptor

CGRP

calcitonin gene-related peptide

CBX

carbenoxolone

EC

endothelial cell

EL1

first extracellular loop

HUVEC

human umbilical vein endothelial cell

MAP

mean arterial pressure

NO

nitric oxide

NOS

nitric oxide synthase

Panx1

pannexin1

PE

phenylephrine

SFK

Src family kinase

SMC

smooth muscle cell

VSMC

vascular smooth muscle cells