|Home | About | Journals | Submit | Contact Us | Français|
Purinergic signaling plays distinct and important roles in the CNS, including the transmission of calcium signals between astrocytes. Gap junction hemichannels are among the mechanisms proposed by which astrocytes might release ATP; however, whether the gap junction protein connexin43 forms these “hemichannels” remains controversial. Recently, a new group of proteins, the pannexins, have been shown to form non-selective, high conductance plasmalemmal channels permeable to ATP, thereby offering an alternative for the “hemichannel” protein. Here we provide strong evidence that in cultured astrocytes pannexin1 but not connexin43 forms “hemichannels”. Electrophysiological and fluorescence microscope recordings performed in wild-type and connexin43-null astrocytes did not reveal any differences in “hemichannel” activity, which was largely eliminated by treating connexin43-null astrocytes with pannexin1-siRNA (Panx1-KD). Moreover, quantification of the amount of ATP released from wild-type, connexin43-null and Panx1-KD astrocytes indicates that down regulation of pannexin1, but not of connexin43, prevented ATP release from these cells.
It has been widely proposed that connexin43 (Cx43), the main gap junction protein expressed in astrocytes, can under specific conditions form functional “hemichannels”, providing a transmembrane pathway for the diffusion of ions and relatively large molecules (reviewed in Harris, 2007; Spray et al., 2006). Evidence in support is primarily pharmacological, showing blockade of uptake of fluorescent dyes (Lucifer Yellow, ethidium bromide, propidium iodide) and of release of intracellular molecules (ATP, glutamate) by compounds known to block gap junction channels; additionally, electrophysiological recordings have demonstrated the presence of large (>200pS, as expected for Cx43 hemichannels) conductance channels in astrocytes (Contreras et al., 2003; Retamal et al., 2007).
Recently, a newly discovered group of proteins, the pannexins (Panx 1, 2, and 3), were cloned from mammalian tissues. Pannexins were classified as gap junction proteins due to their significant but low (~20%) homology to the innexins, the gap junction proteins of invertebrates; they bear no sequence homology with connexins, the gap junction proteins of chordates (reviewed in Scemes et al., 2007).
It is becoming apparent that none of the pannexins readily forms intercellular channels (but see Bruzzone et al., 2003), and that Panx1 in particular forms functional plasmalemmal channels that display properties similar to those that have been attributed to connexin hemichannels (reviewed in Dahl and Locovei, 2006; Spray et al., 2006; Scemes et al., 2008).
In the CNS, Panx1 mRNA and protein were reported to be present in both neuronal and glial cells in vivo and in vitro (Huang et al., 2007a; Ray et al., 2005, 2006). Panx1 forms large conductance (400-500pS) non-selective channels that are permeable to ATP, carboxyfluorescein and YoPro (Bao et al., 2004; Locovei et al., 2006a, 2007) and are sensitive to compounds known to block connexin channels, including carbenoxolone (CBX), flufenamic acid (FFA) and mefloquine (MFQ) (Bruzzone et al., 2005; Iglesias et al., 2008).
Because the overlapping pharmacology reported for pannexins and connexins may confound identification of the molecular substrate of “hemichannel” activity in astrocytes, we have compared the electrophysiological properties and membrane permeability to dyes of astrocytes prepared from wild-type and Cx43-null neonatal mice. We here show for the first time that cultured astrocytes display functional Panx1 channels that are activated by membrane depolarization or following P2X7R stimulation. These channels are sensitive to CBX and MFQ and allow permeation by YoPro and ATP. Because no differences in the activation properties of “hemichannels” were observed between wild-type and Cx43-null astrocytes and because Panx1 siRNA reduces the occurrence of these channels, we conclude that Panx1 is more likely the molecular substrate for “hemichannel” activity in these cells.
We used primary cultures of cortical astrocytes derived from neonatal wild-type (WT) and Cx43-null mice (offspring of Cx43 heterozygotes in C57Bl/6J-Gja1 strain; at least two litters per experiment were used). Animals were maintained at the Albert Einstein College of Medicine; the AECOM Animal Care and Use Committee has approved all experimental procedures used in these studies. Cortices were separated from whole brain embryos (E19-E20) and after meninges removal, tissues were trypsinized (0.1% trypsin at 37°C for 10 min). Cells from each animal were collected by centrifugation and pellet suspended in DMEM supplemented with 10% FBS and 1% antibiotics and seeded in 60mm culture dishes. Genotype of individual cultures was determined by PCR on tail DNA (Dermietzel et al., 2000). Astrocytes were maintained for 2-3 weeks in culture (100% humidity; 95% air; 5% CO2, 37°C) at which time about 95-98% of the cells were immunopositive for glial fibrillary acidic protein.
Solitary WT and Cx43-null astrocytes were plated on cover slips 24-48hrs prior to recordings. Whole cell patch clamp recordings were performed as previously described (Iglesias et al., 2008). Briefly, cells were bathed in external solution containing (mM): NaCl 147, Hepes 10, glucose 13, CaCl2 2, MgCl2 1 and KCl 5, pH 7.4. The pipette solution contained (mM): CsCl 130, EGTA 10, Hepes 10, CaCl2 0.5. Activation of Panx1 channels by voltage was performed using a 10 second ramp protocol from holding a potential of -60mV to +100mV. To analyze the participation of Panx1 channels in agonist-induced P2X7R activation, astrocyte membrane potential was held at -60mV and the P2X7R agonist BzATP (50μM) was superfused for 5-10 seconds, condition that was sub-threshold for total activation of Panx1 channels. After the first response to the agonist, the gap junction channel blockers carbenoxolone (50μM; CBX) and mefloquine (100nM MFQ) were superfused for 5 min prior to the addition of the P2R agonist. Electrophysiological recordings were accomplished using an Axopatch 200B amplifier and pClamp9 software was used for data acquisition and analysis.
Astrocytes (WT and Cx43-null) plated on glass bottomed dishes (MatTek) were bathed for 5 min in phosphate buffered solution (pH 7.4) containing the cell-impermeant dye YoPro-1 (5μM). Cells were then exposed to a solution containing 300 μM BzATP and 5μM of the dye. [Higher concentrations of agonist than those used in the electrophysiological studies are necessary for long lasting and full activation of Panx1 channels and for optimal detection of YoPro fluorescence]. YoPro fluorescence intensity was measured during 500 sec BzATP stimulation, as previously described (Suadicani et al., 2006). The effects of a gap junction channel blocker (MFQ: 10nM) and of a P2X7 receptor antagonist (BBG: 1μM) on BzATP-induced dye uptake were also tested. YoPro fluorescence was captured using a CoolSNAP-HQ2 CCD camera (Photometrics) attached to an inverted Nikon microscope (Eclipse TE-2000E) equipped with a 20X dry objective and FITC filter set using Metafluor software.
Astrocytes were treated with 50nM small interference RNA corresponding to the mouse pannexin1 sequence, as well as with scrambled sequences, using 6μl/1.5ml oligofectamine reagent (Invitrogen), as previously described (Locovei et al., 2007). After overnight exposure, transfection reagents were removed and cells incubated for 30hr in DMEM-FBS medium before use in electrophysiological and dye-uptake studies.
Confluent cultures of WT and Cx43-null astrocytes plated in 35mm dishes were washed twice in PBS and then exposed for 3 min to PBS containing 300μM BzATP. After complete removal and washout of the agonist, cells were bathed in PBS for 2 min before collection of samples of BzATP-induced ATP release. [Agonist concentration and incubation times are based on dye uptake measurements and were optimal for detection of released ATP]. For measurements of intracellular ATP levels, cells were lysed with Tris-buffered solution containing 1% Triton-X and supernatants of whole cell lysates used. The amount of ATP in samples were measured as previously described (Striedinger et al., 2007), using the luciferin/luciferase assay (Molecular Probes - Invitrogen) and a plate reader luminometer (Veritas, Turner Instruments). The amounts of ATP in the samples were calculated from standard curves and normalized for the protein concentration, using the BCA assay (Pierce, Rockford, IL).
Whole cell recordings from solitary astrocytes indicated the presence of voltage activated outward currents in WT cells; the current amplitudes increased at voltages above +20mV and were significantly attenuated by carbenoxolone (CBX) and mefloquine (MFQ), two compounds previously described to block Panx1 channels more effectively than connexin gap junction channels (Bruzzone et al., 2005; Cruikshank et al. 2004, Iglesias et al., 2008). At membrane potentials of +100mV, current amplitudes (1062.0±60.6pA, N=16 cells) were significantly reduced by 61% (419.0±38.2pA, N=12 cells) and by 55% (478.9±41.6pA, N=5 cells) following exposure to 100nM MFQ and 50μM CBX, respectively (P<0.01, ANOVA followed by Dunnett’s test; Fig. 1).
To verify that channel activity recorded in WT astrocytes was not due to activation of currents attributed to Cx43 hemichannels (Contreras et al., 2003; Retamal et al., 2007; Kang et al., 2008), we performed electrophysiological recordings as described above using cortical astrocytes derived from Cx43-null mice. Similarly to what we found for WT astrocytes, Cx43-null cells displayed voltage activated outward currents (Fig.1A). At membrane potentials of +100mV, current amplitudes (1074.0±84.1pA, N=8 cells) recorded from Cx43-null cells were significantly reduced by 64% (386.7±41.1pA; N=5 cells) and 55% (485.5±62.5pA; N=3 cells) following exposure to MFQ (100nM) and CBX (50μM), respectively (P<0.01, ANOVA followed by Dunnett’s test). The amplitudes of voltage-activated currents measured in Cx43-null astrocytes (1074.0±84.1pA; N=8 cells) were virtually identical to those recorded from WT cells (1062.0±60.6pA; N=16 cells; P=0.89, T-test). These data, summarized in Figure 1B, indicate that the channels contributing this outward current at positive potentials are not Cx43 hemichannels.
To obtain evidence for the molecular identity of CBX- and MFQ-sensitive outward currents, we treated astrocytes with Panx1 siRNA. Under the condition of Panx1 knockdown (KD), voltage-activated currents recorded from both WT and Cx43-null astrocytes were greatly attenuated. Following 48-72hr Panx1-KD, current amplitudes measured at +100mV were significantly reduced to 308.0±31.1pA (N=11 Cx43-null cells) and to 373.8±42.5pA (N=8 WT cells) compared to those of untreated (1062.0±60.6pA, N=16 cells; P<0.01, T-test; Fig.1B) or scrambled Panx1 siRNA-treated (1028.0±89.4pA, N=9 cells; supplemental Fig. S1) WT astrocytes. Exposure of Panx1-KD Cx43-null astrocytes to MFQ (100nM) did not further attenuate voltage activated outward currents (230.2±24.2pA, N=5 cells; P>0.05, T-test). These data strongly suggest that Panx1 forms the channels responsible for current activation at positive potentials in astrocytes. [Evidence that Panx1 but not Cx43 forms non-junctional channels in Xenopus oocytes is shown in supplemental Fig. S2].
Activation of P2X7 receptors by high ATP concentration and by the synthetic agonist BzATP leads to opening of Panx1 channels in a macrophage cell line and in Xenopus oocytes co-expressing these two proteins (Iglesias et al., 2008; Locovei et al., 2007; Pelegrin and Surprenant, 2006).
As illustrated in Figure 2A, 5 sec exposure to BzATP (50μM) elicited inward currents in astrocytes held at -60mV. The inward current elicited by BzATP was biphasic, with an initial small inward current followed by a larger one; the second but not the first component was blocked by 100nM MFQ (BzATP: 470.6±18.9pA, N=10 cells; MFQ: 130.8±17.3pA, N=8 cells; P<0.001, T-test; Figs 2A,B). The remaining MFQ-insensitive currents are likely mediated by flux of ions through the cation channel of the P2X7R receptor itself, as we have recently shown for the J774 macrophage cell line (Iglesias et al., 2008).
Similarly to WT astrocytes, the second phase of BzTAP-induced currents (458.2±15.7pA, N=5 cells) in Cx43-null astrocytes was also significantly decreased (131.6±25.4pA, N=4 cells; P<0.01, T-test) by 100nM MFQ (Figs 2 C,D). Knockdown of Panx1 greatly attenuated BzATP-induced currents in Cx43-null (238.9±40.0pA; N=5 cells) and in WT astrocytes (202.5±30.10pA, N= 6 cells), compared to untreated (470.6±18.9pA, N=10 cells; P<0.001, t-test; Figs 2B, D) and scrambled Panx1 siRNA treated (431.6±30.1pA, N=7 cells; supplemental Fig. S1) WT cells.
To evaluate the contribution of Panx1 to astrocyte membrane permeabilization, we exposed astrocytes to BzATP and measured the influx of YoPro-1. As we previously showed for WT spinal cord astrocytes (Suadicani et al., 2006), BzATP induced YoPro uptake in WT and Cx43-null cortical astrocytes that was prevented by BBG, MFQ and by Panx1 siRNA (Figs 3A,B). After 500 second exposure to 300μM BzATP, YoPro fluorescence intensity increased 1.14±0.005 fold (N=72 cells) in WT astrocytes and 1.14±0.004 fold (N=90 cells) in Cx43-null cells. Five min pre-incubation with BBG (1μM) reduced YoPro uptake in WT (1.05±0.002, N=90 cells) as well as in Cx43-null (1.02±0.003, N=90 cells). MFQ (10nM) also prevented YoPro uptake in both WT (1.03±0.003, N=90 cells) and Cx43-null (1.03±0.003, N=90 cells) astrocytes. Similarly, knockdown of Panx1 also prevented BzATP-induced YoPro uptake in WT (1.06±0.001, N=84 cells) and in Cx43-null (1.03±0.01, N=100 cells) astrocytes (Fig. 3B). Scrambled Panx1 siRNA had no effect on BzATP-induced dye uptake from WT astrocytes (1.13±0.004, N=300 cells) and there was no YoPro uptake in the absence of BzATP stimulation (1.03±0.002, N=100 cells).
To evaluate whether deletion of the Gja1 gene affected the expression levels of P2X7R and Panx1, western blot analyses were performed on whole cell lysates of astrocyte cultures derived from at least three different litters of heterozygous matings. No significant changes in protein expression levels of P2X7 and Panx1 were detected when comparing astrocyte cultures derived from WT and Cx43-null sibling mice (supplemental Fig. S3). Thus, these results strongly suggest that the persistence of “hemichannel” currents and membrane permeability to YoPro seen in Cx43-null astrocytes (Figs(Figs11 and and3)3) are not related to increased expression levels of P2X7 receptors and Panx1 that might in principle compensate for the loss of putative Cx43 hemichannels.
Using the luciferin-luciferase assay, we recorded significantly lower BzATP-induced ATP release from Cx43-null (119.0±18.2nM, N=7 experiments) compared to WT astrocytes (205.8±20.5nM, N=10 experiments; P<0.05, T-test; Figure 4A). However, cytosolic ATP levels in Cx43-null (6.96±0.67μM, N=7 experiments) were lower compared to those of WT astrocytes (11.4±1.4μM, N=8 experiments; P<0.05, T-test; Fig. 4B). This result suggests that the lower extracellular ATP levels recorded from Cx43-null astrocytes may reflect their lower cytosolic levels rather than lack of Cx43 hemichannels.
Nevertheless, in terms of fold of ATP release following receptor stimulation in relation to basal non-stimulated condition, no difference was found between WT and Cx43-null astrocytes. After BzATP stimulation, a 2.9 fold increase from basal levels was measured in extracellular ATP from cultured WT astrocytes (from 72.23±7.98nM ATP to 205.8± 20.5nM ATP; N=9-10 experiments; P<0.001, T-test; Fig. 4A). Similarly, a 2.4 fold increase in ATP release was also recorded from Cx3-null astrocytes following BzATP stimulation (from 49.5±6.7 to 119.0±18.2nM ATP, N=5 independent experiments; Fig. 4A; P<0.05, T-test). This result showing similar fold changes in ATP release in both WT and Cx43-null cells provides further support that the pathway mediating BzATP-induced ATP release is unlikely to be Cx43 hemichannels.
To evaluate whether Panx1 could provide the pathway for ATP release from astrocytes, experiments were performed on WT astrocytes untreated and treated with Panx1 siRNA. As shown in Figure 4C, knockdown of Panx1 almost completely prevented BzATP-induced ATP release (from 105.6±8.24 to 117.9±9.9nM; N=10 experiments; P>0.05, T-test). Similarly to Cx43-null astrocytes (Fig. 4B), cytosolic ATP levels in Panx1 siRNA treated WT cells (4.7±0.9μM; N=10 experiments) were also significantly lower (P<0.001, T-test) than in untreated astrocytes (13.7±0.9μM; N=20 experiments; Fig. 4D). Scrambled Panx1 siRNA reduced cytosolic ATP levels (from 20.7±0.8 to 13.3±0.5μM, N=3 experiments, P<0.0001 T-test) but did not prevent BzATP induced ATP release from WT astrocytes (from 53.8±9.0 to 101.5±1.4nM; N=3 experiments, P=0.0004, T-test).
Together, these results support the hypothesis that Panx1 and not Cx43 hemichannels provides a pathway for ATP release from astrocytes.
Communication among astrocytes themselves and with other neural cells relies not only on direct gap junctional contacts but is also mediated through the release of paracrine signals. Among the distinct mechanisms by which astrocytes release “gliotransmitters”, hemichannels formed of gap junction (connexin) or gap junction-like (pannexin) proteins have been suggested to be prominent under certain conditions (reviewed in Parpura et al., 2004; Dahl and Locovei, 2006; Spray et al., 2006; Scemes et al., 2007).
Connexin43 is the most abundant gap junction protein in astrocytes (Dermietzel et al., 2000; Scemes et al., 2000) and openings of Cx43 hemichannels have been proposed to account for glial release of ATP (Stout et al., 2002, Kang et al., 2008) and glutamate (Ye et al., 2003) as well as for the uptake of fluorescent molecules under ischemic and inflammatory conditions (Contreras et al., 2002; Retamal et al., 2007). Evidence presented favoring the involvement of Cx43 in these processes includes blockade by compounds that inhibit Cx43 gap junction channels, presence of channels with appropriately large unitary conductances (about 200pS) in membranes of astrocytes and transfected cells (Retamal et al., 2007, reviewed in Spray et al., 2006). The strongest evidence favoring Cx43 hemichannels came from exogenous expression of GFP tagged Cx43, where channel properties reflected the presence of the tag (Bukauskas et al., 2002; Contreras et al., 2003). Interestingly, metabolic inhibition-induced membrane permeabilization was prevented in Cx43de/del but not in Cx43f/f:GFAP-Cre astrocytes (Contreras et al., 2002), suggesting that loss of Cx43 does not abolish “hemichannel” activity.
Pannexin1 has been reported to form gap junction channels and also to function as hemi-gap junction channels that are sensitive to gap junction channel blockers, including carbenoxolone and flufenamic acid (Bruzzone et al., 2003, 2005). The non-junctional Panx1 channels (pannexons) are voltage sensitive, 400-500pS channels that have been reported to be modulated by intracellular calcium and by mechanical stretch (Bao et al., 2004; Locovei et al., 2006b); these large conductance channels have been proposed to mediate ATP release from erythrocytes, mouse taste buds and astrocytes (Locovei et al., 2006a; Huang et al., 2007b; Scemes et al., 2007). Panx1 can be activated by ATP through the metabotropic P2Y1 and P2Y2 receptors (Locovei et al., 2006b), as well as through the ionotropic P2X7 receptors (Pelegrin and Surprenant, 2006; Locovei et al., 2007; Iglesias et al., 2008).
Panx1 transcripts are found in astrocytes in vitro and in vivo (Huang et al., 2007a; Ray et al., 2005, 2006); however, the extent to which the Panx1 protein forms functional channels in astroglial cells and their properties has not yet been fully investigated. We here show that Panx1 channels likely perform the “hemichannel” function in astrocytes. This is based on our quantitative measurements showing no difference between outward currents activated by strong depolarization or by P2X7 receptor stimulation in WT compared to Cx43-null astrocytes. Moreover, knockdown of Panx1 by siRNA is shown to greatly reduce the occurrence of these currents. In addition, MFQ which has been previously shown not to block Cx43 gap junction channels at concentrations below 20 μM (Cruickshank et al., 2004) and to block Panx1 channels at much lower concentration (Iglesias et al., 2008), is here shown to prevent both voltage-activated and BzATP-induced currents in both WT and Cx43-null astrocytes.
We previously showed that the P2X7R-Panx1 complex provides sites of ATP release, amplifying the extent to which intercellular Ca2+ waves (ICWs) spread among WT and Cx43-null astrocytes when exposed to low divalent cation solution (Scemes et al., 2007; Suadicani et al., 2006). In contrast to this view are the reports showing that the amount of ATP released from cells correlates with the levels of Cx43 expression and that gap junction channel blockers attenuate ATP release from astrocytes and transfected C6 glioma cells (Cotrina et al., 1998; Kang et al., 2008; Stout et al., 2002). However, to our knowledge, no previous study has directly measured ATP levels in WT and Cx43-null astrocytes. In agreement with these previous reports, we also found a correlation between Cx43 expression levels and amount of ATP release. However, we attributed this difference to lower cytoplasmic ATP concentration in Cx43-null astrocytes rather than to inhibited release in the absence of Cx43 hemichannels. Moreover, because knockdown of Panx1 prevented ATP release, it is more likely that Panx1 and not Cx43 hemichannels can provide sites of ATP release that amplifies the distance to which calcium signals spread among astrocytes. This possibility is in agreement with our previous studies showing that knockdown of Panx1 but not of Cx43, prevented the amplification of intercellular calcium waves in astrocytes (Suadicani et al., 2006; Scemes et al., 2007).
In summary, our results strongly support the notion that Panx1 and not Cx43 is the main molecular substrate of “hemichannel”. This is evidenced by the similar “hemichannel” activity measured in WT and Cx43-null astrocytes and by the prevention of such activity following Panx1 knockdown.
We gratefully acknowledge the technical assistance of Ms. Aisha Cordero with cell cultures and mouse genotyping. Our work is supported by NIH (NS054225 to ES, GM-8610 to GD, and NS041282 to DCS).