Two non-vesicular ATP release pathways in the mouse erythrocyte membrane

Preparation of cells

Mice were anesthesized with isofluorane. After decapitation, mouse blood was collected in 5-10ml Krebs solution containing 5mM EGTA. The collected blood was centrifuged at low speed to remove the plasma, leukocytes and platelets. The purified erythrocytes were washed three times by low-speed centrifugation before being suspended at 20% hematocrit (2% in the experiments with dipyridamole) in Krebs as stock. To assure equal number of erythrocytes in the various samples, light transmittance was determined spectrometrically and Krebs solution was added if needed.

Preparation of oocytes and electrophysiological recording were performed as described [18]. A mouse pannexin1 construct was kindly provided by Dr. Rolf Dermietzel (University of Bochum), and Cx46 plasmid was obtained from Dr. D. L. Paul (Harvard University). The plasmid containing mouse Pannxin1 in pCS2 was linearized with Not I. In vitro transcription were performed with polymerases SP6, using the Message Machine kit (Ambion, Austin, TX). mRNAs were quantified by absorbance (260nm), and the proportion of full-length transcripts was checked by agarose gel electrophoresis. In vitro transcribed mRNAs (~40nl) were injected into Xenopus oocytes. Cells were kept in regular Ringer solution OR2 (82.5mM NaCl, 2.5mM KCl, 1mM Na₂HPO₄, 1mM MgCl₂, 1mM CaCl₂ and 5mM HEPES) with 10mg/ml antibiotics streptomycin.

Pannexin1 knock-out mouse generation

Hemizygote Panx1-knockout (KO)first (Panx1^{tm1a(KOMP)Wtsi}) mice, in the C57Bl6 background, was generated and purchased from KOMP at UCDavis, an NIH initiative. Hemizygotes were bred till homozygosity and maintained as Panx1^+/+ and Panx^-/- in the animal facility at Albert Einstein College of Medicine. As previously described [19], Panx1 KOfirst were genotyped by tail PCR using 4 primers (1: 5’GAGATGGCGCAACGCAATTAAT3’; 2: 5’CTGGCTCTCATAATTCTTGCCCTG3’; 3: 5’ CTGTATCACACAACCACTTCAGAGAAGG3’; 4: 5’GAGCTGACCCCTTTCCATTCAATAG3’). The wild type Panx1 allele was targeted by primers 3 and 4 and identified as a 579 bp amplicon, while the transgene was targeted by primers 1 and 2 and identified as a 381 bp amplicon Mice lacking Panx1 are viable, fertile and do not display gross morphological defects of major organs.

Western blot

Erythrocytes derived from 2 Panx1^+/+ and 3 Panx1^-/- adult mice were lysed in HEPES buffered H₂O. Membranes were pelleted by centrifugation at 100,000 g. After resuspension, this step was repeated until the pellet was colorless. The pellet was taken up in lysis buffer (150 mM NaCl; 10 mM Tris-base; 1% TritonX-100; protease inhibitor cocktail; pH 7.4) and total protein measured using the BCA reagents (Thermo Scientific). Samples were electrophoresed in 4-20% mini-gels.

After transfer of proteins to nitrocellulose membranes and one hour incubation in blocking phosphate buffered solution (PBS) containing 0.5% Tween-20 and 2% skinned milk, blotting was performed for two hours at room temperature (RT) using anti-Panx1 (1:1000) antibody 4515 ([11,20] and anti-beta actin (1:2000) antibodies After washes with PBS-Tween-20, membranes were incubated with goat anti-chicken or goat anti-mouse HRP conjugated secondary antibodies (1:2000; Santa Cruz Technology) for one hour at RT. Visualization of bands was performed using X-ray film and a developer.

Electrophysiology

Whole cell membrane current of single oocytes was measured using a two-electrode voltage clamp and recorded with a chart recorder. Both voltage-measuring and current-passing microelectrodes were pulled with a vertical puller (Kopf) and filled with 3M KCl. The recording chamber was perfused continuously with solution. Membrane conductance was determined using voltage pulse. Oocytes expressing mouse pannexin1 were held at -60 or -50 mV, and 5s pulses to +20 or +50 mV were applied to transiently open the channels. Glyburide and dipyridamole were obtained from Sigma-Aldrich and iloprost from Cayman. Stock solutions of glyburide and dipyridamole were in DMSO and iloprost was dissolved in aqueous solution.

ATP-release assay

ATP flux was determined by luminometry. The mouse erythrocytes were pre-incubated in Krebs solution with or without (control) drugs for 10 minutes, and aliquots of 10μl were added into 100μl stimulation solutions (150mM KGlu or 1:1 diluted with water to yield hypotonic 75 mM KGlu) for 10 minutes. After 1-minute low-speed centrifugation (~100xg), 50μl of supernatant were used with a luciferase assay (Promega, Madison, USA) to determine the ATP level. Oocytes, two days after injection of mouse Panx1 messenger RNA were pretreated in OR2 solution with and without drugs for 10 minutes and stimulated by incubation in OR2 solution (negative control), KGlu solutions (positive control), KGlu solution with drug, respectively, for 10 minutes. The supernatant was collected and assayed with luciferase/luciferin. A calibration curve was prepared with ATP supplied with the assay kit to assure that measurements were within the linear range of the assay.

Dye uptake

Erythrocytes were incubated in Krebs and hypotonic KGlu solutions containing 1mM 5,6-carboxyfluorescein for 10min, then washed repeatedly with Krebs solution before viewing under epifluorescence microscopy. Fluorescence images were captured using a Nikon inverted microcope equipped with 40X objective, 488/512 nm filter sets and Metafluor software.

ATP release

Pannexin1 channels [21] can be opened by mechanical stress and by increased extracellular K⁺ concentration independent of the cell membrane potential [22-24]. Figure 1 shows that ATP release from Panx1^+/+ mouse erythrocytes was stimulated by a hypotonic K⁺ solution and the release was attenuated by the Panx1 channel inhibitor probenecid [13]. In contrast to human erythrocytes, where Panx1 channel inhibitors almost completely abolish ATP release [11,25], a substantial release remained in mouse erythrocytes after probenecid treatment. This observation suggests the existence of an alternate ATP release pathway to Panx1. As indicated by Panx1^-/- erythrocytes (Figure 1), this alternate pathway was also activated by hypotonic K⁺. However, in contrast to Panx1^+/+ erythrocytes, the amount of ATP released was lower and no significant inhibition of ATP release by probenecid was observed. A probenecid-insensitive ATP release from human erythrocytes has been reported previously for iloprost-induced release [25]. Figure 2a shows that iloprost-stimulated ATP release by mouse erythrocytes to similar levels regardless whether Panx1 was present (Panx1^+/+) or absent (Panx1^-/-) and that probenecid had no effect on this process in either type of erythrocyte.

Effect of dipyridamole

In a screening test on Panx1 expressing ooctyes we have observed that drugs reported in the literature to affect non-vesicular ATP release also affected Panx1 currents. The Panx1 inhibiting drugs include probenecid, which hitherto had been considered to be a specific transport inhibitor, and also connexin mimetic peptides as off target effect [13,26]. One drug, however, that is well documented to inhibit ATP release and nucleoside flux, dipyridamole [4,27], did not affect Panx1 channels. We, therefore, tested whether the second, probenecid insensitive component of ATP release by erythrocytes was sensitive to dipyridamole. As shown in Figures 2b and 3a, dipyridamole attenuated iloprost-induced ATP release but did not affect Panx1 currents. The effect of dipyridamole was not restricted to iloprost-induced ATP release but also included the hypotonic K⁺-induced release as shown in Figure 2c. This indicates that hypotonic K⁺ stimulates two separate ATP release pathways, one sensitive to probenecid and one sensitive to dipyridamole. In that case, one would assume that the effects of these two drugs are additive. However, no additive effect was observed (Figure 2c). Because the data with Panx1^-/- cells unequivocally demonstrate, that Panx1 cannot be the cause of iloprost-induced ATP release, the lack of additive effects of probenecid and dipyridamole on hypotonic K⁺-induced ATP release must have a different explanation than competition of the drugs on one single target. We, therefore, tested the effects of the two drugs on mouse Panx1 currents. Given separately, probenecid inhibited Panx1 currents while dipyridamole had no effect on the currents. When the drugs were combined, the inhibitory effect was significantly lower than that mediated by probenecid alone (Figure 3). Thus, it appears that the drugs interact. However, presently it is unclear where this interaction takes place. It could be a direct interaction of probenecid with dipyridamole in solution, thereby reducing the effective concentrations at their respective targets, or there could be interference at the targets. For example, dipyridamole, while by itself not interfering with Panx1 currents could compete at the probenecid binding site of Panx1 and consequently attenuate the inhibitory action of probenecid.

Uptake of extracellular tracer molecules

A hallmark of non-vesicular (presumably Panx1-mediated) ATP release is the correlated uptake of extracellularly applied tracer molecules. It has been demonstrated in many cell types, including erythrocytes [10-13], that dyes with an exclusion limit >1 kD are taken up by cells under conditions similar to those inducing ATP release. As shown in Figure 4, mouse Panx1^+/+ erythrocytes exhibited uptake of carboxyfluorescein when stimulated with hypotonic K⁺. In contrast, Panx1^-/- erythrocytes did not take up the dye under the same experimental conditions. Similarly, no dye uptake was observed, when ATP release was stimulated with iloprost in Panx1^-/- cells. These data show, that the Panx1 mediated, probenecid sensitive ATP release pathway allows the concomitant uptake of extracellular dyes. In contrast, the iloprost-induced/dipyridamole sensitive ATP release pathway does not allow the entry of extracellular molecules in a similar size range as ATP. Carboxyfluorescein has an abaxial diameter of about 9 Å as compared to 12 Å for ATP and thus should be small enough for passage through a non-selective, ATP permeable channel. Aside from its effect on ATP release, dipyridamole is known as a nucleoside transport inhibitor [14]. The absence of dye flux in the iloprost-induced ATP release pathway suggests a permeation pathway distinct from that provided by Panx1. The latter is in the form of a large, nonselective pore that below the exclusion limit of 1500 Daltons [26] does not discriminate by size or charge among molecules for transit through the channel. The dipyramidole sensitive pathway consequently is either an ATP selective channel or a transporter (or shares key features with a transporter).

Glyburide (glibenclamide) inhibits the Panx1 channel

ATP release by erythrocytes has been reported to involve the cystic fibrosis transmembrane conductance regulator (CFTR). Originally, it had been considered that CFTR provides the permeation pathway for ATP [28], however, the lack of ATP permeability of the CFTR channel excludes this possibility [29]. Nevertheless, the CFTR inhibitor glyburide (glibenclamide) does interfere with the release of ATP [6] and of epoxyeicosatrienoic acids [15]. Here we tested whether glyburide may have additional targets in the erythrocyte cell membrane. Specifically, because Panx1 channels evidently provide a permeation pathway for ATP, we tested whether glyburide inhibits ATP flux through Panx1 directly. Figure 5 shows, that glyburide attenuated Panx1 currents in oocytes expressing the protein exogenously with an EC₅₀ of 45 μM. Furthermore, glyburide impaired ATP release from Panx1 expressing oocytes in the same concentration range where current inhibition was observed (Figure 5d).