Probenecid, a gout remedy, inhibits pannexin 1 channels

William Silverman, Silviu Locovei, Gerhard Dahl


Probenecid is a well-established drug for the treatment of gout and is thought to act on an organic anion transporter, thereby affecting uric acid excretion in the kidney by blocking urate reuptake. Probenecid also has been shown to affect ATP release, leading to the suggestion that ATP release involves an organic anion transporter. Other pharmacological evidence and the observation of dye uptake, however, suggest that the nonvesicular release of ATP is mediated by large membrane channels, with pannexin 1 being a prominent candidate. In the present study we show that probenecid inhibited currents mediated by pannexin 1 channels in the same concentration range as observed for inhibition of transport processes. Probenecid did not affect channels formed by connexins. Thus probenecid allows for discrimination between channels formed by connexins and pannexins.

  • connexin
  • transport
  • erythrocyte
  • ATP release

probenecid has been used for decades for the treatment of gout. The mechanism of action of the drug is inhibition of a renal tubular transporter, thereby facilitating the excretion of the disease causative uric acid by blocking reuptake (5, 26, 37). Probenecid-sensitive transporters are widespread and are even found in plants (30, 31, 44, 52, 56). The inhibition of the transporter by probenecid is also exploited clinically to increase the effective concentrations of antibiotics, chemotherapeutics, and other medications.

The inhibitory effect of probenecid on organic anion transporters is well established, and the effect is thought to be so specific that the drug is often used as a diagnostic tool, i.e., its effect is typically interpreted as an involvement of an anion transporter in the tested parameter. Accordingly, block of cAMP or cGMP release from erythrocytes (18, 25), ATP release from glia cells (1, 17), and block of dye loss in various cell types (20, 21, 23) by probenecid have been presented as evidence for a role of transporters in these phenomena. However, alternative pathways for the transit of these molecules across the plasma membrane have to be considered. Besides the well-documented vesicular release of ATP, a parallel release through membrane channels must exist, because the release is attenuated by drugs that do not interfere with vesicular release but affect gap junction proteins and because ATP release in several cell types is associated with uptake of dye from the extracellular medium (13).

Special attention has to be given to pannexin 1 as an ATP release channel because of the specific properties of pannexin 1 channels and because of the expression pattern of pannexin 1 (2, 16, 28, 3234, 55, 59). Pannexin 1 channels are highly permeable to ATP and to dyes typically used for dye flux measurements through gap junction channels. These dyes are in the same size range as the Ca2+ indicator dyes whose loss is attenuated by probenecid. Pannexin 1 channels also are mechanosensitive, consistent with a role in Ca2+ wave initiation. Expression of pannexin 1 is found in cells exhibiting ATP release, including erythrocytes, endothelial cells, and astrocytes. Furthermore, the localization at the luminal membrane in epithelial cells is consistent with an ATP release function of pannexin 1 channels.

We therefore tested whether probenecid affects pannexin 1 channels in addition to the drug's action on anion transporters. We also tested the effects of 5-nitro-2(3-phenylpropylamino)-benzoic acid (NPPB) on pannexin 1 channels. NPPB is known as a chloride channel blocker and also has been described to block ATP release from ciliary epithelial cells (36, 45) and mouse mammary cells (47). The results of the present study indicate that probenecid is a powerful inhibitor of pannexin 1 channels. In contrast, connexin channels were not affected by probenecid. NPPB inhibited both pannexin and connexin channels.


Preparation of oocytes.

Xenopus laevis oocytes were prepared as previously described (15). Briefly, oocytes were isolated by incubating segments of surgically removed ovary in 2 mg/ml collagenase type I (Worthington Biochemical) in Ca2+-free oocyte Ringer solution (OR2; in mM: 82.5 NaCl, 2.5 KCl, 1.0 MgCl2, 1.0 CaCl2, 1.0 Na2HPO4, and 5.0 HEPES, pH 7.5) with antibiotics (10,000 U/ml penicillin and 10 mg/ml streptomycin) and stirring at 1 turn/s for 3 h at room temperature. After being thoroughly washed with regular OR2, oocytes devoid of follicle cells and having a uniform pigmentation were selected and stored in OR2 at 18°C.

mRNA and electrophysiology.

mRNA for pannexin 1, connexin46 (Cx46), or Cx32E143 were prepared using the mMessage mMachine in vitro transcription kit (Ambion). Oocytes were injected with 20–40 nl of mRNA (100–1,000 ng/ml) and incubated for 18–42 h at 18°C. Oocytes expressing Cx46 or Cx32E143 were incubated in OR2 plus 5 mM CaCl2 to prevent the channels from opening during the incubation. Oocytes were tested using two-electrode voltage clamp (model OC725C, Warner Instruments, or Geneclamp 500B, Axon Instruments) under constant perfusion according to the protocols described. Electrophysiology data are shown as means ± SE or as box plots.

Preparation of erythrocytes.

Xenopus blood was collected into OR2 plus 5 mM EGTA, pH 7.5, and spun at low speed. The buffy coat (the thin layer of cells atop the packed erythrocytes) was removed, and erythrocytes were washed three times with OR2 plus 5 mM glucose and then resuspended at 20% hematocrit. Cells were diluted into OR2 without antibiotics for use in dye uptake assays.

Dye uptake.

Erythrocytes (75 μl) at 0.1% hematocrit in OR2 were plated onto poly-d-lysine-coated 96-well plates (BioCoat, Becton Dickinson). OR2 alone (25 μl) or with the addition of 4 mM probenecid (Alfa Aesar) was immediately added (final concentration 1 mM), and the cells were allowed to adhere to the plates for 10 min. Solution (50 μl) was removed from the wells, and dye uptake was initiated by the addition of 50 μl of 1.0 mM YoPro-1 iodide (final concentration 0.5 μM) in OR2 (negative control), water (stimulated), or water plus 1.0 mM probenecid (stimulated and inhibited). Images were acquired with a Canon PowerShot S3 IS digital camera with an exposure time of 6 s and an aperture setting of 3.2 attached to the phototube of an inverted fluorescence microscope (model DMIL, Leica).

Extracellular ATP measurements from oocytes.

ATP assay solutions (Luciferin/Luciferase, Sigma-Aldrich) were mixed with supernatants collected from pannexin 1-injected and uninjected cells treated with OR2 or potassium gluconate in the presence of 150 or 500 μM probenecid. Oocytes were 4 days postinjection. Pannexin expression and cell viability were confirmed electrophysiologically. Cells were pretreated for 10 min with probenecid, where applicable, and then isolated for 10 min in 150 μl of the experimental solutions. Supernatant (100 μl) was obtained for each condition. Each condition was done in quintuplicate. Luminescence readings were obtained with a Victor 1420 multilabel counter (PerkinElmer) on a 96-well culture plate.


Probenecid [4-(dipropylsulfamoyl)benzoic acid] was obtained from Alfa Aesar. NPPB was obtained from Tocris. NaCl was obtained from EM Science, and CaCl2 was obtained from J.T.Baker. All other chemicals used were obtained from Sigma-Aldrich.


Pannexin 1, although originally discovered as a “gap junction protein,” rather than forming cell-to-cell channels exerts its physiological role as a freestanding, unapposed membrane channel allowing the flux of molecules between the cytoplasm and the extracellular space (2, 8, 9, 12, 16, 28, 39, 42). The channel can be opened at the resting membrane potential by mechanical stress, by an increase in cytoplasmic Ca2+ concentration, or by ATP through P2Y or P2X7 receptors (2, 33, 34). Experimentally, the channel also can be opened by applied voltage pulses (2, 12). The activation of the channel by voltage is probably nonphysiological, because positive potentials are required for the channel to open. However, because of the experimental convenience, we used this paradigm for channel opening.

Figure 1 shows that the organic anion transport inhibitor probenecid attenuates currents induced by voltage steps in pannexin 1-expressing oocytes in a dose-dependant fashion. Figure 2 shows the dose-response curve for the effect of probenecid on pannexin 1 channel currents. At 1 mM concentration, probenecid completely abolished the currents. In the micromolar range, a steep slope for the probenecid effect was observed with an IC50 of ∼150 μM (Fig. 2). Curiously, at concentrations below 1 μM, a small (<10%) dose-dependent inhibition of the currents was detectable. At these low concentrations, an adaptive phenomenon was observed (Fig. 1B, bottom trace) that was not prominent at higher drug concentrations. Probably there is more than one binding site for probenecid.

Fig. 1.

Effect of probenecid on pannexin 1-mediated currents. Xenopus oocytes were voltage clamped at a holding potential of −40 mV. Brief voltage steps (5.5 s in duration at a rate of 5 per minute) to +60 mV (top trace) elicited small currents (bottom trace) in uninjected oocytes (A) and significantly larger currents in oocytes injected 2 days before with pannexin 1 mRNA (B and C). The pannexin 1-mediated currents were attenuated by probenecid in a dose-dependent fashion (B). Similarly, 5-nitro-2(3-phenylpropylamino)-benzoic acid (NPPB) attenuated pannexin 1 channel currents (C).

Fig. 2.

Dose-dependent inhibition of pannexin 1 channel currents by probenecid and NPPB. The dose-response curves for the 2 drugs reveal an IC50 value for probenecid of ∼150 μM and for NPPB of ∼50 μM. n = 4 for all data points. For curve fitting and determination of EC50, the submicromolar data points were excluded from the analysis. The Hill slope is 1.2 for probenecid and 0.9 for NPPB. mPanx1, mouse pannexin 1.

The chloride channel blocker NPPB has been shown to block ATP release from erythrocytes, mouse mammary cells, and ciliary epithelial cells (36, 45, 47). Erythrocytes express high levels of pannexin 1, and ATP release from these cells is correlated with dye uptake and attenuated by the pannexin inhibitor carbenoxolone (32). To test whether the effect of NPPB on ATP release could be attributable to pannexin 1 channels, we tested NPPB on oocytes expressing pannexin 1. A dose-dependent inhibition of currents carried by pannexin 1 channels was observed (Fig. 1C). In contrast to probenecid but similar to carbenoxolone, NPPB inhibited the pannexin 1-mediated currents incompletely, leaving a residual current of ∼20%. The dose-response curve (Fig. 2) shows that NPPB, with an IC50 of ∼50 μM, was more potent, although less efficacious, than probenecid in attenuating pannexin 1 channel currents.

To test whether the effects of the two drugs are additive, we applied probenecid and NPPB concurrently, first one drug and then both. Figure 3 shows that at lower concentrations, the effects of the drugs did not add arithmetically. At high concentrations, NPPB left a residual current that remained unaffected by the addition of probenecid at a concentration that abolished the current when given alone. Thus it appears that the drugs act competitively on pannexin 1 channels. A final assessment of the nature of the drug interaction, however, needs to await single-channel studies.

Fig. 3.

Combined effects of probenecid and NPPB on pannexin 1 channels. Inhibition of the pannexin 1 currents by probenecid was enhanced by the addition of NPPB when both drugs were applied at their IC50 concentrations (A). When given at maximal concentrations, probenecid did not significantly reduce the residual currents in the presence of NPPB (B). A quantitative analysis of the residual currents remaining after application of the drugs at different concentrations (given as IC values) is shown (C). n = 3 for all data points except IC90, where n = 4. *P < 0.05. ns, Not significantly different.

The effect of NPPB is not specific to pannexin 1 channels. A similar inhibition of membrane currents by NPPB was observed in oocytes expressing Cx46 and Cx32E143 (Fig. 4). The vertebrate gap junction proteins Cx46 and Cx50, when expressed in oocytes, form patent nonjunctional membrane channels in addition to gap junction channels (4, 40). Such channel formation also is observed in oocytes expressing the chimera Cx32E143, where in a Cx32 background the sequence of the first extracellular loop is replaced by that of Cx43 (43). Membrane currents in oocytes expressing either Cx46 or Cx32E143 were not affected by probenecid even at high concentrations but were affected by NPPB in a way similar to pannexin 1 channels (Figs. 4 and 5).

Fig. 4.

Effect of probenecid and NPPB on connexin 46 (Cx46) and Cx32E143 channels. Oocytes expressing the connexins were held at −30 mV, and depolarizing pulses to −20 mV were applied. The resulting membrane currents are shown. A: probenecid does not affect connexin-based membrane channels. At concentrations exceeding those resulting in total inhibition of pannexin 1 channels, probenecid did not affect currents through Cx46 or Cx32E143 channels. B: currents through Cx46 and Cx32E143 channels were inhibited by NPPB similarly to the pannexin 1 channel currents. Because of different activation kinetics, the oocytes expressing connexins were held at depolarized potentials well below the equilibrium potential for potassium. Therefore, closure of the channels resulted in a considerable baseline shift.

Fig. 5.

Effects of probenecid and NPPB on Cx46 and Cx32E143 channels. A: both types of channels were unaffected by probenecid (n = 5). B: dose-dependent inhibition of Cx46 channel currents by NPPB. The IC50 for this effect was ∼50 μM (n = 3).

Probenecid and NPPB have been reported to inhibit the release of cAMP and of ATP from erythrocytes and glia cells. Human erythrocytes express pannexin 1 at high levels (32). We tested whether the drugs affect the release of ATP and the correlated uptake of extracellular dyes by erythrocytes in the same concentration range as they inhibit pannexin 1-mediated currents in oocytes. We choose frog erythrocytes because the presence of nuclei in these cells allows for the convenient measurement of dye uptake with YoPro. This compound is barely fluorescent in aqueous solution, and fluorescence intensity increases ∼2,000-fold on binding of the dye to DNA and RNA.

Erythrocytes from various species, including human, release ATP and take up dye in a low-oxygen environment, when mechanically stressed, or when depolarized with a high concentration of potassium gluconate (6, 32, 50). Figure 6 shows that the osmotic stress-induced uptake of YoPro from the extracellular space by frog erythrocytes was inhibited by probenecid.

Fig. 6.

Dye uptake by frog erythrocytes. Fluorescence micrographs were taken of the fields shown at 1, 6, and 15 min after application of YoPro, with transmitted light shown at right. When erythrocytes were exposed to the dye YoPro (0.5 μM) in the absence of a stimulus, only a few fluorescent spots were detected (A). Because of their size, these spots probably represent free nuclei. When stimulated by osmotic stress (1:1 diluted oocyte Ringer solution OR2), erythrocytes took up the YoPro as indicated by the intense staining of the nuclei (B). Dye uptake was detectable already within 1 min and increased over time. YoPro uptake was inhibited by 1 mM probenecid (C). Higher magnification of stimulated erythrocytes in a combined bright-field and fluorescence micrograph (D).

Pannexin 1 channels mediate the release of ATP from oocytes expressing the protein exogenously (2). To test whether the release is sensitive to probenecid, we determined efflux of ATP from oocytes expressing pannexin 1 by performing a luminometric luciferase assay. Figure 7 shows that the high-potassium-induced ATP release was attenuated by probenecid in a dose range matching that of the effects on pannexin 1 channel currents. As shown previously (2, 3), uninjected oocytes exhibit a moderate increase in ATP release when stimulated with high potassium levels (Fig. 7). The ATP release based on endogenous mechanisms in oocytes has been reported to be vesicular in nature because of its Brefeldin sensitivity (35).

Fig. 7.

ATP release from oocytes expressing pannexin 1. Release of ATP to the extracellular medium by oocytes was measured by luminometry using a luciferase assay. Depolarization of the cells with high-potassium solution (KGlu) resulted in a significant increase in ATP release compared with unstimulated cells. Probenecid attenuated the stimulated release in a dose-dependent way. ATP release from uninjected oocytes is shown at right. Normalized (Norm) luminescence is plotted; n = 7 for all experimental conditions using pannexin 1-expressing oocytes and uninjected oocytes. *P < 0.05. Prob, probenecid.


The data presented demonstrate that probenecid, which has been used for decades clinically for treatment of chronic gout, has an additional or alternative target heretofore known. It is well accepted that probenecid blocks organic anion transport (7, 14, 19). Accordingly, the mechanism of action of the drug in treatment of gout is inhibition of renal reuptake of uric acid, presumably by an anion transporter.

In the present report, however, we show that probenecid inhibits pannexin 1 channels that have no known relationship with transporters. Inhibition of pannexin 1 channels by probenecid has an IC50 of ∼150 μM, similar to that seen for inhibition of various transporters, including the human urate transporter hURAT1 (10, 46, 54, 57, 58). Considering the established effect on organic anion transport by probenecid, the drug either affects pannexin 1 channels and the transporters as separate entities or pannexin 1 is part of a transport protein complex providing the permeation pathway. In any case, transport claims for many phenomena based on probenecid effects have to be reevaluated in light of the present finding, in particular where a transport process was only inferred from the drug action.

How probenecid accesses and affects the pannexin 1 channel is unclear. Although hydrophobic in nature, the drug is sufficiently water soluble to prepare aqueous stock solutions. The drug, therefore, may interact either with hydrophilic aspects of the protein or may access the channel through the lipid bilayer.

Pannexins were discovered as a second family of vertebrate “gap junction proteins” based on the limited sequence homology to the innexins found in invertebrates (39). However, presently there is no evidence that pannexins actually form gap junction channels in vivo (8, 16, 28, 32, 42). Instead, these proteins seem to exert their physiological role exclusively as nonjunctional membrane channels providing a pathway for exchange of molecules between the cytoplasm and extracellular space (16). Most drugs acting as “gap junction blockers” do not discriminate among connexins, pannexins, and innexins. The only somewhat discriminating agent is carbenoxolone, which requires a slightly lower concentration to affect pannexin channels than connexin channels (11). Probenecid, therefore, is the first drug with specificity for channels formed by pannexins. NPPB is unspecific in its effect on gap junction proteins, as are all the others of the lot, including flufenamic acid, niflumic acid, and higher alcohols.

The inhibitory effect of probenecid on the release of ATP from astrocytes and other cells in the absence of other known targets for the drug has served as evidence for the involvement of transporters in the ATP release mechanism. Because of the strong effect of probenecid on pannexin 1 channels, an involvement of these channels in ATP release has to be seriously considered as an alternative, especially in light of the properties of pannexin 1 channels. They are highly permeable to ATP, activated at the resting membrane potential by mechanical stress, by extracellular ATP through P2Y or P2X7 receptors, and by increases in cytoplasmic Ca2+ concentration (2, 31, 32). Thus pannexin 1 channels are ideally suited to exert the role of ATP release channels in the initiation and propagation of intercellular Ca2+ waves. Furthermore, the expression pattern of pannexin 1, including luminal localization, is consistent with such a function.

Additional support for that role also comes from the abundant expression of pannexin 1 in erythrocytes. Erythrocytes release ATP in response to shear stress or in a low-oxygen environment as part of a peripheral control loop of oxygen delivery (6, 50). Under conditions of ATP release, erythrocytes take up dyes from the extracellular medium (32), and as shown in the present study, such dye uptake is inhibited by probenecid. Similarly, ATP release from oocytes expressing pannexin 1 is attenuated by probenecid in the same concentration range as the inhibition of pannexin 1 channel currents. These observations lend further support for a role of the pannexon, the hexameric assembly of pannexins forming a channel (8), as an ATP release channel.

Intercellular Ca2+ waves propagate through gap junction channels (48) and by an extracellular pathway involving ATP release and activation of purinergic receptors (27, 38). Typically, this phenomenon is studied with Ca2+-sensitive indicator dyes. The dyes are loaded into the cells in the form of membrane permeant esters, which are cleaved by cytoplasmic esterases. Although the cleavage product is essentially membrane impermeant, a gradual loss of dye is observed nevertheless. To prevent dye leakage, it is common practice to use probenecid in these studies with the assumption that the drug will inhibit an organic anion transporter responsible for dye loss (20, 22, 51). This practice has implications and will bias the experimental data toward gap junction-mediated wave propagation because these channels are not affected by probenecid (24). The practice of using probenecid also biases the experimental data toward vesicular release of ATP by attenuating, if not eliminating, pannexin 1-mediated ATP release.

In most cell types, with erythrocytes representing a notable vesicle-free exception, ATP release involves two parallel pathways. There is overwhelming evidence for vesicular release, including copackaging of ATP with other transmitters (49, 53) and sensitivity to Brefeldin (35). A series of observations, however, indicates that an additional, channel-based ATP release mechanism must exist. For example, the dye uptake by cells from the extracellular space under conditions of ATP release cannot be explained by a vesicular release mechanism but is consistent with a channel release process. Furthermore, a series of drugs that do not affect exocytosis attenuate ATP release. Although most of the drugs are not very specific in their action on channels, all of them inhibit pannexin 1 channels, most notably carbenoxolone, probenecid, and NPPB. In summary, these data lend more support for an ATP release channel role for pannexin 1.

Pannexin 1, through its interaction with the P2X7 receptor, is involved in the activation of the inflammasome, a protein platform that in turn converts interleukin-1β to its releasable form (29, 33, 41). Inflammation is at the root of gout. Thus it appears that probenecid may benefit gout patients in more than one way. In addition to the lowering of systemic uric acid by its action on tubular transport in the kidney, probenecid may stem inflammation through its inhibition of pannexin 1 channels.


This work was supported by National Institutes of Health (NIH) Grant GM-48610 and a NIH Training Grant (to W. Silverman).


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract