Pannexin 1 Channels as an Unexpected New Target of the Anti-Hypertensive Drug Spironolactone

Abstract

Rationale

Resistant hypertension is a major health concern with unknown etiology. Spironolactone is an effective anti-hypertensive drug, especially for patients with resistant hypertension, and is considered by the World Health Organization (WHO) as an “essential” medication. Although spironolactone can act at the mineralocorticoid receptor (NR3C2), there is increasing evidence of mineralocorticoid receptor (MR)-independent effects of spironolactone.

Objective

Here, we detail the unexpected discovery that pannexin 1 (Panx1) channels could be a relevant in vivo target of spironolactone.

Methods and Results

First, we identified spironolactone as a potent inhibitor of Panx1 in an unbiased small molecule screen, which was confirmed by electrophysiological analysis. Next, spironolactone inhibited α-adrenergic vasoconstriction in arterioles from mice and hypertensive humans, an effect dependent upon smooth muscle Panx1, but independent of the mineralocorticoid receptor NR3C2. Lastly, spironolactone acutely lowered blood pressure, which was dependent on smooth muscle cell expression of Panx1 and independent of NR3C2. This effect, however, was restricted to steroidal MR antagonists as a non-steroidal MR antagonist failed to reduced blood pressure.

Conclusions

These data suggest new therapeutic modalities for resistant hypertension based on Panx1 inhibition.

INTRODUCTION

Pannexin 1 (Panx1) channels have emerged as a major means for release of the nucleotides that mediate purinergic signaling¹, which is involved in multiple physiological and pathophysiological processes such as apoptosis², pyroptosis³, tumor cell metastasis⁴, vasoconstriction⁵, leukocyte migration⁶, insulin release from adipocytes⁷, NMDA receptor activation⁸, and neuronal survival⁹. Of particular relevance to this current work, Panx1 channels are activated by both caspase-dependent cleavage of the carboxyl terminus and cleavage-independent, receptor-mediated activation mechanisms. During apoptosis, caspase 3/7 cleaves the carboxyl terminus of Panx1, resulting in an irreversibly-activated channel that can release ATP as a ‘find-me’ signal to surrounding phagocytes². Alternatively, Panx1 channels, which are also expressed in smooth muscle cells, are reversibly activated in a receptor-dependent mechanism. In vascular smooth muscle cells, we and others have demonstrated that sympathetic-mediated vasoconstriction, and blood pressure, are Panx1-dependent based on genetic and pharmacological perturbation^{5, 10–13}. Specifically, activation of α1-adrenergic receptors results in opening of Panx1 channels that activate downstream purinergic receptors through release of ATP. Small ions and large molecules can permeate through open Panx1 channels, which allows for Panx1 channel activities to be assayed in several ways: by entry into cells of dyes, such as TO-PRO-3 or ethidium bromide; by release of nucleotides, such as ATP and UTP; and by plasma membrane Panx1 channel currents (reviewed in^14–16). Given the paucity of selective compounds capable of targeting Panx1, we used different assays to identify and validate new Panx1 blockers that would be useful in these various pathophysiological contexts, such as treatment resistant hypertension.

Here, using an unbiased screen for small molecules¹⁷ targeting Panx1, we identify spironolactone as a potent inhibitor of Panx1. Spironolactone is an effective anti-hypertensive drug, and is considered by the World Health Organization (WHO) as an “essential” medication to be held by all countries. Interestingly, the addition of spironolactone is beneficial to patients already prescribed three drugs from different classes, including a diuretic, and thus the blood pressure reduction action of spironolactone could be partially attributed to ‘off-target’ effects^{18, 19}. Since Panx1 can regulate α1-adrenergic receptor-mediated vasoconstriction and blood pressure⁵, we assessed whether Panx1 may contribute to the anti-hypertensive effects of spironolactone as an ‘off-target’ effect. The data presented herein indicate that Panx1 is a novel target of spironolactone and potential therapeutic target for patients with resistant hypertension.

METHODS

Detailed experimental methods are available via Online Methods section.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

RESULTS

We previously detailed a medium-throughput flow cytometry based assay, based on TO-PRO-3 dye uptake into Jurkat T cells that are induced to undergo Fas-mediated apoptosis, as a measure for PANX1 channel activity¹⁷. By screening a LOPAC library of small molecules using this assay, we now identify spironolactone as a PANX1 inhibitor. In validation studies, in a dose-dependent fashion, spironolactone blocked both PANX1-dependent ATP release from apoptotic cells and TO-PRO-3 dye uptake into these cells (Figure 1A). Structurally (but not functionally) similar compounds such as aldosterone (or progesterone) failed to inhibit Panx1 channel activity (Supplemental Figure I).

Figure 1. Spironolactone inhibits Pannexin 1 channels.

(A) Inhibition of ATP release or TO-PRO-3 uptake (median fluorescence intensity, MFI) by spironolactone or positive control CBX in apoptotic Jurkat T cells that were treated with anti-Fas (n= 4, mean ± s.e.m.). TO-PRO-3 uptake was specifically scored on Annexin V⁺ apoptotic cells via flow cytometry. ** p<0.01 and **** p<0.0001 by one-way ANOVA, with Bonferroni’s test for multiple comparisons; n.s.: not statistically significant. (B) Inhibition of whole-cell pannexin currents in apoptotic Jurkat cells by spironolactone. (left) Whole-cell current amplitude from membrane potentials (Em) of +80 mV and −50 mV were assessed from apoptotic Jurkat cells after spironolactone treatment. Washing off spironolactone reverses the inhibition and treatment with CBX (positive control for PANX1 inhibition) again inhibits the PANX1 currents. (top right) The current-voltage relationship graphs indicate that spironolactone inhibits PANX1 across the voltage range, without affecting ionic selectivity. (bottom right) Grouped results show that spironolactone (20 µmol/L) inhibited 70.3 ± 4.5 % (at +80 mV) and 58.6 ± 8.1 % (at −50 mV) CBX-sensitive currents (n=7, mean ± s.e.m.). (C) Dose-dependent inhibition of PANX1 whole-cell currents in HEK293T cells co-expressing PANX1(TEV) and Tobacco Etch Virus (TEV) protease, a system in which TEVp cleaves and activates PANX1(TEV) in an apoptosis-independent manner¹⁹. (top) Representative current-voltage relationship graph is shown. (bottom) Grouped results (n=8) show an IC50 of spironolactone as 8.7 ± 1.5 µmol/L at +80 mV and 8.0 ± 1.3 µmol/L at −50 mV. (D) Spironolactone addition directly affects single channel currents in inside-out patches of HEK293T cells expressing an active form of PANX1. Schematic of the inside out patch measurements is shown at the top. Representative inside-out patch recording from HEK293T cell co-expressing PANX1(TEV) and TEV protease before (left) and after (right) spironolactone (20 µmol/L) application, at patch potentials of +60 mV (upper) or −60 mV (lower). C: closed state; O1: level for one open channel; O2: level for two open channels. The traces on the right are shown to depict data indicating that majority of the channels are in the ‘closed state’ (C) after spironolactone addition compared to control conditions. (E) Open probability (NPo) of PANX1 channels was reduced by spironolactone (20 µmol/L, n=7). ** p<0.01 by two-tailed paired t-test. (F) Unitary conductance of PANX1 channels was unaffected by spironolactone (20 µmol/L, n=7). Unitary conductance (mean ± s.e.m.) are 91.0 ± 1.5 pS (control) and 89.5 ± 1.7 pS (spironolactone) at patch potentials between +50 mV and +80 mV, and 13.5 ± 1.9 pS (control) or 11.5 ± 1.5 pS (spironolactone) at patch potentials between −70 mV and −50 mV, respectively. (G) Arterioles from human patients with resistant hypertension showed expression for PANX1 (magenta) that were not seen when secondary antibodies alone (IgG) were used. The internal elastic lamina autofluorescence (green), α-smooth muscle actin (cyan) and DAPI (blue) are visualized in each image. Bottom images are higher magnification images. Arrowheads indicate endothelial Panx1, and arrows point to smooth muscle Panx1. Star (*) indicates lumen of arteriole. Scale bar: 30 µm. (H) Phenylephrine induced vasoconstriction was impaired in human arterioles from patients with treatment resistant hypertension, pretreated with either 5 µmol/L PxIL2P (n=3) or 80 µmol/L spironolactone (n=7 vessels). Mean ± s.e.m. * p<0.05 by two-way ANOVA (PxIL2P vs. untreated).

To test directly the effect of spironolactone on ionic currents through plasma membrane human PANX1 channels, we performed whole-cell recording from anti-Fas-treated Jurkat cells. In these apoptotic cells, spironolactone reversibly inhibited membrane currents with current-voltage properties characteristic of PANX1 channels, reducing currents equally well at both positive and negative potentials (Figure 1B). After washing, the current sensitive to spironolactone was fully blocked by carbenoxolone (CBX), a drug known to inhibit PANX1. Spironolactone inhibited ~60–80% of the maximal CBX-sensitive currents (Figure 1B). Spironolactone dose-dependently inhibited C-terminal cleavage-activated recombinant PANX1 currents in HEK293T cells²⁰ with an IC₅₀ of ~7–8 µmol/L (Figure 1C), approximately the concentration of spironolactone estimated to be achieved in tissues^{21, 22}. We also performed single channel recordings using the “inside-out” patch clamp configuration to test more directly the effects of spironolactone on PANX1 channels in a ‘cell-free’ system. Addition of spironolactone strongly reduced the amount of time that the channels spent in the open state (open probability, depicted as NPo), suggesting that spironolactone can directly inhibit PANX1 channels at the plasma membrane, without requiring other cytoplasmic or nuclear factors (Figure 1D). Spironolactone reduced the average open probability of PANX1 channels by 3–4-fold (at −60 mV, from 2.4 ± 0.7 to 0.8 ± 0.2; at +60 mV, from 2.0 ± 0.5 and 0.5 ± 0.9 (Figure 1E), without affecting channel conductance (Figure 1F). These data suggest that PANX1 is a new and previously unappreciated target of spironolactone, which inhibits the channel at concentrations that are achievable in patients.

Spironolactone is an effective anti-hypertensive drug, especially used in patients with hypertension that is resistant to the first line of therapies. Thus, we examined human arterioles isolated from resistant hypertensive individuals by immunohistochemistry and found that PANX1 channels were expressed in vascular smooth muscle cells, where α-smooth muscle actin is also localized, and endothelial cells (Figure 1G). Since spironolactone can be an effective treatment for high blood pressure in resistant hypertensive humans¹⁸, and because Panx1 channels are involved in regulating α1-adrenergic arteriolar vasoconstriction and blood pressure in mice⁵, we tested whether spironolactone affects α1-adrenergic vasoconstriction in PANX1-expressing vessels from patients with resistant hypertension. Isolated, pressurized resistant hypertensive human arterioles were stimulated with an α1-adrenergic receptor agonist, phenylephrine, with or without spironolactone treatment. In parallel experiments, we used a membrane-permeable peptide denoted PxIL2P that we previously developed to inhibit PANX1-dependent, α1-adrenergic receptor-mediated vasoconstriction⁵. Arterioles from resistant hypertensive humans constricted in response to phenylephrine, and both spironolactone and PxIL2P inhibited phenylephrine-induced vasoconstriction in these isolated human arterioles (Figure 1H).

We next tested the effect of spironolactone on recombinant Panx1 currents induced by α1-adrenergic receptor activation, and found robust inhibition of both mouse and human PANX1 channels (Figure 2A and Supplemental Figure IIA–B). We then tested whether spironolactone inhibited Panx1-mediated α1-adrenergic vasoconstriction using resistance arteries from mice²³. Thoracodorsal arteries were first treated with increasing concentrations of phenylephrine in the presence of a maximum dose of spironolactone (80 µmol/L), which resulted in blunted vasoconstriction (Figure 2B). The effect of spironolactone on the arteries was reversible, as washing out the spironolactone and treating the same arteries with phenylephrine alone showed a typically strong vasoconstriction (diameter reduced by ~70%; Figure 2B). Furthermore, vessel function was not adversely affected by spironolactone or phenylephrine, since the arteries dilated fully when treated with acetylcholine, contracted maximally in the presence of potassium chloride, and were again fully dilated after calcium was removed from the bath (Figure 2B). To determine an IC₅₀ for spironolactone action in this assay, we exposed arteries to increasing doses of spironolactone prior to testing vasoconstrictor actions of increasing concentrations of phenylephrine (Figure 2C). Spironolactone inhibited phenylephrine-induced vasoconstriction in a dose-dependent manner. The IC₅₀ of ~ 18.9 µmol/L for 1 µmol/L phenylephrine determined using whole arteries is within the range of spironolactone concentrations estimated to be achieved in tissues²¹, and was only slightly higher than the IC₅₀ from in vitro electrophysiology measurements (~8 µmol/L) (Figure 1C).

Figure 2. Spironolactone inhibition of Pannexin 1 channels on smooth muscle cells regulates α-adrenergic vasoconstriction.

(A) Exemplar whole-cell recording from a HEK293T cell co-expressing mouse Panx1 channels and α1D adrenergic receptors. Phenylephrine-induced currents (cyan shading) are inhibited by spironolactone (20 µmol/L; green shading) and further by CBX (50 µmol/L; red shading). Inset: whole-cell current at +80 mV during sequential application of phenylephrine, spironolactone and CBX (in the continued presence of phenylephrine). (B) Representative trace of phenylephrine-induced vasoconstriction of a C57Bl/6 thoracodorsal artery. Vasoconstriction was blunted when 80 µmol/L spironolactone was present; after washing off the spironolactone, the thoracodorsal artery regained the ability to contract in response to phenylephrine. To confirm vessel integrity was not affected the spironolactone treatment, 10 µmol/L acetylcholine (ACh) was used for vasodilation, and 40 mmol/L KCl was used for inducing vasoconstriction, followed by maximal dilation achieved via a Ca²⁺ free solution.

(C) Control vessels from C57Bl/6 mice treated with DMSO vehicle control (N=6) or 10 (N=4), 20 (N=5), 40 (N=5), or 80 (N=5) µmol/L spironolactone show dose-dependent inhibition of PE-induced vasoconstriction. Repeated measures two-way ANOVA showed significant differences between DMSO vehicle control vs. 20, 40, or 80 µmol/L spironolactone at 10⁻⁶ – 10⁻³ mol/L phenylephrine. Inset: IC50 for spironolactone was determined to be ~ 18.9 µmol/L at 1 µmol/L phenylephrine (10⁻⁶ mol/L). (D) Endothelial cell deletion of Panx1 retained significant reduction of phenylephrine-induced vasoconstriction upon treatment with 80 µmol/L spironolactone. DMSO vehicle control: N=4 mice; n=6 vessels; spironolactone: N=4; n=7. Repeated measures two-way ANOVA showed significant differences between DMSO vehicle control vs. 80 µmol/L spironolactone at 10⁻⁶ – 10⁻³ mol/L phenylephrine. (E and F) Smooth muscle cell deletion of Panx1 after tamoxifen injection (E), prevented the inhibitory effect of 80 µmol/L spironolactone on phenylephrine-induced vasoconstriction, while this was retained in in Panx1^fl/fl/SMC-CreER^T2+ mice injected with the control vehicle (peanut oil) (F). Tamoxifen injected mice: DMSO vehicle control N=6, n=11; spironolactone treated N=5, n=10. Vehicle injected: DMSO vehicle control N=6, n=11; spironolactone treated N=6, n=9. Two-way ANOVA and significant differences at any dose of phenylephrine are shown. For all graphs: mean ± s.e.m. and * p<0.05.

Next, we tested whether this effect of spironolactone on phenylephrine-induced vasoconstriction was dependent on Panx1 expression in the endothelium or the smooth muscle. We used mice with floxed Panx1 alleles crossed to transgenic mice that express tamoxifen-inducible CreER^T2+ to obtain Panx1 deletion either in endothelial cells (Panx1^fl/fl/EC-CreER^T2+) or vascular smooth muscle cells (Panx1^fl/fl/SMC-CreER^T2+). We previously demonstrated inducible and efficient deletion of Panx1 after tamoxifen injection with these two Cre strains^{5, 6}. Deletion of Panx1 specifically from endothelial cells in the Panx1^fl/fl/EC-CreER^T2+ mice did not affect the ability of spironolactone to inhibit α1-adrenergic vasoconstriction (Figure 2D). By contrast, deletion of Panx1 from vascular smooth muscle cells in the Panx1^fl/fl/SMC-CreER^T2+ mice blunted phenylephrine-induced vasoconstriction and completely abolished the inhibitory effect of spironolactone (Figure 2E). In control Panx1^fl/fl/SMC-CreER^T2+ littermates injected with vehicle alone (without tamoxifen), phenylephrine-induced vasoconstriction and inhibition of vasoconstriction by spironolactone was preserved (Figure 2F). Notably, there were no differences in the endothelial-dependent arterial dilation with acetylcholine among all tested mice (Supplemental Figure IIC). Importantly, vasoconstriction induced by serotonin, which is independent of Panx1 channels⁵, was comparable in mice lacking Panx1 in vascular smooth muscle cells, endothelial cells, and control mice (Supplemental Figure IID). These data suggest that Panx1 expression, specifically in the vascular smooth muscle cells, is a key requirement for spironolactone to inhibit α1-adrenergic receptor-mediated vasoconstriction of resistance arteries. This is an important distinction from large conduit arteries (e.g., aorta) that do not directly regulate blood pressure, do not express Panx1 in smooth muscle cells²⁴, and have α1-adrenergic responses that are not inhibited by spironolactone²⁵. Endothelial Panx1 in conduit arteries has been suggested to regulate vascular tone²⁶; however, we find no effect of Panx1 deletion from endothelial cells on phenylephrine-induced vasoconstriction or acetylcholine-induced vasodilation, further supporting the concept that resistance arteries and conduit arteries employ different mechanisms for sympathetic regulation of vascular tone.

It is well established that spironolactone is an antagonist of the mineralocorticoid receptor NR3C2, acting in the kidney and vascular smooth muscle cells to lower blood pressure^27–30. However, rapid, non-genomic and seemingly mineralocorticoid receptor (MR)-independent ‘off-target’ effects of spironolactone have been reported as contributing to its larger beneficial anti-hypertensive actions^31–35. These non-MR effects have largely remained unexplained. Therefore, we examined whether the inhibitory effect of spironolactone on α1adrenergic, Panx1-dependent vasoconstriction required the mineralocorticoid receptor, NR3C2. We tested this via four different means. First, we used NR3C2 siRNA to knockdown NR3C2 expression in Jurkat cells and tested TOPRO-3 uptake after induction of apoptosis. TO-PRO-3 uptake occurred comparably in control and NR3C2 knockdown cells, and was inhibited equally by spironolactone (Figure 3A and Supplemental Figure IIIA). Second, we crossed mice carrying floxed alleles of Nr3c2 with SMC-CreER^T2+ mice, and after tamoxifen injection to induce Nr3c2 deletion (SMC Nr3c2-KO), we assessed spironolactone effects on α1-adrenergic vasoconstriction. In arteries from tamoxifen-treated Nr3c2^fl/fl/SMC-CreER^T2+ mice, spironolactone was still able to blunt vasoconstriction induced by phenylephrine (Figure 3B). We confirmed that Nr3c2 was indeed deleted in thoracodorsal arteries after tamoxifen injection (Supplemental Figure IIIB) and that these thoracodorsal arteries were fully responsive to Panx1-independent vasodilation or vasoconstriction (Supplemental Figure IIIC). Third, we tested the effect of spironolactone on phenylephrine-induced uptake of TO-PRO-3 into arteries from C57Bl/6 control mice, smooth muscle Nr3c2-deleted (SMC Nr3c2-KO) mice, or smooth muscle Panx1-deleted (SMC Panx1-KO) mice (Figure 3C). Similar to control vessels, thoracodorsal arteries lacking Nr3c2 retained the ability to take up TO-PRO-3 in response to phenylephrine, and this dye uptake was abolished by spironolactone. This phenylephrine-induced dye uptake was dependent on Panx1, as it was not observed in arteries lacking Panx1 in the vascular smooth muscle cells. Together these data indicate that the spironolactone mediated inhibition of α1-adrenergic signaling in these assays occurs independent of the mineralocorticoid receptor.

Figure 3. Spironolactone acutely lowers blood pressure by inhibiting Pannexin 1 channels in a mineralocorticoid receptor Nr3c2-independent manner.

(A) Jurkat cells transfected with control siRNA or NR3C2 siRNA were induced to undergo apoptosis via anti-Fas treatment (250 ng/ml) in the presence or absence of spironolactone, and TO-PRO-3. Dye uptake was measured using flow cytometry (left). Knockdown efficiency of NR3C2 is shown on the right. Data are presented as mean ± s.e.m. (n=4). ** p<0.01 and **** p<0.0001 by one-way ANOVA, with Bonferroni’s multiple comparison test (left) or by two-tailed unpaired t-test (right). (B) Spironolactone significantly blunted phenylephrine-induced vasoconstriction in mice with smooth muscle cell deletion of Nr3c2 by injection with tamoxifen. DMSO vehicle control: N=5 mice; n=9 vessels; spironolactone (80 µmol/L): N=5; n=9. Repeated measures two-way ANOVA showed significant differences between untreated vs. spironolactone at 10⁻⁷ – 10⁻³ mol/L phenylephrine. * p<0.05. (C) Phenylephrine-induced uptake of TO-PRO-3 (red) in thoracodorsal arteries from C57Bl6 and SMC Nr3c2-KO mice was prevented by pretreatment with 80 µmol/L spironolactone (blue, DAPI counterstained to mark nuclei from smooth muscle cells). However, SMC Panx1-KO mice failed to take up TO-PRO-3, and this is not affected by vehicle or spironolactone treatment. (D) Continuous recording of blood pressure (5 min averages) reveals a significant drop in mean arterial pressure (MAP) following intraperitoneal (i.p.) injection of 40 mg/kg spironolactone in C57Bl6 control (N=5 mice), SMC Nr3c2-KO (N=5 mice), and hypertensive BPH/2 mice (N=5 mice). In contrast, spironolactone did not decrease MAP in SMC Panx1-KO mice (N=5). Data points are presented as averaged MAP taken every 5 min (mean ± s.e.m.). Dashed green line indicates time of injection. Blood pressure returned to pre-injection values 24 hr after injection. (E) Mean arterial pressure (MAP) is significantly reduced from baseline in C57Bl6 control, SMC Nr3c2-KO, and hypertensive BPH/2 mice following injection of 40 mg/kg spironolactone. (Δ MAP = 2-hr baseline MAP, 30-min post-injection MAP). However, injection of 40 mg/kg finerenone, a non-steroidal mineralocorticoid antagonist, or vehicle (DMSO) failed to reduce the blood pressure in all genotypes. Notably, spironolactone, did not reduce MAP in SMC Panx1-KO mice. N=5 mice per group (N=4 for each group for BPH/2 mice); mean ± s.e.m. (one-way ANOVA; * p<0.05; *** p<0.001). Boxed: Representative whole-cell recording from a HEK293T cell co-expressing human PANX1(TEV) and TEV protease shows that finerenone (50 µmol/L) does NOT inhibit PANX1 currents. Inset shows whole-cell currents at +80 mV upon sequential application of finerenone and CBX (% inhibition is 15.3 ± 5.3 for currents at +80 mV and 30.1 ± 15.6 for −50 mV; n=6).

We next asked whether the blood pressure changes induced by spironolactone administration in vivo relate to Panx1. For this, we focused on the short-term/acute, non-genomic effects of spironolactone on blood pressure expected from Panx1 regulation of sympathetic vascular responses, recognizing that it would be difficult to distinguish those from other MR-dependent genomic changes on BP under longer term treatment conditions. Spironolactone evoked a rapid decrease in mean arterial pressure (MAP) when injected into C57Bl/6 control and SMC Nr3c2-KO, whereas spironolactone did not reduce blood pressure in SMC Panx1-KO mice or mice with global deletion of Panx1 (Panx1^−/−,¹⁷) (Figure 3D–E and Supplemental Figure IIIE and H). We also tested the effect of spironolactone on spontaneously hypertensive BPH/2 mice, with elevated mean arterial pressure that is largely attributed to sympathetic overdrive³⁶. In these BPH/2 mice, again, spironolactone induced a strong reduction in mean arterial pressure.

We also used additional pharmacological agents to differentiate effects mediated by Panx1 and the mineralocorticoid receptor. Recently, non-steroidal MR antagonists, such as finerenone, have been developed to overcome the side-effects of spironolactone that causes it to be discontinued during long term treatments (primarily due to spironolactone acting as a partial progesterone receptor agonist and causing gynecomastia). Finerenone did not affect PANX1 currents (Figure 3E) and did not reduce mean arterial pressure in any of these groups of mice, producing effects similar to vehicle injection (DMSO) (Figure 3E and Supplemental Figure IIIF). We also tested trovafloxacin, a distinct small molecule Panx1 inhibitor previously identified in a screen¹⁷; injection of trovafloxacin caused a decrease in mean arterial pressure in C57Bl/6 control, SMC Nr3c2-KO, and BPH/2 hypertensive mice, whereas it had no effect in mice lacking Panx1 in the vascular smooth muscle cells (SMC Panx1-KO mice, Supplemental Figure IIIG and I). We further verified our earlier results that SMC Panx1-KO mice have reduced blood pressure, which is most pronounced at night when sympathetic activity is typically highest⁵ and we confirmed that the BPH/2 hypertensive mice also had significantly higher mean arterial pressure (Supplemental Figure IIID). Collectively, these in vitro and in vivo data indicate that spironolactone can act via Panx1 channels to acutely lower blood pressure, independently of the mineralocorticoid receptor NR3C2.

In vivo, spironolactone is converted to metabolites including canrenone (sold as an anti-hypertensive drug) and 7-alpha-thiomethylspirolactone (7-α-TMS)³⁷. In addition, the spironolactone-derived compound eplerenone is a steroidal MR antagonist that is more MR-specific, but less potent than spironolactone, and also used clinically³⁸. We found a dose-dependent decrease in PANX1-dependent TO-PRO-3 uptake by anti-Fas treated, apoptotic Jurkat cells exposed to canrenone and eplerenone (Figure 4A, Supplemental Figure IVA), and again both of these effects were independent of NR3C2 (Figure 4A). Canrenone and eplerenone also inhibited ATP release in a dose-dependent fashion (Supplemental Figure IVB). Likewise, and similar to spironolactone, canrenone and 7-α-TMS inhibited PANX1 currents at hyperpolarized (−50 mV) and depolarized (+80 mV) membrane potentials (Figure 4B). Of note, eplerenone also inhibited PANX1 currents at concentrations similar to those achieved in vivo, but was less effective than spironolactone in inhibiting Panx1 in all of the above assays (Figure 4A and B, Supplemental Figure IV)³⁹. To determine the specificity of these compounds for Panx1 over other topologically related channels⁴⁰, we tested their effects on currents of Pannexin 2 (Panx2) channels or Connexin 43 (Cx43) hemichannels by using whole-cell recordings. We found that spironolactone, canrenone, or eplerenone did not inhibit recombinant Panx2 or Cx43 currents in HEK293T cells (Figure 4C–D). These data suggest that spironolactone, as well as its metabolites and analogs, can all inhibit Panx1 channels.

Figure 4. Metabolites and analogs of spironolactone inhibit Panx1 channel function, but not other nucleotide-release channels.

(A) Jurkat cells transfected with control siRNA or NR3C2 siRNA were induced to undergo apoptosis via anti-Fas treatment (250 ng/ml) in the presence or absence of spironolactone or its metabolites and analogs. Representative histograms of TO-PRO-3 uptake are shown on the left. TO-PRO-3 uptake was measured using flow cytometry (right). Data are presented as mean ± s.e.m. (n=4). TO-PRO-3 uptake by apoptotic Jurkat cells was significantly (p<0.0001) reduced by spironolactone (green), canrenone (blue), and eplerenone (purple), compared to vehicle controls (black); no statistical significance was found between scramble and NR3C2 siRNA groups by the same treatment of spironolactone and derivatives (two-way ANOVA, with Bonferroni’s test). (B) Inhibition of CBX-sensitive whole-cell currents by spironolactone and its derivatives (mean ± s.e.m.). Whole-cell currents were obtained from HEK293T cells co-expressing human PANX1(TEV) and TEV protease. All compounds caused voltage-independent inhibition of PANX1 currents. (C) Spironolactone and its derivatives show negligible effect on mouse pannexin 2 (Panx2) channel currents. Whole-cell currents were recorded from HEK293T cells exogenously expressing mouse Panx2. Cells were treated with spironolactone (20 µmol/L), canrenone (20 µmol/L), or eplerenone (20 µmol/L), and followed by CBX (50 µmol/L) exposure. Exemplar I-V curves and grouped percent inhibition (mean ± s.e.m.) are presented. (D) Spironolactone and its derivatives display negligible effect on Cx43 hemichannel currents. Whole-cell currents were recorded from HEK293T cells exogenously expressing Cx43. Cells were treated with spironolactone (20 µmol/L), canrenone (20 µmol/L), or eplerenone (20 µmol/L), and followed by gadolinium (Gd³⁺, 100 µmol/L) inhibition. Exemplar I-V curves and grouped percent inhibition (Gd³⁺-sensitive, mean ± s.e.m.) are presented.

DISCUSSION

Collectively, these studies progress from small molecule screen through target validation to whole animal and human vessel studies to demonstrate that smooth muscle cell Panx1 channels are an unexpected and potentially functionally relevant target for the effects of spironolactone in α1-adrenergic vasoconstriction and blood pressure regulation. The implications for this discovery are significant for several reasons. First, hypertension is a major driver of cardiovascular risk affecting 40% of adults worldwide, yet roughly half of the hypertensives are not controlled with current therapy; those with this resistant hypertension often respond well to MR antagonist treatment, such as spironolactone^{18, 19}. The data presented here on the effect of spironolactone in acutely lowering blood pressure through its effects on Panx1 in the resistance arteries is broadly relevant to understanding mechanisms of resistant hypertension and to the benefits of compounds typically associated with MR antagonism. Indeed, the inhibitory effects of spironolactone on Panx1-mediated α1-adrenergic vasoconstriction in mice and resistant hypertensive humans provides an explanation for prior observations suggesting inhibition of adrenergic responses after spironolactone administration in animals and in humans^{25, 41, 42}. Although Panx1 channels are expressed in various types of cells, such as smooth muscle cells⁵, endothelial cells⁶, leukocytes², and adipocytes⁷, spironolactone inhibition of α1-adrenergic vasoconstriction and reduction of blood pressure were abolished by deletion of Panx1 specifically in smooth muscle cells. Importantly, blood pressure was reduced in normotensive and hypertensive mice, in a vascular smooth muscle Panx1-dependent manner by treatment with spironolactone. These effects appear to be specific to steroidal MR antagonists since the non-steroidal MR antagonist finerenone did not inhibit PANX1 or acutely reduce blood pressure. A comprehensive evaluation of the specific structural features within steroid or steroid-like compounds that enable Panx1 channel inhibition (e.g., spironolactone and derivatives, CBX) versus those that do not affect Panx1 function (e.g., aldosterone, progesterone) awaits further study. We note that both spironolactone and CBX¹³ inhibit PANX1 channels in cell-free patches, suggesting direct effects of these steroidal compounds on channel function. Additionally, we were only able to use male mice in this study because the SMMHC-CreER^T2+ cre recombinase is located on the Y chromosome⁴³ leaving open the possibility of sex differences in Panx1-dependent anti-hypertensive effects of spironolactone. Second, pannexin channels mediate a variety of clinically significant physiological processes^{2, 3, 5–9}, and we have now identified spironolactone as a potent Panx1 inhibitor that can be used to explore further the role and therapeutic potential of Panx1. Third, although spironolactone is an effective antihypertensive, it has side effects limiting its use (e.g., particularly hyperkalemia, due to its potassium sparing actions in the kidney); thus, identification of novel inhibitors targeting Panx1 channels in vivo could be relevant for lowering of blood pressure in such individuals. Finally, beneficial MR-independent effects of spironolactone and related compounds have been widely reported, but remain poorly understood (e.g., in heart failure)^31–35; the unexpected discovery of Panx1 as an alternative target for this class of drugs suggests additional possible mechanism worthy of further analysis for those elusive actions.

NOVELTY AND SIGNIFICANCE.

What Is Known?

• Spironolactone is a widely used mineralocorticoid receptor (MR) inhibitor and effective anti-hypertensive drug, particularly for treatment-resistant hypertensive patients.

• Off-target effects of spironolactone have been thought to also contribute to its effectiveness in treatment-resistant hypertensive patients.

• Pannexin1 (Panx1), a nucleotide release channel, has been shown to regulate various physiological processes, including receptor-mediated vasoconstriction.

What New Information Does This Article Contribute?

• Spironolactone is a potent inhibitor of Panx1 channels.

• α-adrenergic receptor-mediated vasoconstriction of arterioles from treatment-resistant hypertensive patients is inhibited by spironolactone.

• Spironolactone acutely reduces blood pressure in mice, an effect independent of the mineralocorticoid receptor but requires Panx1 in smooth muscle cells (SMC).

Treatment-resistant hypertension is a major problem around the world that can be ameliorated by spironolactone, a drug class that is often used as key second line of treatment. While the primary action of spironolactone is linked to the MR, the mechanisms for its efficacy in acute reduction of blood pressure and the various non-genomic effects of spironolactone remain unexplained. First, we detail the unexpected discovery that Panx1 channels are additional, potentially therapeutically relevant, molecular targets of spironolactone. We identified spironolactone as a potent inhibitor of Panx1 in an unbiased small molecule screen. Panx1 has been shown to be important for αadrenergic vasoconstriction and indeed spironolactone inhibited α-adrenergic vasoconstriction in hypertensive human arterioles and mouse peripheral arterioles. This effect was lost when Panx1 expression was specifically deleted from vascular SMC but was fully retained in mice lacking the MR in SMC. Likewise, at the whole animal level, spironolactone caused an acute lowering of blood pressure that was dependent on expression of Panx1 in SMC, but independent of MR. These data identify Panx1 as a novel target of spironolactone, and suggests the potential for new therapeutic modalities for treatment-resistant hypertension based on Panx1 channel inhibition.

Nonstandard Abbreviations and Acronyms

Panx1

Pannexin 1

MR

Mineralocorticoid Receptor

NMDA Receptor

N-methyl-D-aspartate Receptor

LOPAC

Library of Pharmacologically Active Compounds

CBX

Carbenoxolone

NPo

Open Probability

DMSO

Dimethyl Sulfoxide

7-α-TMS

7-alpha-thiomethylspirolactone

Panx2

Pannexin 2

Cx43

Connexin 43

Em

Membrane Potential

TEVp

Tobacco Etch Virus Protease

HEK

Human Embryonic Kidney

PE

Phenylephrine

Gd³⁺

gadolinium

IEL

Internal Elastic Lamina

MAP

Mean Arterial Pressure