Carbenoxolone and 18β‐glycyrrhetinic acid inhibit inositol 1,4,5‐trisphosphate‐mediated endothelial cell calcium signalling and depolarise mitochondria

Background and Purpose Coordinated endothelial control of cardiovascular function is proposed to occur by endothelial cell communication via gap junctions and connexins. To study intercellular communication, the pharmacological agents carbenoxolone (CBX) and 18β‐glycyrrhetinic acid (18βGA) are used widely as connexin inhibitors and gap junction blockers. Experimental Approach We investigated the effects of CBX and 18βGA on intercellular Ca2+ waves, evoked by inositol 1,4,5‐trisphosphate (IP3) in the endothelium of intact mesenteric resistance arteries. Key Results Acetycholine‐evoked IP3‐mediated Ca2+ release and propagated waves were inhibited by CBX (100 μM) and 18βGA (40 μM). Unexpectedly, the Ca2+ signals were inhibited uniformly in all cells, suggesting that CBX and 18βGA reduced Ca2+ release. Localised photolysis of caged IP3 (cIP3) was used to provide precise spatiotemporal control of site of cell activation. Local cIP3 photolysis generated reproducible Ca2+ increases and Ca2+ waves that propagated across cells distant to the photolysis site. CBX and 18βGA each blocked Ca2+ waves in a time‐dependent manner by inhibiting the initiating IP3‐evoked Ca2+ release event rather than block of gap junctions. This effect was reversed on drug washout and was unaffected by small or intermediate K+‐channel blockers. Furthermore, CBX and 18βGA each rapidly and reversibly collapsed the mitochondrial membrane potential. Conclusion and Implications CBX and 18βGA inhibit IP3‐mediated Ca2+ release and depolarise the mitochondrial membrane potential. These results suggest that CBX and 18βGA may block cell–cell communication by acting at sites that are unrelated to gap junctions.


| INTRODUCTION
Cell-cell communication is a central component of endothelial function that is required for propagated vasodilation, transfer of signals from activated cells and emergent signalling (Bagher & Segal, 2011;Lee et al., 2018;Longden et al., 2017;McCarron et al., 2019;Socha, Domeier, Behringer, & Segal, 2012;Tallini et al., 2007). Among key signalling molecules that are transferred between cells are inositol 1,4,5-trisphosphate (IP 3 ) and cytoplasmic Ca 2+ . Changes in IP 3 and cytoplasmic Ca 2+ concentration decode information held in extracellular activators and encode intracellular signals that regulate the production of NO, prostacyclin and signalling peptides that diffuse to smooth muscle cells (Tran & Watanabe, 2006).
In the endothelium, Ca 2+ increases begin as highly localised subcellular events caused by the opening of a single or multiple IP 3 receptors in the internal store (Bagher et al., 2012;Ledoux et al., 2008;Sonkusare et al., 2012;. These local signals rapidly grow and propagate among cells to transmit information. However, the mechanisms that scale the signals to propagate waves and enable cell-cell communication are not well understood, even though they are critical to permit Ca 2+ to act as a communicator with wide reach (Behringer, Socha, Polo-Parada, & Segal, 2012;Billaud et al., 2014;Emerson & Segal, 2000a;Emerson & Segal, 2000b;Ledoux et al., 2008;Sonkusare et al., 2012;Taylor & Francis, 2014).
Among the most widely used pharmacological agents to study the role of gap junctions in cell-cell communication are the connexin and gap junction blockers 18β-glycyrrhetinic acid (18βGA) and its derivative carbenoxolone (CBX). Derived from the liquorice root Glycyrrhiza glabra, 18βGA (see Bodendiek & Raman, 2010) blocks a wide range of connexins such as Cx43 (Guan, Wilson, Schlender, & Ruch, 1996), Cx46 and Cx50 (Bruzzone, Barbe, Jakob, & Monyer, 2005). CBX is a derivative of 18βGA and is perhaps the most widely used broadspectrum connexin channel and gap junction inhibitor.
To investigate whether or not gap junctions play a role in endothelial IP 3 -mediated Ca 2+ signal propagation between cells, we aimed to disrupt normal gap junction function pharmacologically using CBX and 18βGA. IP 3 -evoked intercellular Ca 2+ waves were measured in the endothelium of intact mesenteric resistance arteries after stimulation with either ACh or photorelease of caged-IP 3 (cIP 3 ). cIP 3 provides precise spatial and temporal control of the site of cell activation and Ca 2+ release. Paired cellular responses to ACh or cIP 3 were analysed before and after various pharmacological interventions with CBX and 18βGA. Intercellular Ca 2+ waves were blocked by CBX and 18βGA, but this occurred by inhibition of IP 3 -evoked Ca 2+ release rather than block of gap junction-mediated signal propagation. The inhibition of IP 3 -evoked Ca 2+ release by CBX and 18βGA was reversible and was unaffected by the presence of small or intermediate K + -channel blockers. Furthermore, CBX and 18βGA each also rapidly and reversibly collapsed the mitochondrial membrane potential. These results suggest that CBX and 18βGA act at sites outwith gap junctions by inhibiting IP 3 -mediated Ca 2+ release and depolarising mitochondrial membrane potential (ΔΨ M ). Care is required in the use of these drugs when IP 3 -mediated Ca 2+ signalling is being investigated. • Carbenoxolone and 18β-glycyrrhetinic acid are widely used to study gap junctions in cell-cell communication.
What is the clinical significance Sprague-Dawley rats (10-12 week old; 250-350 g), from an in-house colony, were used for the study. The animals were housed three per cage, and the cage type was North Kent Plastic model RC2F with nesting material "Sizzle Nest." A 12:12 light dark cycle was used with a temperature range of 19-23 C (set point 21 C) and humidity levels between 45% and 65%. Animals had free access to fresh water and SDS diet RM1 (rodent maintenance). The enrichment in the cages was aspen wood chew sticks and hanging huts.
Animals were killed by cervical dislocation and the mesenteric bed removed. All experiments were performed using first-or secondorder mesenteric arteries. Controls and experimental treatments were carried out in the same tissue, so blinding and randomisation were not used.

| Image acquisition
Two imaging systems were used. The first was a Nikon Eclipse TE300 inverted microscope fitted with a CoolLED pE-300 LED illumination system (488 and 561 nm excitation) and custom designed, dual FITC/ TRITC filter sets (Figure 1a). A 40× 1.3 NA Nikon S Fluor oil-immersion objective lens was used for Ca 2+ imaging experiments, while a 100× 1.3NA Nikon S-Fluor lens was used in experiments imaging mitochondrial membrane potential. The second imaging system was a Nikon Eclipse FNI upright microscope equipped with a Nikon Fluor 40× 0.8 NA water immersion objective lens and a pE-4000 CoolLED system (470 nm). This system was used for K + -channel blocking experiments. All images were acquired by Andor iXon EMCCD F I G U R E 1 ACh-evoked Ca 2+ increases are reproducible. (a) Schematic of widefield microscopy for endothelial cell imaging of intact arteries. (b) Representative Ca 2+ images and kymograph illustrating temporal dynamics of ACh (50 nM)-evoked endothelial Ca 2+ activity. Ca 2+ images show raw fluorescence (left), ΔF/F 0 maximum intensity projection (middle), and temporally colour-coded projection of active Ca 2+ wave fronts (determined by sequential subtraction). The kymographs show changes in Ca 2+ levels across scanlines spanning four (red) or three (orange) cells. (c) Example of raw and pseudocoloured Ca 2+ images and corresponding single-cell Ca 2+ traces (black line average) illustrating the response of a single field of endothelial cells to repeat application of ACh (50 nM, 30-min equilibration between recordings). (d) Summary data showing no significant changes in the number of cells activated by successive ACh applications (left; 306 ± 25 cells for repeat 1, 311 ± 16 cells for repeat 2, n = 5) and the mean amplitude of the Ca 2+ response (right; 0.26 ± 0.04 ΔF/F 0 for repeat 1, 0.26 ± 0.05 ΔF/F 0 for repeat 2; n = 5). All image scale bars = 50 μm cameras (1024 × 1024) using MicroManager v1.4.22 (Edelstein et al., 2014).
Photolysis of cIP 3 was achieved using a Rapp Optoelectronics flash lamp (00-325-JML-C2) at 300 V, which produced light of $1 ms duration. The flashlamp output was passed through a 395 nm short pass filter into a 1250 μm diameter light guide (Figure 3a). The light-guide was coupled to the epi-illuminator of the TE300 microscope, and the output was focussed on the endothelium using broadband light. For each imaging session, broadband light was used to identify the position of the uncaging region ($70 μm diameter) and determine which endothelial cells were directly activated by the spot photolysis system.
In some experiments, the extent of IP 3 uncaging was graded by attenuating the photolysis light power using neutral density filters placed in the excitation path. The neutral density filters had ODs of 0.5 (27% transmission at 395 nm; product code NE505B; Thor Labs, UK), 0.2 (63% transmission at 395 nm; NE502B; Thor Labs, UK), or 0.1 (80% transmission at 395 nm; NE501B; Thor Labs, UK). These experiments were performed such that the 27% transmission was recorded first, followed by 63% transmission with 15 min rest between photolysis events, and so on.

| Experimental protocols
In experiments that examined the effect of CBX and 18βGA on IP 3mediated Ca 2+ release, ACh-or cIP 3 -evoked endothelial Ca 2+ activity was measured at 10 Hz. Baseline Ca 2+ activity was recorded for 30 s, and then endothelial Ca 2+ activity evoked by ACh (50 nM) or photolysis of cIP 3 . The same arteries were then incubated with CBX (100 μM, 5 min) or 18βGA (40 μM, 45 min). ACh/cIP 3 -evoked Ca 2+ activity was then recorded again. In separate experiments, this protocol was repeated with an additional washout period of 1 h (PSS, 1.5 mlÁmin −1 ) before an additional recording was taken.
In experiments assessing the effect of K + -channel blockade on endothelial Ca 2+ signalling, ACh-evoked (50 nM) Ca 2+ activity was assessed in the absence and then the presence of either the K Ca 2.x channel blocker, apamin (100 nM, 10 min pre-incubation) or the K Ca 3.1 channel blocker, TRAM-34 (1 μM; 10 min pre-incubation).
After their introduction, K + -channel blockers remained in the PSS until washout, as indicated. In all experiments, there was a minimum of 15 min between successive stimulations for responses to recover.
In a separate series of experiments, the effects of CBX and 18βGA on ΔΨ M were investigated while changes in the plasma membrane potential were prevented. In these experiments, a high K + /Ca 2+ -free PSS (79.7 mM NaCl, 2 mM MOPS, 70 mM KCl, 1.2 mM NaH 2 PO 4 , 5 mM glucose, 0.02 mM EDTA, 2 mM NaPy, 1 mM MgCl, 1 mM EGTA) was used to prevent plasma membrane potential changes.
In experiments where cell viability was assessed, propidium iodide (1.5 μM) was added into the PSS, 100 images were acquired and an average image intensity projection generated using Fiji (Schindelin et al., 2012). Propidium iodide was then washed out (10 min) with PSS and the experiment continued.
In experiments in which Ca 2+ store content was assessed, the SERCA inhibitor cyclopiazonic acid (CPA; 5 μM) was applied in a Ca 2 + -free bath solution. By inhibiting SERCA, CPA disrupts the store uptake-leak equilibrium so that the leak may be measured as a rise in cytoplasmic Ca 2+ concentration and integrated to determine the store content. In these experiments, CBX was used to inhibit IP 3 receptor activity and the effectiveness of block confirmed by the absence of a response to ACh (50 nM). The bathing media was then changed to Ca 2+ -free PSS containing CPA and the whole-field Ca 2+ signal profile measured over the next 15 min. The area under the Ca 2+ discharge curve was calculated as a measure of store Ca 2+ content and compared to controls.

| Ca 2+ signal analysis
Single-cell Ca 2+ signals were extracted from Ca 2+ imaging data as previously described (Wilson, Lee, & McCarron, 2016). In brief, automated Fiji macros were used to extract cell coordinates and track cell positions between datasets. Single-cell Ca 2+ signals were then extracted and processed using a custom algorithm written in the Python programming language (Wilson, Lee, & McCarron, 2016;Wilson, Saunter, Girkin, & McCarron, 2015;Wilson, Saunter, Girkin, & McCarron, 2016). Raw fluorescence (F) signals were converted to baseline-corrected fluorescence intensity (F/F 0 ) by dividing each intensity measurement by the average value of a 100-frame baseline period at the start of each trace. F/F 0 signals were smoothed using a 21-point third-order polynomial Savitzky-Golay filter, and key signal parameters (e.g., amplitude, frequency, number of cells, and time of event) extracted automatically. Analysis of cIP 3 -evoked Ca 2+ activity was restricted to those cells in which cIP 3 was photolysed. This was achieved by applying a mask restricted to the photolysis region. The photolysis region occupied a fraction of the overall field, so these experiments had a lower number of cells per experiment than those of ACh-evoked signalling.
To visualise Ca 2+ wave propagation, we created images of active Ca 2+ wavefronts by calculating ΔF/F 0 for each image in the recording. For cIP 3 -evoked Ca 2+ experiments, a maximum intensity projection of the first 3 images immediately following uncaging was taken, ensuring that only signal from the uncaging area is presented.
This only differs in Figure 5, where a maximum intensity projection of the first 5 s immediately following uncaging is presented for each experimental condition to compare propagation extent. For ACh experiments, a maximum intensity projection of the 60 s after ACh onset was taken. A JET LUT was then applied to the images. Since all experiments were paired, images were contrast matched for control and treatment. To visualise mitochondria, images were loaded into FIJI and an unsharp mask applied, the background was subtracted, a Gaussian blur was applied, and the local contrast was enhanced. To get a fluorescence intensity trace, images were stabilised, and a region of interest was placed over the mitochondria of interest.

| Data and statistical analysis
Graphical summary data represent averaged, paired responses in arteries from ≥5 different animals. Data are summarised as mean ± SEM. Data were assessed for variance homogeneity (F-test) before statistical tests were performed. Raw peak F/F 0 responses were analysed statistically using either a paired Student's t-test or a

| RESULTS
In the endothelium, muscarinic receptor stimulation, using the physiological agonist ACh (50 nM), evoked heterogeneous increases in Ca 2+ .
The Ca 2+ rise propagated regeneratively, initially within and subsequently between cells, to generate multicellular Ca 2+ waves ( Figure 1a,b, Video S1). These Ca 2+ waves are the result of IP 3 -dependent Ca 2+ release from intracellular stores Wilson, Lee, & McCarron, 2016). In control experiments, repeated application of ACh evoked reproducible increases in Ca 2+ and propagating waves (Figure 1c,d). There was no difference in the number of cells or the amplitude of responses on each activation with ACh.
It is unclear how these waves are transmitted between neighbouring endothelial cells. A prime candidate for the transmission is the movement of small molecules such as Ca 2+ or IP 3 through gap junctions between endothelial cells (Pohl, 2020). To explore the role of gap junctions in the intercellular propagation of Ca 2+ waves, we exam- These results initially appeared to be consistent with a contribution of gap junctions to the propagation of endothelial cell Ca 2+ waves. However, the decrease in amplitude of ACh-evoked Ca 2+ signals occurred approximately uniformly across all endothelial cells-an unexpected finding, as these drugs would not be expected to reduce Ca 2+ signals in cells directly activated by ACh. These results raised the possibility that CBX and 18βGA may each directly inhibit IP 3 -evoked Ca 2+ release.
To determine if CBX and 18βGA interfere with the ability of IP 3 to evoke Ca 2+ release, the effects of the drugs on Ca 2+ signals evoked by the photolysis of cIP 3 were examined (Figures 3 and 4). Uncaged IP 3 bypasses plasma membrane receptors to directly activate IP 3 receptors. Photolysis of cIP 3 , in a 70 μm diameter spot, triggered an immediate rise in Ca 2+ in the photolysis region followed by multicellular Ca 2+ waves that propagated across cells away from the photolysis spot (Figure 3a (c) show raw baseline Ca 2+ images, ACh-evoked Ca 2+ activity images (pseudocoloured max ΔF/F 0 ), and corresponding single-cell Ca 2+ traces (black line average) obtained from the same field of endothelial cells before and after incubation with the indicated inhibitor; (b, d) (left panels) paired summary data plots showing significant decrease in the number of cells activated by ACh, for CBX (316 ± 36 cells for control, 93 ± 60 cells for CBX; n = 5) and for 18βGA (349 ± 27 cells for control, 129 ± 64 cells for 18βGA; n = 5); panels (b) and (d) (right panels) show the mean amplitude of the ACh-evoked Ca 2+ response before and after CBX (0.29 ± 0.08 ΔF/F 0 for control, 0.04 ± 0.03 ΔF/F 0 for CBX; n = 5) and 18βGA (0.29 ± 0.05 ΔF/F 0 for control, 0.04 ± 0.02 ΔF/F 0 for 18βGA; n = 5). * P<0.05, significantly different as indicated; paired t-test. All image scale bars = 50 μm magnitude of initiating cIP 3 -evoked Ca 2+ release occurring after increasing 18βGA incubation times shows a strong correlation (gradient of 0.82 and R 2 value of 0.95; Figure 5d). As the decrease in outward signal propagation was the same after intervention with either 18βGA (Figure 5b) or a decrease in photolysis light intensity (Figure 5a), this suggests that a major mechanism of action of the reported gap junction blockers is to inhibit IP 3 -mediated Ca 2+ release in the vascular endothelium of the mesenteric arteries.
CBX and 18βGA have each been reported to evoke cell death (Hasan et al., 2016;Lee et al., 2010;Yu et al., 2014). To investigate whether CBX and 18βGA decreased IP 3 -evoked Ca 2+ release by inducing cell death, the reversibility of the drugs was examined. IP 3evoked Ca 2+ signalling evoked by cIP 3 or ACh was examined before incubation, after incubation, and after washout (1 h) of CBX ( Figure 6) or 18βGA, (Figure 7). The inhibitory effects of CBX on Ca 2+ release evoked by photolysis of IP 3 (Figure 6a To further test whether CBX and 18βGA caused cell death, we used propidium iodide staining as an assay of cell membrane permeability and apoptosis. Neither CBX nor 18βGA caused an increase in propidium iodide staining ( Figure S1). Thus, in the present study, CBX and 18βGA did not evoke endothelial cell death, as measured by the reversibility of the IP 3 -evoked Ca 2+ responses and by propidium iodide staining.
F I G U R E 3 cIP 3 -evoked increases in endothelial Ca 2+ levels are reproducible. (a) Schematic of localised photolysis of cIP 3 with simultaneous widefield endothelial cell imaging of intact arteries. (b) Representative Ca 2+ images and kymograph illustrating temporal dynamics of cIP 3 -evoked endothelial Ca 2+ activity. Ca 2+ images show raw fluorescence (left), temporally colour-coded projection of active Ca 2+ wave fronts (determined by sequential subtraction, middle; photolysis region shown by dotted line), and the polar coordinates used for the kymograph. (c) Example of raw and pseudocoloured Ca 2+ images and corresponding single-cell Ca 2+ traces (black line average) illustrating the response of a single field of endothelial cells to repeat photolysis of cIP 3 (30-min equilibration between recordings). Arrow indicates uncaging event. (d) Summary data showing no significant differences in the number of cells activated by successive cIP 3 photolysis events (left; 24 ± 1 cells for repeat 1, 24 ± 1 cells for repeat 2, n = 5) and the mean amplitude of the Ca 2+ response (right; 0.80 ± 0.10 ΔF/F 0 for repeat 1, 0.67 ± 0.05 ΔF/F 0 for repeat 2; n = 5). All image scale bars = 50 μm CBX and 18βGA are known to inhibit small (SK) and intermediate (IK) conductance K + channels  which may alter the plasma membrane potential and have consequences for Ca 2+ store refilling (McCarron, Flynn, Bradley, & Muir, 2000). A block of store refilling could explain the effects of CBX and 18βGA on IP 3 -evoked Ca 2+ release. To determine if the inhibitory effects of CBX and 18βGA arose from K + -channel-dependent changes in membrane potential, IP 3 -evoked endothelial Ca 2+ responses were recorded in the absence and presence of apamin (100 nM, Figure 8a,b), an SK blocker, or TRAM-34 (1 μM, Figure 8c,d), an IK blocker.
As shown in Figure 8a apamin did not alter ACh-evoked Ca 2+ signals, while CBX abolished the response in these same preparations (Figure 8a,b). Again, the effect of CBX was reversible on washout. The mean amplitude of ACh-evoked Ca 2+ signals and the number of AChresponsive cells (Figure 8b) confirms this.
TRAM-34 also failed to alter ACh-evoked endothelial Ca 2+ signalling (Figure 8c,d). The mean amplitude of ACh-evoked Ca 2+ signals and number of ACh-responsive cells (Figure 8d) were unaltered by the K + channel blockers but were subsequently inhibited by CBX. As neither apamin nor TRAM-43 altered IP 3 -mediated Ca 2+ release, it is unlikely that the inhibitory effects of CBX and 18βGA on IP 3 -evoked Ca 2+ release were mediated by K + channel inhibition. The store content was unaltered in the absence and presence of CBX (100 μM, 5 min) as measured using the area under the whole field Ca 2+ signal intensity curve upon addition of CPA (5 μM, 15 min) in a Ca 2+ -free PSS (Figure 8e). The effectiveness of CPA-induced store depletion was confirmed by the absence of a response to ACh (50 nM; not shown).

The concentration of TMRE in mitochondria is governed by
Nernstian function of the mitochondrial membrane potential and plasma membrane potential. To ensure that the effect of CBX and 18βGA arose from depolarisation of ΔΨ M rather than depolarisation of the plasma membrane potential, the plasma membrane potential was clamped using a high K + PSS and the experiments repeated ( Figure 9c). Ca 2+ was omitted from the bathing solution to prevent smooth muscle contraction. In high K + -PSS, CBX, or 18βGA each again rapidly depolarised ΔΨ M (Figure 9c), as revealed by the loss of punctate TMRE staining. As the endothelial plasma membrane potential was clamped by the high K + -PSS, the effect of CBX or 18βGA is on the mitochondria.
Taken together, these data suggest that CBX and 18βGA have pronounced effects on endothelial function by inhibiting IP 3 -evoked Ca 2+ release and depolarising ΔΨ M .

| DISCUSSION
Ca 2+ signals in the endothelium propagate regeneratively among cells to provide the long distance communication essential to coordinate normal vascular function (Lee et al., 2018;Longden et al., 2017;McCarron, Lee, & Wilson, 2017;Tallini et al., 2007;Wilson, Lee, & McCarron, 2016). Movement of small molecules such as IP 3 or Ca 2+ through gap junctions is proposed to underlie Ca 2+ signal propagation and aberrant gap junction function may participate in cardiovascular disease development (Christ, Spray, el-Sabban, Moore, & Brink, 1996;Pohl, 2020). The link between gap junctions and cardiovascular disease has generated a substantial interest in determining the F I G U R E 6 Inhibition of IP 3 -mediated Ca 2+ release by CBX is reversible. Effect of CBX incubation (100 μM, 5 min incubation) and washout (1 h, PSS) on (a-c) cIP 3 -evoked (5 μM) and ACh-evoked (d-f) endothelial cell Ca 2+ signalling. Panels (a) and (d) show cIP 3 -evoked Ca 2+ activity images (pseudocoloured max ΔF/F 0 ), and corresponding single-cell Ca 2+ traces (black line average) obtained from the same field of endothelial cells before and after incubation with, and after washout of, CBX. Arrows indicate uncaging event. (b, e) Mean Ca 2+ response from each cell in the endothelial field shown under each condition. Points are colour coordinated according to plotting density; (c and f ) paired summary data plots showing the effect of CBX incubation and washout on the number of cells activated (left) by cIP 3 (c; 18 ± 2 cells in control vs. 17 ± 2 cells after CBX and 18 ± 2 cells after CBX washout; n = 5) and ACh (f; 315 ± 25 cells in control vs. 76 ± 18 cells after CBX and 332 ± 15 cells after CBX washout; n = 5). The mean amplitude of the Ca 2+ response (c) for cIP 3 was 0.67 ± 0.08 ΔF/F 0 in control and 0.18 ± 0.03 ΔF/F 0 after CBX, and 0.48 ± 0.07 ΔF/F 0 following washout (n = 5). The mean amplitude of the Ca 2+ response (f) for ACh was 0.27 ± 0.05 ΔF/F 0 in control, 0.018 ± 0.007 ΔF/F 0 in CBX and 0.50 ± 0.04 ΔF/F 0 after CBX washout (n = 5). * P<0.05, significantly different as indicated; paired one-way ANOVA with Tukey's multiple comparisons test. All image scale bars = 50 μm The mechanisms by which CBX and 18βGA block gap junctions are unclear (see Willebrords, Maes, Crespo Yanguas, & Vinken, 2017).
18βGA-mediated inhibition of Cx43 may occur via dephosphorylation of type 1 or type 2A protein phosphatases (Guan et al., 1996), and direct interaction with the connexin has also been proposed to occur (Davidson & Baumgarten, 1988). There have been no studies clearly defining the mechanisms behind CBX inhibition of connexin channels (Leybaert et al., 2017). There are several reports of "off-target" effects which may account for some of the effects of 18βGA and CBX on cell-cell communication. Glycyrrhetinic acids bind strongly to mineralocorticoid and glucocorticoid receptors (Armanini, Karbowiak, & Funder, 1983;Kratschmar et al., 2011), inhibit 11β-hydroxysteroid dehydrogenase and act in anti-inflammatory roles through these pathways (Morsy et al., 2019). CBX also shows high affinity for the mineralocorticoid receptor (Armanini, Karbowiak, Krozowski, Funder, & Adam, 1982).
An alteration in K + -channel activity  by CBX and 18βGA could alter the plasma membrane potential and store refilling, providing an explanation for the decreased IP 3 -evoked Ca 2+ release. However, in the present study, there was no effect of either an SK-channel blocker (apamin) or IK-channel blocker (TRAM 34) on IP 3 -mediated Ca 2+ release. This suggests that inhibition of K + -channel activity is an unlikely explanation of CBX-and 18βGA-mediated inhibition of IP 3 -mediated Ca 2+ release in mesenteric artery endothelium.
Another unexpected finding in the present study was the rapid ΔΨ M collapse induced by each of the gap junction blockers. The collapse of ΔΨ M will have wide ranging effects on cell signalling. The F I G U R E 8 Inhibitory action of CBX is not due to blockade of small or intermediate conductance K + channels nor is store content affected by CBX. Effect of (a, b) small (apamin, 100 nM, 10 min incubation) and (c, d) intermediate conductance (TRAM-34, 1 μM, 10 min incubation) K +channel block on ACh-evoked (100 nM) endothelial cell Ca 2+ signalling. (e) Effect of CBX on store content, measured using CPA (5 μM in Ca 2+ free PSS, 15 min). Panels (a) and (c) show ACh-evoked Ca 2+ activity images (pseudocoloured max ΔF/F 0 ), and corresponding single-cell Ca 2+ traces (black line average) obtained from the same field of endothelial cells for a control recording, after incubation with TRAM-34, after incubation with CBX, and after washout of all drugs; (b) paired summary data plots showing the effect of incubation of apamin and washout on the number of cells activated by ACh (left; 157 ± 6 cells in control, 157 ± 6 cells after apamin, 68 ± 9 cells in CBX and 157 ± 7 cells after washout, n = 6). The mean amplitude of the Ca 2+ response (right) was 0.35 ± 0.06 ΔF/F 0 in control, 0.35 ± 0.05 ΔF/F 0 after apamin, 0.04 ± 0.01 ΔF/F 0 after CBX and 0.41 ± 0.03 ΔF/F 0 after washout (n = 5). (d) Paired summary data plots showing the effect of incubation of TRAM-34 and washout on the number of cells activated by ACh (left; 151 ± 7 cells in control, 151 ± 7 cells after TRAM-34, 83 ± 7 cells after CBX, 148 ± 8 cells after washout; n = 7) and the mean amplitude of the Ca 2+ response (right) (0.5 ± 0.1 in control, 0.44 ± 0.09 ΔF/F 0 after TRAM-34,. 0.06 ± 0.02 ΔF/F 0 after CBX and 0.50 ± 0.08 ΔF/F 0 after washout ΔF/F 0 ; n = 7). (e) Summary data showing the effect of CBX incubation on Ca 2+ store content. *P<0.05, significant effect of CBX; paired one way ANOVA with Tukey's multiple comparisons test. All image scale bars = 50 μm F I G U R E 9 CBX and 18βGA each rapidly depolarise the mitochondrial membrane potential. (a) Endothelial cells from en face mesenteric artery preparations were stained with Cal-520 (5 μM, grey) and TMRE (150 nM, red) to visualise the mitochondrial membrane potential (ΔΨ M ). (b) Mitochondria were imaged for 1.5 min while administering PSS (control), CBX (100 μM), or 18βGA (40 μM) at 1.5 mlÁmin −1 under constant flow. Fluorescence intensity traces from individual mitochondria (designated by green arrows in the baseline image) are shown from across the treatment period for PSS, CBX, and 18βGA administration, indicated with by a bar over the trace. (c) Experiments were repeated in Ca 2+ -free, high K + PSS (control), CBX in Ca 2+ -free, high K + PSS and 18βGA in Ca 2+ -free, high K + PSS, and fluorescence intensity traces from individual mitochondria again shown. Examples from single experiments are shown from n = 5 biological replicates yielding similar results. Scale bars = 25 μm depolarisation leading to Hsp90 inhibition-mediated caspase 8 activation (Yang, Myung, Kim, & Lee, 2012) or cytochrome c release and caspase 3 activation (Lee, Kim, Lee, Han, & Lee, 2008). 18βGAinduced mitochondrial membrane changes, and apoptosis occurs in human bladder cancer (Lin et al., 2011), human endometrial stromal (Yu et al., 2014), and human hepatoma cell lines (Hasan et al., 2016).
CBX also induced ΔΨ M collapse in liver mitochondria, resulting in mitochondrial permeability transition and apoptosis (Salvi et al., 2005).
While CBX and 18βGA each depolarised ΔΨ M , we did not observe endothelial cell apoptosis in the present study at the concentrations and incubation times used, as shown by the lack of propidium iodide-positive staining ( Figure S1) and the reversibility of the drug effects on Ca 2+ signalling and ΔΨ M depolarisation. CBX has better water solubility than 18βGA (Leybaert et al., 2017), and therefore, the washout of CBX was more effective than that of 18βGA. Notwithstanding, we did observe that leaving the drug on longer than the $10 min for CBX or $1 h for 18βGA caused a significant, irreversible increase in resting Ca 2+ concentration in some cells (data not shown).
Our results therefore raised the possibility that CBX and 18βGA inhibit IP 3 -mediated Ca 2+ release by ΔΨ M depolarisation. However, depolarisation of ΔΨ M by CBX or 18βGA occurred rapidly (within 90 s) while inhibition of IP 3 -mediated Ca 2+ release developed more slowly (5 min for CBX; 45 min for 18βGA). The differences in time course suggests that ΔΨ M depolarisation alone does not explain the inhibition of IP 3 -evoked Ca 2+ release and that CBX or 18βGA block IP 3 receptors.
Together, our study questions the usefulness of CBX and 18βGA in studies on IP 3 -mediated signal transduction via gap junctions in intact arterial tissue. CBX and 18βGA each inhibit IP 3 -mediated Ca 2+ release and depolarise ΔΨ M .