Multidrug resistance proteins preferentially regulate natriuretic peptide‐driven cGMP signalling in the heart and vasculature

cGMP underpins the bioactivity of NO and natriuretic peptides and is key to cardiovascular homeostasis. cGMP‐driven responses are terminated primarily by PDEs, but cellular efflux via multidrug resistance proteins (MRPs) might contribute. Herein, the effect of pharmacological blockade of MRPs on cGMP signalling in the heart and vasculature was investigated in vitro and in vivo.


| INTRODUCTION
cGMP is a common intracellular second messenger responsible for conveying many of the cardiovascular actions of NO and natriuretic peptides (Kuhn, 2016;Papapetropoulos et al., 2015). Generated by NO-sensitive (GC-1 and GC-2) and natriuretic peptide-sensitive (GC-A and GC-B) GCs, this ubiquitous molecule possesses a number of cytoprotective roles including vasodilator, anti-leukocyte, anti-platelet, anti-fibrotic and anti-hypertrophic actions (Preedy et al., 2020).
Cellular levels of cGMP are dynamically regulated in a spatial and temporal fashion, the result of a balance between GC-mediated formation and enzymatic hydrolysis by PDEs. Indeed, targeting of these synthetic (e.g. sGC stimulators and NO donors) and degradative (e.g. PDE5 inhibitors) pathways has proven clinically effective in a number of cardiovascular diseases including heart failure and pulmonary arterial hypertension (Galie et al., 2005;Ghofrani et al., 2013;Roberts et al., 1992). The discrete localisation of these synthetic and degradatory enzymes enables compartmentalisation of cGMP-regulated processes to distinct cellular localities. This phenomenon has been established largely in cardiomyocytes and vascular smooth muscle cells (VSMCs) through the utilisation of cGMP-specific FRET sensors (Calamera et al., 2019;Russwurm & Koesling, 2018;Sinha et al., 2013;Subramanian et al., 2018;Thunemann et al., 2013) but also by functional pharmacological assessment of selelective GC ligands (i.e. NO vs. natriuretic peptides) and PDE inhibitors (Baliga et al., 2018;Bubb et al., 2014;Kokkonen & Kass, 2017;Moltzau et al., 2014;Zhao et al., 1999). Intriguingly, these cutting-edge approaches have also given rise to the possibility that cellular efflux plays a further, active role in terminating the biological actions of cGMP (Krawutschke et al., 2015) (and also in intercellular transfer, Menges et al., 2019). Indeed, appreciation of the compartmentalisation and cellular efflux of sibling cAMP in vitro and in vivo is comparatively well established (Sassi et al., 2008;Sinha et al., 2013;Vettel et al., 2017;Zoccarato et al., 2015). Cellular efflux of cyclic nucleotides is thought to depend on the action of multidrug resistance proteins (MRPs) (Jedlitschky et al., 2000;Schuetz et al., 1999). These energy-dependent, transmembrane transporters belong to the superfamily of large adenosine-5 0 -triphosphate (ATP)-binding cassette (ABC) proteins, which include 10 MRPs, two sulfonylurea receptors and the cystic fibrosis transmembrane-conductance regulator (Dean et al., 2001;Deeley et al., 2006). ABC transporters are unidirectional (i.e. transport from the intracellular to the extracellular space) and ATP dependent and have a hugely diverse substrate profile that includes xenobiotics and therapeutic agents (Dean et al., 2001;Deeley et al., 2006). Two MRP isozymes are thought to primarily underpin efflux of cGMP and cAMP; MRP4 (ABCC4) and MRP5 (ABCC5), along with a modest contribution from MRP8 (ABCC11; Slot et al., 2011). Both transporters are widely expressed in the cardiovascular system, being found in VSMCs, endothelial cells and cardiomyocytes Slot et al., 2011), and are inhibited by MK571. Indeed, this pharmacological tool, alongside MRP4 À/À and MRP5 À/À mice, has been used to establish cellular cGMP efflux as a key mechanism for regulation of cGMP signalling. For example, MRP4 expression is significantly increased in both VSMCs and endothelial cells at the mRNA and protein levels in lungs from patients with pulmonary hypertension (PH) compared with controls and is the same in a murine model of pulmonary hypertension. In accord, administration of MK571 or MRP4 (Abcc4) gene deletion causes resistance to hypoxia-induced What is already known • Cyclic nucleotide compartmentalisation spatially and temporally constrains cGMP and cAMP signalling.
• cGMP and cAMP cellular efflux may represent an important mechanism for regulating cyclic nucleotide signalling.

What this study adds
• Pharmacological inhibition of multidrug resistance proteins (MRPs) promotes cGMP signalling in the heart and vasculature.
• cGMP production by membrane-spanning GCs, triggered by natriuretic peptides, is governed preferentially by MRP activity.

What is the clinical significance
• Augmenting cGMP signalling is established to be beneficial in a number of cardiovascular diseases.
• MRPs inhibition might represent alone or adjunct therapy to optimise the therapeutic benefits of cGMP. pulmonary hypertension (Hara et al., 2011). Notably, the effect of MRP inhibition is accentuated in the presence of sildenafil, suggesting that in the absence of PDE degradation, MRP activity is significantly elevated and takes on a more substantial role in cGMP inactivation.
Indeed, genetic deletion of MRP4 leads to significantly increased heart weight, cardiomyocyte size and left ventricular (LV) wall thickness with a concomitant increase in cardiac atrial natriuretic peptide (ANP; Nppa) gene expression (Sassi et al., 2012). Such mice also exhibit an extended bleeding time demonstrating that MRP4 promotes platelet aggregation by decreasing intracellular cAMP and cGMP (Borgognone & Pulcinelli, 2012). Interestingly, MRP5 expressio is increased in the hearts of patients suffering from ischaemic cardiomyopathy and dilated cardiomyopathy, suggesting that this MRP isozymes may contribute to the pathology and that inhibition of MRP5 might be an attractive target to increase beneficial intracellular cyclic nucleotide concentrations (Dazert et al., 2003) Yet, MRP5 À/À mice do not have an overt phenotype, a finding that is also true for MRP4 À/À /MRP5 À/À animals (Borst et al., 2007). This is at odds with the adverse cardiovascular phenotype in MRP4 À/À mice (Sassi et al., 2012). Therefore, a greater understanding of the physiological and pathological role(s) of MRPs in the regulation of cyclic nucleotide efflux is needed if these transporters are to be established as key players in terminating cGMP signalling and, consequently, as potential target in promoting the cytoprotective functions of cGMP for therapeutic benefit in cardiovascular disease.
Herein, in order to establish (patho)physiological role(s) for MRPs in regulating cGMP efflux and signal termination in the cardiovascular system, we have used functional pharmacological characterisation of vascular smooth muscle proliferation, isolated blood vessel reactivity, acute BP responses and an experimental model of heart failure in the absence of presence of MRP inhibition.  (Lilley et al., 2020). All animal procedures were conducted in accordance with the UK Home Office Animals (Scientific Procedures) Act of 1986 and were approved by a local animal welfare and ethical review board. Animals were housed in a temperaturecontrolled environment with a 12-h light-dark cycle. Food and water were accessible ad libitum. For in vivo experimentation, animals were randomly assigned to interventions and the experimenter was blinded to treatment.
For cell-and tissue-based studies, interventions were randomly assigned but the experimenter was not blinded to treatment.

| Cell proliferation
Human coronary artery smooth muscle cells (hCASMCs; Passages 5-7; Lonza, Basel, Switzerland) were seeded onto six-well plates at a density of 30,000 cells per well. Cells were initially grown for 24 h in routine cell culture medium  Lonza) with SmGM™-2 BulletKit containing FBS (5%), human EGF insulin, human fibroblastic growth factor and gentamicin/ amphotericin B (proprietary concentrations). Subsequently, the cells were serum starved (FBS: 0.5%) for 48 h to synchronise cell cycles.
Cells were then incubated in medium containing 5% serum in the presence of vehicle (sterile ddH 2 O) or pharmacological treatment, and cells (stained with trypan blue diluted 1:1 to identify dead cells) counted at 0-, 24-, 48-, 72-and 96-h intervals using a haemocytometer.
Individual plates were set up with the same treatments, in duplicate wells, one for each time point. Cells seeded at an initial density of 30,000 cells per well grew at an exponential rate under control conditions without becoming confluent ( Figure 1). This proliferative profile was explored in the absence and presence of MK571 (30 nM to 30 μM), ANP (100 nM), the NO donor diethylenetriamine-NONOate (Deta-NO; 10 μM) or combinations thereof.

| cGMP assay
Intracellular and extracellular cGMP concentrations were measured by ELISA (GE Healthcare, Hatfield, UK) following 24 h of treatment in the same cells used for the proliferation assays described above. Each well was treated with the non-selective PDE inhibitor IBMX (100 μM; Sigma Aldrich) for 30 min prior to cell and supernatant harvest to prevent breakdown of cGMP during cell processing. Cell pellet (for intracellular cGMP) and supernatant (for extracellular cGMP) were frozen and stored at À80 C until use. In vivo cGMP measurements were determined using an identical ELISA (as above), determined in whole heart homogenates (for intracellular cGMP) and plasma (for extracellular cGMP), treated with IBMX (100 μM; as above) and frozen and stored at À80 C until use. cAMP concentrations were also evaluated in heart homogenates and plasma by ELISA (GE Healthcare). Extraction of cyclic nucleotides and ELISA were conducted as the per manufacturer's instructions.

| Organ bath pharmacology
Mice (C57/BL6; male and female; 20-30 g) were killed by cervical dislocation. The thoracic aortae were carefully removed, cleaned of connective tissue and cut into three to four ring segments of approximately 4 mm in length. Aortic rings were mounted in 10-ml organ baths or ACh produced relaxations of less than 50% of the phenylephrinetone, tissues were discarded. After another wash period, the vessels were again contracted with phenylephrine (EC 80 ) and cumulative concentration response curves to MK571 (100 nM to 30 μM) or probenecid (1 μM to 1 mM) constructed; responses to MK571 were also studied in the presence of the NO-sensitive GC (GC-1/2) antagonist ODQ (5 μM) or the G-kinase inhibitor KT5823 (2 μM). In further studies, tissues were pre-contracted with an EC 80 concentration of phenylephrine and then concentration-response relationships for ACh (1 nM to 3 μM), ANP (1 pM to 300 nM), the NO donor Sper-NO (1 nM to 30 μM) or the β-adrenoceptor agonist isoprenaline (300 pM to 300 nM) determined in the absence and presence of a subthreshold concentration (3 μM; 15 min pre-incubation) of MK571. with body temperature maintained at 37 C. The left common carotid artery was exposed, isolated and cleaned of any surrounding tissue. A small incision was made in the carotid, and the catheter inserted whilst submerged in saline to ensure no bubble form inside the catheter tip.

| Acute BP measurement in anaesthetised mice
The tip of the catheter was placed into the aortic arch, securely fastened and the transmitter body placed s.c. on the right flank. The incision was stitched and each animal received post-operative analgesia consisting of 0.3-μg Vetergesic ® (buprenorphine; Abbott Laboratories Ltd) in 0.5-ml saline (s.c.). Animals were left to recover for a minimum of 7 days, after which haemodynamic recordings were taken for 64 h over the weekend to minimise noise disturbance. Mean arterial BP (MABP), heart rate (HR) and activity were recorded for 2 min at 15-min intervals using Dataquest Art Acquisition System (Data Sciences International). Baseline haemodynamic measurements were taken 1 week before dosing.
The daily drinking water consumption was measured each day, and, where appropriate, water bottles were replaced with bottles containing MK571 at a concentration that entailed each mouse received 25 mgÁkg À1 Áday À1 (based on a mean consumption of 4 mlÁmouse À1 Áday À1 ). A 24-h time period was used for analysis starting from 1 pm Saturday and ending 1 pm Sunday (12 h light/dark), with dosing commencing 24 h before the start of this time period.
Mice were placed under an operating microscope in the supine position with body temperature maintained at 37 C. An incision was made in the abdominal cavity and the abdominal aorta was separated from the surrounding tissue at the suprarenal level by moderate dissection. Aortic constriction was performed by tying a 4-0 surgical thread against a 25-gauge needle between the superior mesenteric and renal arteries. This produces a $30% constriction of the luminal diameter. For sham operations, the 4-0 surgical thread was passed under the aorta and removed without tying it against the needle. To close internal and external incisions, 6.0 absorbable and nonabsorbable sutures were used, respectively, and local anaesthetic (lidocaine hydrochloride; 2.0% w/v; Dechra, UK) was administered.
Mice were monitored regularly for 3 days post-surgery. MK571 (25 mgÁkg À1 Áday À1 ) was delivered in the drinking water (changed every other day, based on average water consumption) starting at Day 21 following abdominal aortic constriction and maintained throughout the remaining 3 weeks of study. To

| Histology, staining and imaging
For murine studies, the isolated left ventricles were cut transversely below the mitral valves, fixed in 10% formalin for 24 h and then stored in 70% ethanol before embedding in paraffin wax and sectioning (6-μm transverse sections). Human heart samples from 'healthy'

| Picrosirius red staining
Tissue slides were dewaxed, rehydrated and stained with Picrosirius Red (0.1% w/v) to visualise collagen fibres in the heart. Images were obtained using a Nanozoomer S210 Slide Scanner (Â40 objective, Hamamatsu, Japan) and analysed by threshold analysis using ImageJ (National Institutes of Health).

| Quantitative RT-PCR and immunoblotting
Left ventricles were snap frozen, broken down using a pestle and mortar and then homogenised using QIAshredder technology. RNA was extracted using a Fibrosis Tissue extraction kit (as above; Qiagen) and quantified using a NanoDrop spectrophotometer (ThermoFisher

| Statistical analyses
The data and statistical analysis comply with the recommendations on experimental design and analysis in pharmacology (Curtis et al., 2018).
All data are reported as mean ± SEM, where n denotes the number of mice used in each group. Statistical analyses were conducted using GraphPad Prism (Version 8; GraphPad Software, USA). For the comparison of two groups of data, a two-tailed, unpaired Student's t-test was used. When comparing three or more groups of data, either a one-way or two-way ANOVA followed by a Šidák multiple comparisons test was used with adjustment for multiplicity. A P-value < 0.05 was considered statistically significant, and the P-values presented in     Figure S1). These findings suggest that efflux of cGMP through MRP4/MRP5 (i.e. MK571-inhibitable MRPs) regulates the cellular response of hCASMC to NO donors and ANP, but that activation of the membrane-spanning GC-A by ANP is subject to significant cGMP efflux that is sensitive to MRP blockade.

| Effect of MRP inhibition on vascular reactivity
To determine if cell-based findings were mirrored in a functional phar-

| Effect of MRP inhibition on acute changes in BP
As a whole, in vitro observations suggest that MRP4 and/or MRP5 play a key role in regulating cGMP-driven responses in the vasculature and that this phenomenon is more important in with respect to natriuretic peptide/particulate GC signalling compared with NO-sensitive GC/cGMP processes. To determine if such an influence was also functionally apparent in vivo, the effect of MRP inhibition on NO-and ANP-induced acute changes in BP was explored in anaesthetised mice.
Administration of MK571 at doses ranging from 0.1 μgÁkg À1 to 3 mgÁkg À1 caused little or no change in MABP or HR (Figure 3a,b).
Accordingly, the highest dose (3 mgÁkg À1 ) of MK571 was used in subsequent experiments to assess the effect of MRP inhibition on haemodynamic responses to NO and ANP. The NO donor SNP (1, 3 and 10 μgÁkg À1 ; i.v. bolus) and ANP (1, 10 and 100 μgÁkg À1 ; i.v. bolus) were chosen due to their ability to elicit dose-dependent reductions in MABP (Madhani et al., 2006). SNP produced a dose-dependent decrease in MABP, which was not significantly altered when combined with MK571 ( Figure 3c). SNP also elicited a dose-dependent increase in HR, but this was not changed by MK571 (Figure 3d). ANP also produced dose-dependent decreases in MABP, however in this T A B L E 2 Absolute and fold change in EC 50 values for vasorelaxant responses to ACh, atrial natriuretic peptide (ANP), spermine-NONOate (Sper-NO) and isoprenaline (Iso) in mouse isolated aorta in the absence and presence of the multidrug resistance protein (MRP) inhibitor MK571 (3 μM) F I G U R E 3 Changes in mean arterial BP (MABP; a, c, e) and heart rate (b, d, f) in anaesthetised mice following bolus doses of MK571 (0.1 μgÁkg À1 to 3 mgÁkg À1 , i.v.; a, b), or bolus doses of sodium nitroprusside (SNP; 1-10 mgÁkg À1 , i.v.; c, d) or atrial natriuretic peptide (ANP; 1-10 μgÁkg À1 , i.v.; e, f) in the absence and presence of MK571 (3 mgÁkg À1 ). Data are shown as mean ± SEM. Statistical analysis by two-way ANOVA across the entire dose-response curve. *P <0.05 (adjusted for multiplicity) case, the hypotensive response to ANP was significantly enhanced in the presence of MK571 (Figure 3e). The effects of ANP on HR were not significantly different following MK571 administration (Figure 3f).
These data imply that MRP inhibition has a preferential ability to promote ANP/GC-A/cGMP signalling in the vasculature compared with NO/GC-1/cGMP pathway, but that basally, MRP inhibition does not significantly alter cGMP-driven vasodilation, perhaps arguing that in healthy animals intracellular levels of cGMP are not sufficient to be affected by MRP-driven efflux.

| Effect of MRP inhibition on chronic changes in blood pressure
In vivo studies in anaesthetised mice suggested that MRP inhibition acutely enhances the hypotensive activity of ANP, but not NO, substantiating earlier in vitro observations. Investigations were therefore undertaken to establish whether MK571 was able to modulate BP in conscious telemetered mice in vivo over a 24-h period. Mice Pressure overload produced the expected increase in LV wall thickness and LV dilatation, and reduced contractility (i.e. fractional shortening and ejection fraction), characteristics of an heart failure phenotype (although this did not result in a significant reduction in cardiac output or stroke volume, Figures 5 and 6). Remarkably, the vast majority of these indices of disease severity were reversed in the presence of MK571, manifesting as both improved cardiac morphology and enhanced myocardial contractility (Figure 5a-f). This was mirrored by qPCR analysis of the cardiac expression of prohypertrophic and pro-fibrotic markers which were increased in response to pressure overload, with the exception of SERCA2, which was decreased in line with a previous report (Nagai et al., 1989).
These effects were reversed in the presence of MK571 ( Figure S2).
The anti-fibrotic actions of MRP inhibition were confirmed by assessment of collagen deposition ( Figure S3). Importantly, MK571 did not significantly reduce MABP following abdominal aortic constriction, indicating that the beneficial actions of MRP inhibition are not due to indirect effects on BP.

| Expression of MRP4 and MRP5 in heart failure
Molecular analysis revealed contrasting effects of pressure overload on MRP subtype expression. MRP4 mRNA levels trended towards a reduction in experimental heart failure, and this was corroborated at the protein level ( Figure S4). However, expression of MRP5 mRNA was significantly increased in response to abdominal aortic constriction ( Figure S4). These changes were largely mirrored in tissue from patients with heart failure with the exception of MRP4 expression in ischaemic patients, suggesting that this mouse model closely mimics the human medical condition ( Figure S4).

| Effect of MK571 on cGMP levels in vivo following pressure overload
To evaluate whether the advantageous outcome associated with To elucidate potential functional role(s) of MRPs in the cGMPmediated inhibition of hCASMC proliferation, the MRP4/MRP5 inhibitor MK571 (Chen et al., 2002;Jedlitschky et al., 2012;Reid et al., 2003) was employed. MK571 produced a concentrationdependent reduction in hCASMC growth in the absence of exogenous cGMP-elevating agents. Because healthy VSMCs do not express NO synthase or natriuretic peptides, it is likely this innate anti-proliferative action of MK571 is a consequence of reducing export of basal cGMP production by the inherent GCs, a phenomenon well established for the NO-sensitive enzymes, but less so for the particulate isozymes (Potter & Garbers, 1992;Wolin et al., 1982).
Moreover, a concentration of MK571 that did not significantly reduce proliferation per se augmented the anti-proliferative effects of subthreshold concentrations of both NO and ANP. Such findings imply that cellular efflux is a key mechanism regulating the antiproliferative effects of cGMP, regardless of the synthesising GC (i.e. NO or natriuretic peptide sensitive). This concept was corroborated, to a certain degree, by measurements of intracellular and extracellular cGMP levels. Here, MRP inhibition alone tended to increase the intracellular:extracellular cGMP ratio, indicative of a reduction in transport out of the cell. However, the low basal cGMP concentrations entailed that such changes were modest at best. The NO donor, Deta-NO, tended to increase the mean intracellular cGMP concentrations, but MK571 was unable to increase this further. In contrast, the effect of MRP inhibition on ANP-mediated rises in cGMP was more clear cut. Interestingly, addition of ANP actually reduced the intracellular:extracellular cGMP ratio, driven by a subtle rise in intracellular levels but a dramatic increase in extracellular cGMP concentrations. In addition, in the presence of MK571, this balance was restored to a basal phenotype in which intracellular cGMP was significantly greater than extracellular cGMP. This pronounced effect of MRP inhibition on ANP responses fits with the particulate nature of this cognate GC, as synthesis of cGMP in the juxtamembrane region by GC-A should be in proximity to MRPs and thereby more overtly influenced by cGMP efflux. The lack of effect of MK571 and NO donor alone, or combination of the two, on cGMP F I G U R E 6 Mean arterial BP (MABP; a), heart weight to tibia length ratio (HW/TL; b), left ventricular weight to tibia length ratio (LVW/TL; c) and left ventricular mass (LV mass; d) in sham mice and animals undergoing abdominal aortic constriction (AAC; 6 weeks) in the absence and presence of MK571 (25 mgÁkg À1 Áday À1 ; p.o.). Data are shown as mean ± SEM. Statistical analysis by one-way ANOVA with Šídák post hoc test. *P <0.05 (adjusted for multiplicity) levels suggest that the NO-sensitive GC generates cGMP in a more diffuse, cytoplasmic region that is less sensitive to efflux via MRPs.
These observations confirm and extend previous reports indicating that cGMP extrusion plays a critical role in terminating signalling in vascular tissue (Krawutschke et al., 2015), although herein it is demonstrated that natriuretic peptide-stimulated cGMP production is more susceptible to MRP-mediated regulation. Published work also supports the notion that cGMP produced by the membrane-spanning GCs is maintained in proximity to the membrane by PDE2 and PDE3, thereby accentuating the consequences of cellular efflux Fischmeister et al., 2006;Krawutschke et al., 2015).
Echoing findings in cell-based experiments, MK571 and the structurally distinct MRP inhibitor probenecid (Hamet et al., 1989) elicited concentration-dependent relaxations of mouse aorta. Again, the ability of MRP inhibitors per se to relax vessels probably reflects intrinsic turnover of cGMP production and turnover as a result of NO production in the endothelium and GC-1/2 activity in the smooth muscle.  (Huang et al., 1995;Madhani et al., 2003;Scotland et al., 2005). Whether this is a consequence of cGMP signal compartmentalisation within the vascular smooth muscle remains to be determined. Specifically, that ACh-dependent NO release stimulates VSMC GC-1/2 but the cGMP generated is not of sufficient magnitude to be affected by MRPs, whereas cGMP production stimulated by exogenous NO (i.e. Sper-NO) is more global, activating the entire GC-1/2 enzyme pool. This interpretation is supported by previous findings that increasing concentrations of ACh do not elevate extracellular cGMP in rat aorta, in contrast to the NO donor, SNP (Schini et al., 1989). These data imply that greater cytosolic cGMP is generated by pharmacological concentrations of NO, spilling over into the membrane region and resulting in increased extrusion by MRPs. It should also be noted that the 3 μM (subthreshold) concentration of MK571 employed in these studies produced little or no effect on vascular tone per se, whereas higher concentrations had a significant vasorelaxant activity. However, at higher concentrations of MK571, it was not possible to maintain precontraction with phenylephrine. Indeed, an arguably larger vascular effect of MRP inhibition has been shown in murine aorta in response to the NO donor, S-nitrosoglutathione, using 10 μM MK571 (Krawutschke et al., 2015). Thus, although the increases in potency observed in Krawutschke et al. and  MPRs also functionally regulate cAMP signalling in the vasculature (Sassi et al., 2008).
Can this preferential activity of MRPs to regulate natriuretic peptide-sensitive GC signalling be explained by subcellular colocalisation? Currently, there are little or no data defining MRP protein-protein interactions with membrane-spanning GCs. However, MRP4 is localised in caveolin-1-enriched membrane fractions in vascular smooth muscle (Sassi et al., 2008) and both GC-A and GC-B colocalise with caveolins in the cardiovascular system (Chen et al., 2012;Doyle et al., 1997). In some situations, this phenomenon may also be of relevant to NO-sensitive GC because of its ability to translocate to the plasma membrane via the chaperone Hsp 90 (Venema et al., 2003).
In addition, a fraction of endogenous GC-1/2 within the heart is found in caveolin-3-positive caveolae (Tsai et al., 2012). When present in the caveolae, the enzyme is protected against oxidation that can occur from vascular injury, which inactivates the catalytic haem domain in the β subunit. Therefore, the enzyme remains active compared with its cytosolic counterparts (Tsai et al., 2012). Overall, this demonstrates that not only can MRP inhibition potentiate particulate GC produced cGMP but also in disease environments characterised by oxidative stress, MRP inhibition may also potentiate cGMP generated from membrane located sGC, amplifying the signal, thus expanding the therapeutic potential of MRP inhibitors.
The vasorelaxant effect of MK571 per se appears to be an isolated vessel phenomenon, because MRP inhibition did not significantly affect BP in anaesthetised or conscious mice, despite the compound having a relatively long plasma half-life (2-3 h in humans; Margolskee, 1991) and the dose used (25 mgÁkg À1 Áday À1 ) having previously been shown to effectively reverse experimental pulmonary hypertension (Hara et al., 2011). However, MK571 did exhibit a preference to potentiate GC-A activation as evidenced by significantly enhanced reductions in BP elicited by acute bolus doses of ANP (but not NO). This suggests in situations where endogenous circulating levels of NPs are increased (e.g. heart failure and pulmonary hypertension) that MRP inhibition will potentiate particulate GC activity, and the lack of an observable effect of MRP inhibition on MABP, in healthy mice, may be due to the low basal circulating concentrations of NPs. This should not be a surprise as the MRP4 À/À , MRP5 À/À and MRP4 À/À /MRP5 À/À mice lack an overt basal phenotype, suggesting that under healthy conditions, MRPs are not essential to vascular homeostasis (Borst et al., 2007). However, longer term MRP4 deficiency results in cardiac hypertrophy, as observed in aging MRP4 À/À mice, which is not apparent in the same strain at 3 months (Sassi et al., 2012). Thus, in health, this mechanism may not be crucial to the maintenance of cardiovascular homeostasis in vivo, whereas its role may be triggered in cardiovascular disease characterised by enhanced circulating levels of natriuretic peptides (Potter et al., 2006).
To explore the latter hypothesis, the effect of MK571 on experimental heart failure, induced by pressure overload, was investigated.
Abdominal aortic constriction produced significant systolic dysfunction, LV hypertrophy and LV dilatation (without any overt changes in cardiac output or HR to be beneficial in models of heart failure (Frantz et al., 2013;Patrucco et al., 2014).
Exploration of MRP expression in experimental heart failure and tissue from heart failure patients revealed a reduction in MRP4 levels, but a contrasting increase in MRP5, the latter matching a previous report (Dazert et al., 2003). Further investigation will be required to determine whether MRP4 or MRP5 plays a more important in regulating cGMP signalling in health and disease and to elucidate whether these expressional variations represent a host defence mechanism to promote cGMP signalling (i.e. MRP down-regulation) or potentially a pathogenic mechanism (i.e. MRP up-regulation). Such changes might also be argued to have opposing outcomes with respect to cAMP, because this cyclic nucleotide is predominantly detrimental in heart failure. Indeed, the cyclic nucleotide transporting specificity of MRP4 and MRP5 remains controversial. MRP4 is considered a predominantly cAMP transporting protein despite the fact that MRP4 inhibition results in elevated intracellular cGMP levels (Hara et al., 2011;Sassi et al., 2008). Indeed, MRP4 has a fivefold lower K m for cGMP compared with cAMP, suggesting that it may actually have a preference for cGMP over cAMP (Chen et al., 2001). However, the V max for cAMP extrusion is approximately twice as high as cGMP, raising the possibility that at lower cAMP concentrations, MRP4 transports cGMP, but at higher cAMP concentrations, cAMP extrusion is favoured. This theory appears to be supported in practice by studies that show increasing cAMP levels decrease cGMP efflux, presumably via competition (Hamet et al., 1989;Patel et al., 1995). In fact, it is possible that this competitive interaction underpins some of the crosstalk between the cGMP and cAMP systems, which is often thought to exclusively originate from the activity of PDE2 (cGMP stimulated) and PDE3 (cGMP inhibited) (Conti & Beavo, 2007). In contrast, MRP5 is commonly accepted as a predominantly cGMP transporting protein because of its significantly lower K m (180-fold) and V max (20-fold) for cGMP compared with cAMP (Jedlitschky et al., 2000). In the vasculature, MRP4 and MRP5 are both expressed on endothelial cells and VSMCs, so in terms of localisation, both appear to be present to regulate cGMP efflux (Dazert et al., 2003;Sassi et al., 2008;Xu et al., 2004). Overall, these findings indicate that MRP4 and MRP5 are capable of regulating vascular homeostasis, with the inhibition of both proteins likely to elicit an additive, more advantageous effect. Elucidation of the precise role of each in cardiovascular homeostasis will almost certainly necessitate the use of MRP4 À/À and MRP5 À/À mice (Borst et al., 2007) (Ahmed et al., 2007;Billiar et al., 1992;Patel et al., 1995). One explanation for this difference lies with the time and concentrations of the NO-sensitive GC agonist used. Thus, longer term exposure to supramaximal NO concentrations results in excessive cGMP production that spills over in to the juxtamembrane compartment and is susceptible to MRP extrusion. However, in a more physiological setting, we conclude that this phenomenon is largely exclusive to the particulate cyclase(s), which synthesise cGMP adjacent to the membrane.
In summary, these data support that concept that MRP inhibition has ability to promote cytoprotective cGMP signalling in the heart and vasculature, and provide additional evidence verifying the importance of MRPs in contributing to cGMP compartmentalisation. Since MRP inhibitors are already in the clinic (i.e. probenecid), repurposing of such molecules may offer a rapid and inexpensive means by which to enhance cGMP-targeted agents for cardiovascular diseases such as heart failure and pulmonary hypertension. Indeed, synergy between cGMP elevating agents (e.g. PDE and neutral endopeptidase inhibitors) has been well characterised in experimental models of pulmonary hypertension and patients with the disease (Baliga et al., 2008Hobbs et al., 2019). This gives credence to the idea that combination of natriuretic peptide elevating agents (e.g. neutral endopeptidase inhibitors) with MRP inhibitors is also likely to be of supplementary therapeutic benefit. Furthermore, in the setting of pulmonary hypertension, blockade of both cGMP-and cAMP-transporting MRPs might also be argued to hold additional promise. However, the ultimate goal of utilising MRP inhibitors as drugs is fraught with difficulty. The selectivity of current inhibitors leaves much to be desired (e.g. MK571 blocks lymphotoxin signalling), MRP4/MRP5 are known to transport a number of key biological molecules (e.g. ADP, ATP, 5-hydroxytryptamine, steroids and eicosanoids) and both transporters have a widespread distribution, both within and outside the cardiovascular system (Zhou et al., 2008). Thus, despite the promise of MRP inhibition as an additional mechanism to pharmacologically augment the cardioprotective and vasoprotective actions of cGMP, far more refined and targeted interventions will be necessary to harness this effectively.

A.J.H. is a scientific advisory board member/consultant for Palatin
Technologies Inc. and Novo Nordisk and has received research support from Palatin Technologies Inc. for an unrelated project.

RIGOUR
This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research as stated in the BJP guidelines for Design & Analysis, Immunoblotting and Immunochemistry and Animal Experimentation, and as recommended by funding agencies, publishers and other organisations engaged with supporting research.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request. Some data may not be made available because of privacy or ethical restrictions.