Nitric oxide modulates cardiomyocyte pH control through a biphasic effect on sodium/hydrogen exchanger-1

Abstract Aims When activated, Na+/H+ exchanger-1 (NHE1) produces some of the largest ionic fluxes in the heart. NHE1-dependent H+ extrusion and Na+ entry strongly modulate cardiac physiology through the direct effects of pH on proteins and by influencing intracellular Ca2+ handling. To attain an appropriate level of activation, cardiac NHE1 must respond to myocyte-derived cues. Among physiologically important cues is nitric oxide (NO), which regulates a myriad of cardiac functions, but its actions on NHE1 are unclear. Methods and results NHE1 activity was measured using pH-sensitive cSNARF1 fluorescence after acid-loading adult ventricular myocytes by an ammonium prepulse solution manoeuvre. NO signalling was manipulated by knockout of its major constitutive synthase nNOS, adenoviral nNOS gene delivery, nNOS inhibition, and application of NO-donors. NHE1 flux was found to be activated by low [NO], but inhibited at high [NO]. These responses involved cGMP-dependent signalling, rather than S-nitros(yl)ation. Stronger cGMP signals, that can inhibit phosphodiesterase enzymes, allowed [cAMP] to rise, as demonstrated by a FRET-based sensor. Inferring from the actions of membrane-permeant analogues, cGMP was determined to activate NHE1, whereas cAMP was inhibitory, which explains the biphasic regulation by NO. Activation of NHE1-dependent Na+ influx by low [NO] also increased the frequency of spontaneous Ca2+ waves, whereas high [NO] suppressed these aberrant forms of Ca2+ signalling. Conclusions Physiological levels of NO stimulation increase NHE1 activity, which boosts pH control during acid-disturbances and results in Na+-driven cellular Ca2+ loading. These responses are positively inotropic but also increase the likelihood of aberrant Ca2+ signals, and hence arrhythmia. Stronger NO signals inhibit NHE1, leading to a reversal of the aforementioned effects, ostensibly as a potential cardioprotective intervention to curtail NHE1 overdrive.

cardiac physiology. Using manipulations that include nNOS knockout, adenoviral nNOS gene delivery, nNOS inhibition, and pharmacological NO titrations, we demonstrate that low levels of NO activate NHE1, whereas higher doses switch to a net inhibitory effect. These dynamic effects were not dependent upon S-nitros(yl)ation 16,37 but, instead, involved a careful balancing of activatory cGMP and inhibitory cAMP signals. We show that PKG and PKA can phosphorylate the activatory Ser703 at the C-terminus of NHE1 in vitro, but Ser648, an inhibitory residue, is more strongly phosphorylated by PKA. Finally, we demonstrate how the biphasic modulation of NHE1 by NO affects Ca 2þ handling, using the example of spontaneous Ca 2þ waves.

Adult ventricular myocyte isolation
Ventricular myocytes were isolated from hearts by enzymatic digestion and mechanical dispersion. This study used myocytes from wild-type male Sprague-Dawley rats (300-325 g), or 4-to 5-month-old nNOS-/knockout mice and their wild-type littermates, bred under license (PPL 30/3374). nNOS knockout was achieved by targeted disruption of exon 2 of nNOS. 38 Rats were sacrificed by stunning followed by cervical dislocation, and mice were sacrificed by cervical dislocation and exsanguination in accordance with UK Home Office regulations (Schedule I of A(SP)A 1986), approved by national and University ethics committees. Hearts were rapidly removed and rinsed in isolation solution (in mM): 120 NaCl, 4 KCl, 1.2 MgCl 2 , 11 glucose, 10 HEPES, 2 NaH 2 PO 4 , 2.5 pyruvate pH 7.4 at 37 C; supplemented with heparin sodium (5 units/L), and then mounted on a Langendorff perfusion system. Once cleared of blood, the heart was perfused with collagenase type II (0.9 mg/mL, Lorne Laboratories, UK) and protease type XIV (0.09 mg/mL, Sigma, UK) for 10 min, cut into pieces, triturated with a Pasteur pipette and dissociated cells filtered through a 250-mm nylon mesh.

Neonatal ventricular myocyte isolation and culture
Neonatal rat ventricular myocytes were isolated from 1-to 2-day-old Sprague-Dawley rats. Animals were euthanized by cervical dislocation and exsanguination according to Schedule 1 of the Animals (Scientific Procedures) Acts 1986. Experiments were approved by Oxford University ethical review boards and conform to the guidelines from Directive 2010/63/EU. Cells were isolated from ventricular tissue of excised hearts by enzymatic digestion and plated onto fibronectin-coated Ibidi m-slides (Ibidi, Germany) then cultured as described previously. 39

Solutions and superfusion
Solutions were delivered at 37 C to a custom-made Perspex superfusion chamber with a coverslip glass bottom (for experiments on adult myocytes) or an Ibidi m-slide (for experiments on cultured cells). Normal Tyrode contained (in mM) 135 NaCl, 4.5 KCl, 1 CaCl 2 , 1 MgCl 2 , 11 glucose, 20 Hepes at pH 7. 4. In ammonium-containing solutions, NaCl was iso-osmotically replaced with NH 4 Cl. In acetate-containing solutions, NaCl was iso-osmotically replaced with NaAcetate.

Dual microperfusion
The boundary between microstreams was visualized by including 10 mM sucrose into one of the two microstreams. Flows were adjusted to produce a sharp boundary, which was positioned across the middle of a myocyte. 45

Fluorescence imaging
To image cAMP levels, the FRET-sensor H187 was excited at 405 nm and fluorescence was measured at 480 ± 10 nm and 530 ± 10 nm in xymode on a Zeiss LSM700 confocal system. To image NO levels, DAR-4M was AM-loaded into cells (10 mM DAR-4M AM, for 10 min), and its fluorescence was excited at 514 nm and emission detected at 580 ± 20 nm on a Leica TCS NT system. 47 To image Ca 2þ waves, Fluo3 was AM-loaded into myocytes (5 mM Fluo3, for 10 min), and its fluorescence was excited at 488 nm and detected >520 nm on a Leica TCS NT confocal system in xt mode. To measure pH i , myocytes were AMloaded with cSNARF1 (10 mM, for 10 min), and fluorescence was excited at 530 nm and detected simultaneously at 580 ± 10 nm and 640 ± 10 nm on an inverted Olympus microscope with an Orca 05G CCD Camera (Hamamatsu, Japan) and Optosplit (Cairn Research, UK).

Expression of NHE1 C-terminus (His182)
The C-terminus of NHE1 was expressed in a bacterial (BL21-SI) system, as described previously. 48 Bacteria were cultured in LBON media (bacto tryptone 10 g/L, bacto yeast extract 5 g/L, AMP 100 mg/L) at 30 C. After reaching an optical density of 0.6 (600 nm), expression was induced by adding 300 mM NaCl and incubation for 4 h. His182 was purified as described previously. 48 See Supplementary material online for Kinase reactions, Western blotting, Phos-tag gel analysis, S-nitros(yl)ation (iodoTMT) assay, organo-mercury enrichment, antibodies, and mass spectrometry.

Statistics
Differences tested by two-way ANOVA at 5% significance, followed by post hoc test (Fisher's LSD). Data are reported as means ± SEM, and repeats as 'number of cells/number of animals'. Flux data are normally distributed continuous variables. For non-normally distributed data comparing two groups, statistical analysis was done using a Mann-Whitney U-test. 3. Results 3.1 PKA and PKG produce distinct patterns of NHE1 phosphorylation NO signalling in the myocyte can activate PKG via the canonical pathway involving cGMP 17,27 and, under certain conditions, PKA through a rise in [cAMP]. [28][29][30] To study how NHE1 responds biochemically to these kinases, the C-terminus (wherein most modulatory influences converge) of the human protein was expressed in a bacterial system. 48 This fragment, thenceforth referred to as His182, was reacted in vitro with PKA or PKG, and phosphorylation was analysed by Phos-tag gel blotting ( Figure 1A). Both PKA and PKG were able to phosphorylate His182, but the mobility shifts were distinct, indicating that the kinases act differentially on NHE1. Two important residues linked directly to regulating NHE1 transport activity, 31,36 and predicted substrates for PKA and PKG, 14 are Ser648 and Ser703. To test if these residues are phosphorylated, kinase reactions were subject to analysis by mass spectrometry ( Figure 1B). This revealed that His182 was phosphorylated on multiple serine residues, including Ser648 and Ser703. Consistent with the Phos-tag analysis, mass spectrometry revealed a differential pattern of His182 phosphorylation by PKA and PKG ( Figure 1C). Western blotting using antibodies raised against phosphorylated serine confirmed phosphorylation of His182 by both PKA and PKG, albeit with distinct bands ( Figure 1D).
To investigate how the phosphorylation status of Ser648 and Ser703 responds to PKA and PKG, antibodies recognizing specific phosphoserine motifs were used for western blotting (see Supplementary material online, Figure S1 for loading controls). An antibody raised against the 14-3-3 binding motif detected Ser703 phosphorylation by PKA and PKG, albeit with a distinct immunoreactivity pattern ( Figure 1E). In contrast, an antibody raised against the Akt phosphorylation consensus sequence, i.e. Ser648 on NHE1, showed substantially greater immunoreactivity towards PKA-treated His182, and only weak signal with PKG ( Figure 1F). These findings indicate that both PKA and PKG can phosphorylate Ser703, which underpins NHE1 activation, 36 but that PKA is considerably more selective for Ser648, a residue responsible for NHE1 inhibition. 31 To investigate how these kinases are recruited physiologically by NO to regulate cardiac NHE1, the next experiments measured transport activity in ventricular myocytes subjected to pharmacological or genetic interventions that target PKG-and PKA-mediated cascades.

Basal NO production activates cardiac NHE1
nNOS knockout mice were used to investigate the effect of basal NO production on pH i control by NHE1 (see Supplementary material online, Figure S2 for confirmation of nNOS gene ablation). Resting pH i was measured in enzymatically isolated cardiomyocytes loaded with the pH reporter dye, cSNARF1, and superfused in physiological CO 2 /HCO - buffered solution. The frequency distribution of resting [H þ ] i in nNOS-/myocytes was, compared to cells from wild-type littermates, broader and extended into the acidic range. Thus, myocytes lacking basal NO production are less able to defend a favourable (mildly alkaline) pH i ( Figure 2A). Intrinsic buffering capacity, attributable largely to proteins, was no different between the two experimental groups (Supplementary material online, Figure S3; measured by a step-wise NH 4 Cl removal method 49 ) To measure NHE1 activity, the transporter was activated by lowering pH i , attained by means of an ammonium prepulse solution manoeuvre. 1 For these experiments, myocytes were superfused with Hepes-buffered solution to inactivate HCO -3 -dependent transport. Under these conditions, recovery from an acidic pH i is mediated primarily by NHE1, as demonstrated by its sensitivity to dimethylamiloride (DMA; 30 mM), an NHE1 inhibitor ( Figure 2B). Given that NHE1 is exquisitely sensitive to [H þ ], data were presented as pH i -flux curves, and comparisons are made at matching levels of pH i . NHE1 flux, calculated as the product of buffering capacity and the rate of pH i change during the recovery phase, was slower in nNOS-/-myocytes compared to wildtype cells ( Figure 2C), demonstrating that basal NO production has an activatory effect on NHE1. There was no effect of the nNOS knockout on the DMA-insensitive component of flux.
Genetic ablation of nNOS may have exerted its activatory effect through a reversible post-translational effect driven by NO, or through a slower up-regulation of NHE1 protein. These two possible mechanisms were tested by measuring the acute response of NHE1 flux to nNOS inhibition with the compound AAAN. 50 Wild-type rat myocytes treated with 5 mM AAAN had a left-shifted NHE1-pH i curve ( Figure 2D), indicating that basal nNOS activity is exerting a reversible, post-translational effect on NHE1.
Further testing of the effect of nNOS activity on NHE1 used rat myocytes infected with an adenovirus encoding for an nNOS-eGFP fusion protein. Cells incubated overnight with virus produced a good yield of construct expression, as determined by eGFP fluorescence ( Figure 2E, inset). Compared to cells incubated overnight with virus containing only the eGFP gene (i.e. sham infection), adenoviral nNOS gene delivery hastened pH i recovery to a more alkaline resting pH i , and increased NHE1 flux ( Figure 2F). Overall, these findings demonstrate that basal nNOSderived NO exerts an activatory influence on NHE1.

Exogenously delivered NO exerts a dose-dependent effect on NHE1
The actions of NO on NHE1 were further studied using an exogenous source of NO delivered by superfusion. SNP has widely been used as a short half-life NO donor, producing substantial NO release peaking near 600 mM within tens of minutes of addition to superfusates (Supplementary material online, Figure S4). Recovery from an acid-load (Hepes-buffered superfusates) was first measured under control conditions, and repeated in the presence of SNP (1 mM). Exposing the myocytes to a high concentration of NO resulted in a slowing of pH i recovery and a more acidic resting pH i ( Figure 3A). This attenuation of NHE1 flux was not due to an NO-independent, time-dependent attrition, because a consecutive pair of control measurements produced a similar recovery (Supplementary material online, Figure S5). The inhibitory effect of SNP was absent in cells pre-treated with 1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one (ODQ; 6 mM), an irreversible inhibitor of GC, indicating the involvement of cGMP signals ( Figure 3B). These data suggest that a high concentration of NO released from SNP inhibits NHE1 flux, in apparent contradiction to the activatory effect of basal NO production by nNOS described earlier (see Figure 2). SNP (1 mM) had no effect on Na þ -HCO 3 cotransport (NBC), the other major acid-extruder in myocytes, measured in CO 2 /HCO 3 buffered superfusates containing DMA to block NHE1 ( Figure 3C).
The actions of NO on NHE1 were investigated further using NOC12, a donor with longer half-life, thus releasing a steadier flow of NO. Kinetic modelling using published stability constants indicates that myocytes superfused with 5 mM NOC12 are exposed to >300 nM NO (Supplementary material online, Figure S4). At this concentration, NO exerted an inhibitory effect on the rate of pH i recovery and acidified steady-state pH i ( Figure 3D), akin to the effect of SNP. The inhibitory effect of 5 mM NOC12, which could be described in terms of a left-shift in Western blots were performed using an antibody raised against phosphorylated serine residues. (E) Western blot performed using antibody against the phosphorylated 14-3-3 binding motif (detecting Ser703 in NHE1). (F) Western blot performed using antibody against phosphorylated Akt substrate (Ser648 in NHE1). Note: blots for PKA-and PKGreacted His182 were prepared with the same amount of substrate and developed simultaneously with identical exposure time. See Supplementary material online, Figure S1 for loading controls.
the NHE1-pH i curve by $0.1 units, was absent in myocytes pretreated with ODQ, confirming that the exogenously released NO was signalling via cGMP ( Figure 3E). Although many previous studies have used micromolar doses of donors to interrogate NO signalling, the concentration of NO attained may be considered supra-physiological. 26 To measure the effects of lower NO concentrations, NOC12 was combined with a scavenger, CPTIO, to titrate [NO] down to the low-nM range. 44 A kinetic model determined that combining 300 mM NOC12 with 100 mM CPTIO (plus 300 mM urate to scavenge the by-product NO 2 ) could maintain [NO]<30 nM 27,44 (Supplementary material online, Figure S4). At this lower [NO], pH i recovery was faster and progressed to a more alkaline resting pH i ( Figure 3F) indicating NHE1 activation ( Figure 3G),  Figure S2). NHE1-dependent fluxes were slower in nNOS knockout mice (P < 0.0001, two-way ANOVA). (D) NHE1 flux measured in wild-type rat ventricular myocytes. Hepes-buffered superfusates, 37 C. Treatment with selective nNOS inhibitor AAAN (5 mM; N = 16 from five rats) reduced activity relative to control (N = 18 from five rats) (P < 0.0001, two-way ANOVA). (E) Ammonium prepulse performed on adenovirally infected rat ventricular myocytes cultured overnight. Hepes-buffered superfusates, 37 C. Infected cells were identified by eGFP fluorescence (inset). (F) NHE1 activity was accelerated (P < 0.0001, two-way ANOVA) in myocytes infected with nNOS gene (N = 16 cells from four independent infections) compared to sham infected cells (N = 28 from five independent infection).
which is now consistent with the effect of nNOS-derived NO. This effect, observed as a right-shift in the NHE1-pH i curve, was absent in myocytes pretreated with ODQ ( Figure 3G). Thus, high and low concentrations of NO produce opposite effects on NHE1 activity, yet both involve cGMP signals.

Biphasic modulation of NHE1 by NO is a special property of the adult cardiomyocyte
To determine if the biphasic modulation of NHE1 by NO is a more general property of NHE1-expressing cells, measurements were performed on cultured neonatal rat ventricular myocytes and on a human breast cancer cell line MDA-MB-468, both of which produce NHE1 fluxes that are comparable in magnitude to those in adult ventricular myocytes. However, the biphasic effect of NO on NHE1 was not observed in neonatal myocytes (Supplementary material online, Figure S6A and B) or MDA-MB-468 cells (Supplementary material online, Figure S6C and D). These findings argue that NO signals modulate NHE1 in a highly context-sensitive manner, involving elements that are present in adult cardiac myocytes, but not necessarily in other cell types. Consequently, expression systems and in vitro-based assays are not appropriate for interrogating the mechanisms of cardiac NO-NHE1 interplay; instead, the relevant cascades must be studied in primary adult myocytes.

NO signals can be transmitted onto remote NHE1 targets and do not involve S-nitros(yl)ation
One possible mechanism through which NO may be regulating NHE1 is S-nitrosylation or S-nitrosation. Indeed, many protein targets have been proposed to undergo this post-translational modification, 16,18,37 although the physiological significance of some of these observations has been questioned. 51 To test whether sarcolemmal NHE1 is substrate for direct NO reactions, immunoblotting using the so-called TMT-method 52 was performed on membrane fractions of cardiac tissue treated with either high or low [NO], delivered by Langendorffperfusion with 5 mM NOC12 or 300 mM NOC12 plus 100 mM CPTIO, respectively (Supplementary material online, Figure S7A). There was no evidence of any increase in nitros(yl)ation at the band corresponding to NHE1 in response to low or high [NO] ( Figure 4A), arguing against S-nitros(yl)ation. As a positive control, an increase in nitros(yl)ation was observed in whole-cell lysates subject to treatment with Snitrosoglutathione (Supplementary material online, Figure S7B). The lack of NHE1 S-nitros(yl)ation by high levels of [NO] was confirmed using the complementary technique of organo-mercury resin-assisted capture that is considered a more specific test for S-nitros(yl)ated residues (SNO) 53 ( Figure 4B). Blots were quantified as the ratio of SNO signal to input (NHE1), and this analysis revealed no significant difference between control and NOC12-treated hearts ( Figure 4C). As confirmation of the method's resolving power to detect changes in protein nitros(yl)ation, lysates pretreated with DTT or GSNO produced measurable changes on blots (Supplementary material online, Figure S8).
The lack of evidence for a direct S-nitros(yl)ation of NHE1 suggests the involvement of cGMP signals and is consistent with the ODQsensitivity of NO actions on NHE1 ( Figure 3E and G). If the diffusible messenger cGMP were responsible for the coupling between NO and NHE1, then a spatially-confined source of NO would be expected to influence the activity of more remote NHE1 targets. However, with increasing distance from NO-activated GC, the concentration of cGMP will decay as a result of degradation by phosphodiesterases (PDEs). Since NO exerts a biphasic effect on NHE1, spatially-decaying [cGMP] may flip from being an inhibitory signal to becoming activatory. To test for remote signalling by NO, a dual microperfusion system was used to deliver two sharply separated microstreams at a perpendicular angle to a myocyte. One microstream contained NOC12 to release NO, whilst the other contained CPTIO to scavenge any NO spill-over ( Figure 4D). The fluorescent dye DAR-4M reported a rise in [NO] that was spatially restricted to the NOC12exposed half of the cell ( Figure 4E). Thus, using dual microperfusion, it is possible to deliver NO to one side of a myocyte, and observe any remote actions on NHE1 at the opposite end.
To trigger NHE1 activity, myocytes were first acid-loaded, by means of ammonium prepulse, and then dually microperfused. One half of the myocyte was exposed to NOC12 at a dose (5 mM) that would inhibit NHE1, if applied uniformly to a cell ( Figure 3E). To investigate how this localized source of NO influences NHE1 remotely, the NOC12containing microstream also included DMA to inactivate NHE1 locally; consequently, any pH i recovery produced by the cell must be attributable to NHE1 activity at the other half of the myocyte. Measurements of NHE1 flux showed a stimulatory effect of NOC12 when the donor was delivered at a distance from NHE1 ( Figure 4F). This remote NO-NHE1 coupling was absent in cells pre-treated with ODQ, indicating the involvement of diffusible cGMP signals ( Figure 4G)

Crosstalk between cGMP and cAMP produces biphasic modulation of NHE1 by NO
Based on the results of in vitro studies using His182 (Figure 1), the pattern of NHE1 phosphorylation produced by PKG is distinct to that evoked by PKA. Thus, it is possible to explain biphasic NO regulation of cardiac NHE1 by dose-dependent recruitment of kinases. PKG was found to be more potent at phosphorylating Ser703 (linked to activation; Figure 1E) than Ser648 (linked to inhibition; Figure 1F), in comparison to PKA which acted strongly on Ser648. 54 Notwithstanding this in vitro evidence, it is not intuitive to explain how a NO signal, acting through its canonical second messenger cGMP, could orchestrate NHE1 activation at low concentrations, but inhibition with stronger stimulations. A plausible mechanism by which a cGMP signal could switch its polarity of action may involve the recruitment of cAMP beyond a threshold [cGMP]. Indeed, previous studies have shown that cGMP-evoking stimuli can produce a secondary rise in [cAMP], arising from the inhibitory effect of cGMP on the cAMP-degradating enzyme phosphodiesterase-3 (PDE3). 55 To determine if the cGMP signal evoked with 5 mM NOC12 is sufficient to produce a detectable rise in [cAMP], myocytes were first infected with virus containing the gene for H187, a FRET-based cAMP probe. Superfusion with 5 mM NOC12 produced a measurable rise in [cAMP], reported as a response that was equivalent to the effect of lowdose ($1 nM) isoprenaline (Iso), a beta-agonist ( Figure 5A). Thus, a sufficiently large NO signal can evoke a cAMP response in adult myocytes ( Figure 5B) The effects of cAMP on NHE1 were tested in wild-type rat myocytes stimulated with either 0.5 nM Iso or 20 mM 8Br-cAMP, a membrane-permeable cAMP derivative. Engaging cAMP-dependent signalling with these agents produced an acute inhibition of NHE1 ( Figure 5C and D), consistent with previous studies showing a PKA-dependent phosphorylation of the inhibitory site at Ser648. 3 Thus, a cAMP signal emerging from a NO-evoked rise in [cGMP] may explain the net inhibitory effect of high [NO] on NHE1. To test this, myocytes were exposed to 5 mM NOC12 to activate cGMP signalling, but one group of cells was pretreated with 50 mM dideoxyadenosine (ddA), a  membrane-permeable inhibitor of adenylyl cyclase, which is expected to prevent the secondary [cAMP] rise. NHE1 activity was significantly higher in myocytes that were blocked from evoking a cAMP signal, representing the state of NHE1 modulation in response to cGMP alone ( Figure 5E).
The coupling between NO-NHE1 was further interrogated by introducing membrane-permeable cGMP derivatives in the absence of an NO donor. The analogue 8Br-cGMP activates canonical cGMP targets, such as PKG, and also inhibits PDE3. In contrast, 8pCPTcGMP, unlike endogenous cGMP, is not broken down by PDEs and also does not inhibit PDE3. 56 When applied to myocytes, 20 mM 8Br-cGMP and 10 mM 8pCPTcGMP produced opposite effects on NHE1 flux ( Figure 5F). The activatory effect of 8pCPTcGMP, which is unable to evoke secondary cAMP signals, was consistent with the actions of low [NO] ( Figure 3G), whereas the effect of 8Br-cGMP is more akin to that of high [NO] ( Figure 3E). Thus, the biphasic effect of NO on NHE1 is a result of a [NO]-dependent rise in cGMP, which per se is activatory on NHE1, but switches to becoming inhibitory when a sufficient degree of PDE3 inhibition allows [cAMP] to build-up. This mechanism exploits the distinct patterns of NHE1 phosphorylation produced by PKG and PKA (Figure 1).

NO control of NHE1 activity produces a biphasic effect on Ca 2þ wave frequency
The triggering of NHE1 activity at low pH i evokes a substantial influx of Na þ ions, in the range of several mM/min, which can Ca 2þ -overload the cardiomyocyte through a well-documented mechanism involving NCX ( Figure 6A). Beyond a threshold of Ca 2þ loading, myocytes will produce spontaneous Ca 2þ release from the sarcoplasmic reticulum (SR), detected as cytoplasmic Ca 2þ waves. Modulation of Na þ -influx via NHE1 by NO signals is therefore expected to influence the frequency of spontaneous Ca 2þ waves.
To trigger NHE1 activity, the cytoplasm of myocytes was uniformly acidified by exposure to 30 mM acetate, which enters and acidifies the cytoplasm and triggers the activation of NHE1 ( Figure 6B). As NHE1 attempts to extrude the acid-load, Na þ ions enter the myocyte ( Figure 6C). Ca 2þ release events were recorded in Fluo3-loaded myocytes, superfused with solutions containing 5 mM Ca 2þ (to raise the likelihood of spontaneous release). The frequency of Ca 2þ waves was calculated from the interval between events detected by imaging in linescan mode ( Figure 6D). In the absence of NO donors, Ca 2þ wave frequency increased by a factor of two within 5 minutes of acidification ( Figure 6D). The protocol was repeated in the presence of either low [NO] (300 mM NOC12 plus 100 mM CPTIO) to stimulate NHE1 or high [NO] (5 mM NOC12) to inhibit NHE1. It is well-established that NO can affect Ca 2þ waves through various NHE1-independent mechanisms (e.g. on RyR2 channels), 37,57,58 and to correct for these actions, the response of Ca 2þ wave frequency to acetate was normalized to baseline frequency. In the presence of high [NO], the acetate-evoked rise in Ca 2þ wave frequency was attenuated (to only $150% of resting levels), consistent with an inhibitory effect on NHE1-dependent Na þ -influx ( Figure 6D). Thus, inhibition of NHE1 with high doses of NO is antiarrhythmogenic, similar to the pharmacological effect of NHE1 inhibitors such as cariporide. 59 In contrast, low [NO] resulted in a more robust acetate-evoked rise in Ca 2þ wave frequency (to $300% of resting levels), explained by the larger Na þ influx through activated NHE1 ( Figure 6D).

Discussion
We provide the first report of a physiologically relevant molecule exercising a biphasic effect on cardiac NHE1, the heart's most powerful pH i regulator and major Na þ entry pathway. NHE1 activity has a broad remit of actions on cardiac physiology because its activity influences pH i homeostasis and Ca 2þ signalling, and thence a myriad of downstream effects. Although resting pH i in myocytes is normally kept near 7.2, deviations can occur pathologically, such as during ischaemia, or physiologically in response to intra-or extracellular signals. 54,[60][61][62] Controlled shifts in pH i are implemented physiologically by changing transmembrane H þ ion traffic carried by transporters, such as NHE1. Unlike other regulators of NHE1, NO does not operate through a surface receptor, and thus the coupling to its downstream enzyme (GC) has less 'gain' than in the case of G protein-coupled cascades. Instead, the physiological outcomes of NO signalling are fine-tuned by regulating the pattern of gas release. Other examples of dose-dependent actions of NO on the heart have been reported for chronotropy, 63,64 but these are the ensemble of effects on various proteins, and not necessarily a convergence onto a single identifiable target.
The regulatory scope of NO signalling on NHE1 can be quantified from the dynamic range of its effect, calculated as the ratio of the highest to the lowest activation state. Based on results such as those in Figure 3, the dynamic range for the NO effect on NHE1 is a factor of two, offering good leverage on cardiac physiology. At the lower end of this activity range, NHE1 is still able to maintain a low intracellular [H þ ] that is conducive for cellular physiology, but this reduction in homeostatic prowess would permit some degree of pH i fluctuations, serving as bona fide H þsignals within the so-called permissive range. 1 When maximally activated, cardiac NHE1 can produce an unprecedented magnitude of H þ extrusion to defend against acid-challenges and maintain an alkaline pH i .
Intriguingly, the biphasic effect of NO on NHE1 was not observed in neonatal rat ventricular myocytes or in a breast cancer cell line (Supplementary material online, Figure S6). This finding indicates that the biochemical details of the NO-NHE1 cascade cannot be resolved in expression systems (e.g. with the use of NHE1 mutants) or with cell-free approaches. Here, we were able to obtain mechanistic insights by studying the responses of primary adult myocytes to pharmacological or genetic manipulations of the putative NO-NHE1 cascade (Figures 2-6), and interpreting these findings in terms of how PKA and PKG differentially phosphorylate the C-terminus of NHE1, as determined using in vitro assays ( Figure 1).
The modulation of NHE1 activity by NO is not exercised by S-nitros(yl)ation, tested biochemically by two independent methods ( Figure 4A-C), but through the enzyme GC instead (Figures 3 and 4) Canonically, this pathway signals through the diffusible messenger cGMP, but in the heart, adequate cGMP-inhibition of PDE3 can allow a secondary rise in [cAMP], which we confirmed with a FRET-based sensor of cAMP ( Figure 5A). Results of in vitro studies ( Figure 1A-C) show that serine residues of the regulatory C-terminus of NHE1 are substrates for cAMPand cGMP-dependent kinases. Using antibodies against specific phosphorylation sites, we show that PKA and PKG have distinct selectivity for serine residues. Specifically, PKG showed a preference for phosphorylating Ser703, a residue linked to NHE1 activation. PKA, in contrast, was able to produce a strong phosphorylation of Ser648. Our functional data explain how NO signals are able to use these kinases to orchestrate NHE1 activation and inhibition at low and high [NO], respectively, by recruiting cGMP in a dose-dependent manner, and triggering cAMP only . . . . . . . . . . . with the higher concentration ( Figure 7). Using pharmacological inhibition and knockdown, we show that basal NO production by nNOS was activatory on NHE1. Adenoviral delivery of nNOS, which preferentially targets the sarcolemma, 65 or a carefully titrated low (<30 nM) dose of NO (NOC12þCPTIO) produced a further stimulatory effect ( Figure  7A). 44 In the latter experiments, NO concentrations attained inside myocytes are likely to be attenuated further by chemical reactions involving proteins such as myoglobin ($200mM myoglobin will scavenge NO with a rate constant of 22 mM/s 66 ). We therefore argue that the low [NO] concentration delivered by mixing NO donor and scavenger is within the physiological range. 26 NHE1 activation by this level of NO signalling could be explained by PKG phosphorylation of Ser703, without affecting Ser648 ( Figure 1E). Exposure to higher [NO] (>300 nM) released from donors (SNP or NOC12 alone) switched the polarity of NO signalling to net inhibition because of the secondary recruitment of cAMP signalling (Figure 3). Activation of PKA is expected to increase Ser648 phosphorylation, an inhibitory modification ( Figure 1F). Indeed, a previous report showing NHE1 inhibition by NO released from a high dose of SNP 21 is likely to have involved cAMP signals ( Figure 5B). Thus, the crossover in the polarity of the NO-NHE1 cascade occurred around 100 nM, postulated to be above the normal physiological range of NO signalling. 26 As a test of our model, a cGMP-analogue that cannot inhibit PDE3 (and hence has no effect on [cAMP]) was found to activate NHE1 ( Figure 5F), whereas stimuli that evoked cAMP signals but without a change in cGMP, caused NHE1 inhibition ( Figure 5C and D). Furthermore, it was possible to reverse the effect of a nominally inhibitory NO signal (i.e. domain of high [NO]) by imposing a diffusion distance between the source of NO and NHE1, along which the cGMP signal decays and becomes inadequate to raise [cAMP] (Figure 4).
Given that the actions of NO on NHE1 are highly contextdependent, as shown by their manifestation in adult but not neonatal myocytes (Figure 3, Supplementary material online, Figure S6), it is plausible that pathologic remodelling of myocytes may alter NO-NHE1 coupling further. Patterns of myocardial NO signalling can undergo changes in heart failure, 67-69 hypertension, 70 and ischaemia/reperfusion injury 68,71,72 as a result of altered NOS isoform expression 67,68,70,71 and subcellular distribution 67,68 or abnormal intracellular chemistry (e.g. NO scavengers 73 and NOS co-factors 19 ) In failing hearts, for instance, raised nNOS activity at the sarcolemma 67,68 may produce a more intensive release of NO near NHE1, potentially resulting in inhibition.
The tightening of pH i control by NO-activated NHE1 would improve the responsiveness of the myocyte's pH i -regulatory apparatus to acidic disturbances, such as in periods of increased metabolic demand or failure of perfusion. However, the benefit of this augmentation must be weighed against the additional energetic burden on the Na þ /K þ ATP pump required to restore the inward transmembrane [Na þ ] gradient. The

NHE1-driven increase in intracellular [
Na þ ] will also affect Ca 2þ signalling by increasing SR Ca 2þ loading. At best, this would exert a positive inotropic effect, but more sustained overloading will eventually precipitate an arrhythmia as a result of spontaneous Ca 2þ release. Considering the injurious nature of the latter phenomenon, there is a biological niche for agents that protect against NHE1 over-drive. High volumes of NO release would produce such an effect, as shown by the inhibitory effect on Ca 2þ wave frequency ( Figure 6D). We speculate that basal-to-low levels of NO serve to increase heart contractility by stabilizing pH i at a more alkaline level, and by increasing cellular Ca 2þ loading through increased Na þ entry. However, excessive activation of NHE1 can be acutely deleterious to the heart, as demonstrated by ischaemia-reperfusion injury 74 when cells become Ca 2þ overloaded 11 and arrhythmic. 75 Chronically overactive NHE1 has been linked to maladaptive hypertrophy, a risk factor in heart failure. 76,77 Thus, enhanced NO signalling may serve as a natural NHE1 inhibitor, producing more quiescent Ca 2þ signalling by curtailing Na þ influx and allowing the cell to undergo a mild acidification.
Modulation of NHE1 by NO signalling may be relevant to the interpretation of clinical trials, such as the GUARDIAN trial (cariporide) 78 because the benefit of targeting NHE1 will depend on the extent to which the protein is functionally active. Patients and animal models with HFpEF (heart failure with preserved ejection fraction) have greatly increased nitrosative stress 69 which may be equivalent to levels of NO that inhibit NHE1. In these states, pharmacological inhibition of NHE1 would yield a reduced therapeutic effect, and drugs such as cariporide may not be as effective as anticipated. Compared to HFpEF, myocardial ischaemia is postulated 72 to involve more modest increases in NO levels. If such a nitrosative stress activates NHE1, a higher dose of cariporide would be required to adequately suppress Na þ influx. This may explain why the sole beneficial outcomes noted in the GUARDIAN trial were for the highest doses (120 mg) of cariporide. 78 However, without information about the state of NO signalling, it is not possible to perform a meta-analysis of patients stratified by endogenous NHE1 activity.
In summary, the biphasic effect of NO provides unprecedented flexibility in controlling NHE1 locally in the heart, and with it, the capacity to modulate cardiac Ca 2þ signalling and other pH-sensitive processes.

Supplementary material
Supplementary material is available at Cardiovascular Research online. The various experimental manoeuvres performed in this study are ranked by strength of NO signal. Higher NO stimulations will evoke greater cGMP production. At higher [cGMP], inhibition of PDE3 permits a steady rise in [cAMP]. The balance between cGMP and cAMP signals sets the level of NHE1 activity. Based on literature values for [NO] measured in vivo, NO-evoked NHE1 activation is expected to be within physiological signalling, whereas NO-evoked NHE1 inhibition would occur over the pathophysiological range of elevated [NO]. (B) Schematic of the signalling pathways, evoked by NO, that regulate NHE1 activity by means of a differential activation of PKG-and PKA-dependent cascades.