Oxidation of Protein Kinase A Regulatory Subunit PKARIα Protects Against Myocardial Ischemia-Reperfusion Injury by Inhibiting Lysosomal-Triggered Calcium Release

Supplemental Digital Content is available in the text.

O xidative stress plays a pivotal role in the pathogenesis of ischemia-reperfusion (I/R) injury, with early bursts of reactive oxygen species (ROS) initiating a cascade of deleterious cellular processes that promote cell death and cardiac dysfunction. 1,2 Paradoxically, prevention of ROS generation by inhibiting specific oxidase systems exacerbates I/R injury, 3,4 suggesting that some degree of ROS formation is necessary for cardioprotection. 2,5 Evidence that ROS underpin the effects of preconditioning or some cardioprotective compounds [6][7][8] supports this conclusion, as does the general failure of antioxidants to reduce reperfusion injury after coronary angioplasty 9 or improve clinical outcomes in patients with acute myocardial infarction or heart failure. 10 Although it is known that ROS signaling is mediated largely through covalent modification of specific cysteine thiols within redox-sensitive proteins, 11 the exact mechanisms through which they exert their cardioprotective actions remain unclear.
Protein kinase A (PKA) is 1 of the master regulatory molecules in the heart. Under physiological conditions, PKA contributes to the cardiac response to catecholamine stimulation through catalyzed phosphorylation of proteins involved in excitation-contraction coupling, metabolism, and cardiomyocyte hypertrophy. 12,13 In disease states, however, persistent activation of PKA signaling, or altered expression of PKA isotypes, has been linked to maladaptive remodeling, pathological hypertrophy, and the progression to heart failure, 14 making pharmacological targeting of PKA an attractive therapy for the treatment of cardiac disease.
The ability for PKA to regulate a multitude of cellular processes occurs through differential expression and localization of 2 distinct isotypes (type-1 and type-2) composed of 2 catalytic (PKA cat ) and 2 regulatory subunits (RIα or RIβ and RII, respectively). 15 Although all PKA isotypes depend on cAMP binding for activation, recent work has shown that PKARIα (type-1 PKA) possesses 2 cysteine residues within the RIα subunits that are sensitive to ROS-mediated oxidation. 16,17 Studies in isolated hearts and cardiomyocytes, using exogenous oxidants, have shown that oxidation of these cysteines leads to formation of an interprotein disulfide bond within the RIα subunit, 17 which may enhance the holoenzyme's catalytic activity, independent of cAMP, or promote PKARIα subcellular targeting. 16,18 Beyond this, however, little else is known about the endogenous triggers of PKARIα disulfide formation in the myocardium or how PKARIα oxidation affects cardiac function.
Here, we provide the first evidence for endogenous induction of PKARIα disulfide formation in the heart, occurring after I/R in both humans and mice. Using high spatial and temporal resolution imaging modalities, in conjunction with a "redox dead" PKARIα knock-in (KI) mouse model, 19 we demonstrate that disulfide modification targets PKARIα to the lysosome, where it acts as a gatekeeper for two-pore channel (TPC)-mediated Ca 2+ release and prevents inappropriate triggering of Ca 2+ release from the sarcoplasmic reticulum (SR). In the postischemic heart, we find that inhibition of lysosomal Ca 2+ release by oxidized PKARIα is crucial for limiting infarct size and preserving cardiac function during reperfusion, offering a novel target for the design of cardioprotective therapeutics.

METHODS
Supporting data and methods can be found in Methods in the Data Supplement and will be made available, on reasonable request, by contacting the corresponding author.

Human Samples
Biopsies of the right atrial appendage were obtained before and after cardiopulmonary bypass and reperfusion in patients undergoing on-pump coronary artery bypass surgery at the John Radcliffe Hospital (Oxford, United Kingdom). The study was approved by the Research Ethics Committee (reference no. 07/Q1607/38), and all patients gave written, informed consent.

Animals
"Redox dead" PKARIα KI mice (C57BL/6 background), in which the nucleotides encoding for cysteine at position 17 were mutated to nucleotides encoding for serine (Cys17Ser), were generated as previously described. 19 Only male mice

Clinical Perspective
What Is New? • We offer the first evidence that ischemia/reperfusion injury, in humans and in mice, induces PKARIα (regulatory subunit Iα-containing protein kinase A) oxidation and disulfide formation. • Disulfide formation enhances PKARIα intracellular anchoring and promotes compartmentation of the holoenzyme complex to the lysosome, where it acts as a negative regulator of two-pore channeldependent calcium release. • Using genetic loss of PKARIα disulfide formation, we demonstrate that this newly identified regulatory mechanism serves as a crucial, adaptive response to myocardial ischemia/reperfusion injury by inhibiting excess calcium release and limiting infarct size.

What Are the Clinical Implications?
• Inhibition of lysosomal two-pore channel-dependent calcium release by oxidized PKARIα prevents myocardial cell death in response to ischemia/ reperfusion, revealing a previously unrecognized mechanism of cardioprotection that could be exploited for therapeutic intervention.

Statistical Analysis
All experimentation and data analysis, apart from immunoblots, were conducted blinded to genotype and intervention. Data were checked for normality of distribution before statistical analysis using a Shapiro-Wilk normality test. Comparisons between data were performed using either a Student t test or ANOVA with Bonferroni correction (normally distributed) or using the Mann-Whitney test or Kruskal-Wallis test (nonnormally distributed). For Ca 2+ handling data in cardiomyocytes, analyses were carried out in RStudio using a hierarchical statistical method, 20 taking into consideration clustering of single cells per animal and correcting for this in the statistical analysis. The incidence of spontaneous Ca 2+ release events was compared using a Fisher exact test. A P value <0.05 was considered statistically significant.

Myocardial I/R Promotes PKARIα Disulfide Formation
Although PKARIα disulfide formation is known to occur in the heart in response to exogenous oxidant treatment, [16][17][18] no evidence for endogenous induction of PKARIα oxidation and disulfide formation has been reported. We therefore aimed to determine whether, in humans and mice, disease states associated with increased ROS production would promote PKARIα disulfide bond formation. Atrial tissue biopsies taken from patients undergoing on-pump cardiac surgery showed a minimal degree of PKARIα disulfide formation before cardiopulmonary bypass; however, in samples acquired from the same patient minutes after cardioplegia and reperfusion, the PKARIα disulfide state was found to be significantly increased ( Figure 1A and 1B). Left ventricular (LV) tissue obtained from mice undergoing transient coronary artery ligation in vivo also displayed markedly enhanced PKARIα disulfide formation compared with sham-operated mice, with increased PKARIα oxidation seen in both the I/R and the remote region of the LV ( Figure 1C and 1D). Control experiments in human and mouse tissue showed a reduction of high molecular weight bands when samples were treated with reducing agents ( Figure IA and IB in the Data Supplement), confirming that the higher molecular weight bands were disulfide dimerized PKARIα. In mice, a second band just above the putative RIα monomer was also found after reduction. Further experiments revealed this to be a nonspecific band no longer present when PKARIα was purified using cAMP-affinity capture ( Figure IC in the Data Supplement). Unlike oxidative modifications that lead to protein degradation, 21 PKARIα oxidation was not associated with loss of total PKARIα protein levels (as observed in Figure IA and IB in the Data Supplement), suggesting that this modification is regulatory in nature and likely has a functional role during I/R-injury.

RIα Disulfide Formation Enhances PKA Intracellular Anchoring Through A-Kinase Anchoring Protein Binding Without Affecting Catalytic Activity
For several kinases, regulatory oxidation of cysteine thiols increases their catalytic activity. 21 To test whether this was the case for PKARIα, we used real-time monitoring of PKA catalytic activity by the genetically encoded AKAR3ev fluorescence resonance energy transfer (FRET) biosensor, which we expressed in cultured adult LV cardiomyocytes isolated from PKARIα "redox dead" KI mice or their WT littermates. Before use of the KI mouse model for mechanistic studies, detailed cardiac characterization was undertaken to rule out gross structural cardiac remodeling ( Figure  As demonstrated in Figure 2A, genetic substitution of a serine for 1 of the critical, disulfide-forming cysteines in the RIα subunit (Cys17Ser) 19 prevented KI mice from forming PKARIα disulfide bonds, either under basal conditions or in response to H 2 O 2 treatment. By contrast, freshly isolated WT cardiomyocytes showed a significant proportion of RIα in the disulfide state under basal conditions (52.6±3.6%), which was further increased by treatment with H 2 O 2 (83.0±2.1%). Despite this marked difference in PKARIα disulfide state between WT and KI cardiomyocytes, we found no change in the normalized FRET ratio over the course of the 8-minute H 2 O 2 incubation ( Figure 2B and 2C). Addition of saturating doses of forskolin and 3-isobutyl-1-methylxanthine at the end of each protocol confirmed that the sensor responded appropriately to a rise in intracellular cAMP and further indicated that no oxidant-induced potentiation of forskolin/3-isobutyl-1-methylxanthine activation occurred ( Figure 2C).
Cardiomyocyte culture itself, which was necessary to allow adenoviral gene transduction of the FRET sensor, was associated with a significant increase in the proportion of RIα disulfide formation (up to 72.1±5.8% after 24 hours in culture; Figure 2D). PKARIα was found to be highly oxidized following culture or storage of cells under all ex vivo conditions assessed (Table II in the Data Supplement). As the near-complete induction of disulfide bond formation by culturing could have accounted for the failure of PKARIα activity to increase in response to H 2 O 2 , we also tested whether, in WT cardiomyocytes, PKARIα exhibited greater intrinsic catalytic compared with KI, as evaluated using the H89-inhibitable fraction. As shown in Figure 2E, the FRET response to H89 did not differ between WT and KI cardiomyocytes, consistent with the overall conclusion that disulfide formation has no direct effect on PKARIα catalytic activity.
Given that disulfide bonds form within the A-Kinase Anchoring Protein (AKAP)-binding domain of the RIα subunit, 22 we asked whether PKARIα intracellular anchoring was impacted by the oxidation state. To assess this, we conducted fluorescence recovery after photobleaching experiments-which offer the robust capability of measuring protein diffusion and mobility in live cells 23 -in PKARIα knock-out (prkar1a -/-) mouse embryonic fibroblasts expressing green fluorescent proteintagged WT or mutant RIα proteins. The use of prkar1a -/mouse embryonic fibroblasts allowed us to monitor changes in intracellular anchoring of green fluorescent protein-tagged PKARIα in the absence of endogenous PKARIα, which might compete for available AKAP-binding sites. Nonreduced immunoblotting confirmed the presence of disulfide bond formation in cells expressing PKARIα(WT) or in cells expressing PKARIα(H24A), a mutation known to substantially reduce PKARIα AKAPbinding affinity without affecting disulfide formation, 22 but not in PKARIα(C17S)-expressing cells ( Figure IIIA in the Data Supplement).
Compared with PKARIα(WT), PKARIα(C17S) showed a higher degree of green fluorescent protein-PKARIα diffusive exchange within the photobleached region of interest, as indicated by the higher recovery index ( Figure 3A). This difference was reflected quantitatively as a reduction in the immobile (ie, anchored) fraction of PKARIα in C17S-expressing cells compared with WT ( Figure 3B and 3C), indicating that, in the absence of disulfide formation, less PKARIα is restricted to intracellular compartments. The relative reduction in the immobile fraction for PKARIα(C17S)-expressing cells was equivalent to that found in PKARIα(H24A)-expressing cells ( Figure 3C). Diffusion rate constants for the mobile To test whether the reduction in immobile PKARIα(C17S) was a result of a loss of disulfide-dependent anchoring to endogenous AKAPs, we repeated the fluorescence recovery after photobleaching experiments in cells expressing each of the constructs (WT or C17S) in combination with the RIα anchoring disruptor (RIAD), which prevents PKARIα interaction with AKAPs, with 50-fold selectivity over PKARII. 24 In cells expressing PKARIα(WT), disruption of AKAP binding by RIAD led to a reduction in the immobile fraction of PKARIα to levels comparable with PKARIα(C17S)expressing cells ( Figure 3D and 3E). The effect of RIAD was present only in PKARIα(WT)-expressing cells, with no significant effect of RIAD on cells expressing the "redox dead" RIα (C17S) mutant ( Figure 3E). As before, we saw no effect of the C17S mutant or RIAD on the diffusion rates of the mobile fraction of RIα ( Figure  IIIC in the Data Supplement). Thus, the disulfide state of RIα appears to influence the extent to which PKA is anchored within the cell, through AKAP binding, without affecting the diffusion rate of PKARIα's cytosolic fraction or its catalytic activity.

PKARIα Disulfide Formation Localizes the Holoenzyme to Lysosomal Microdomains in Cardiomyocytes
If PKARIα disulfide formation affects AKAP-mediated intracellular anchoring, then the disulfide state would also be expected to influence PKARIα subcellular compartmentation; however, to date, the identity of these compartments has remained elusive. Taking advantage of the oxidizing conditions of cell culture, which was shown to induce near-complete PKARIα disulfide formation, we determined whether the disulfide state influenced PKARIα subcellular compartmentation in cardiomyocytes using immunofluorescence imaging in cultured WT and KI LV cardiomyocytes. In WT cardiomyocytes, PKARIα was found to colocalize with mitochondria, the nucleus ( Figure

ORIGINAL RESEARCH ARTICLE
and LAMP2-positive lysosomes ( Figure 4A and 4B). Whereas confocal microscopy was sufficient to demonstrate that localization to the mitochondria and nucleus was unaffected by the loss of PKARIα disulfide formation in KI cardiomyocytes ( Figure IVB in the Data Supplement), superresolution stimulation emission depletion microscopy was required to quantify the extent of PKARIα association with the lysosome and assess the redox dependence of this interaction. Stimulation emission depletion imaging allowed accurate identification of lysosomes ( Figure 4C through 4E)-whose average diameter is less than the 200-nm resolution of standard confocal imaging-and significantly improved quantification of PKARIα fluorescence intensity at nanometer distances. 25 By quantifying PKARIα fluorescence intensity at increasing radial distances from lysosomal foci ( Figure 4F; radial incre-ments=70 nm as determined in Figure VA in the Data Supplement) and comparing that with the PKARIα fluorescence intensity measured at randomly generated coordinates, we were able to demonstrate significant clustering of PKARIα to within 70 nm of lysosomes in

ORIGINAL RESEARCH ARTICLE
WT cardiomyocytes ( Figure 4G, Figure VB in the Data Supplement). Using the same approach, we found that PKA cat also clustered near lysosomes ( Figure 4H and 4I), indicating that the entire holoenzyme complex was present in the lysosomal microdomain when PKA was highly oxidized. By contrast, in the absence of disulfide formation (ie, KI cardiomyocytes), PKARIα no longer clustered near lysosomes ( Figure 5A). Clustering of PKA cat was also found to be reduced in KI cardiomyocytes, albeit to a lesser extent ( Figure 5B; P value for significant interaction=0.009). The loss of PKA clustering to the lysosome, however, did not appear to affect gross lysosomal distribution ( Figure VC in the Data Supplement). We next assessed whether clustering of PKARIα to the lysosomal microdomain was mediated by AKAP binding. For this, stimulation emission depletion imaging experiments were repeated using the RIAD disruptor peptide in neonatal rat ventricular myocytes, which are more easily cultured and transfected than adult mouse cardiomyocytes. As with adult LV cardiomyocytes, control-transfected neonatal rat ventricular myocytes showed a high degree of PKARIα clustering to LAMP2-positive lysosomes ( Figure 5C and 5D). RIAD transfection significantly reduced this colocalizationparticularly at the nearest measurable distancewhereas transfection of cells with SuperAKAP-IS, a potent and specific disruptor of PKA-RII:AKAP interactions, did not ( Figure 5C and 5E).
Collectively, these data provide strong evidence that induction of PKARIα disulfide formation facilitates localization of the holoenzyme complex to the lysosome of cardiomyocytes in a manner that is AKAP-dependent.

Intracellular Ca 2+ Release Is Regulated by PKARIα Through Its Interaction With the Lysosomal TPCs
Lysosomes are known to couple with the mitochondria 26 and the cardiomyocyte SR, 27 forming structural microdomains through which lysosomal Ca 2+ release may affect Ca 2+ handling by these organelles. 26,28 We initially assessed, therefore, whether loss of lysosomal-localized PKARIα in KI cardiomyocytes affected mitochondrial or SR Ca 2+ handling (eg, release and reuptake) under steady-state conditions. As before, we found that the conditions required to assess intracellular Ca 2+ handling in WT cardiomyocytes led to near-complete induction of PKARIα disulfide formation (94.5±2.3%; as reported in Table II  Under steady-state pacing (3 Hz; 35±1°C), fura-2-loaded cardiomyocytes showed no significant differences in the intracellular Ca 2+ transient amplitude or diastolic intracellular Ca 2+ levels between genotypes, although a mild increase in the rate of intracellular Ca 2+ decay was observed in KI cardiomyocytes ( Figure 6A through 6D). Derivation of the sarcoplasmic/endoplasmic reticulum Ca 2+ ATPase (SERCA) and Na + /Ca 2+ exchangerdependent rate constants for free intracellular Ca 2+ decay indicated a mild enhancement of SERCA-dependent uptake of Ca 2+ into the SR in KI cardiomyocytes ( Figure 6E), independent of phospholamban phosphorylation or an altered phospholamban:SERCA ratio (Figure VIIA through VIID in the Data Supplement). However, measurement of total SR Ca 2+ content, using rapid caffeine application, showed no genotype-dependent differences ( Figure 6F), indicating that the modest increase in SERCA-mediated Ca 2+ reuptake had no significant effect on SR Ca 2+ loading under these conditions. In agreement with these findings, echocardiographic parameters of LV function were similar in both genotypes (as reported in Table I in the Data Supplement). Equally, we saw no genotype differences in fractional shortening or the rate of relaxation of isolated cardiomyocytes ( Figure VIIE through VIIG in the Data Supplement) in the peak and kinetics of the Ltype Ca 2+ current ( Figure 6G and 6H) or in the Na + /Ca 2+ exchanger current ( Figure 6I). Nevertheless, KI cardiomyocytes displayed a higher incidence of spontaneous Ca 2+ release events during a pause from steady-state pacing ( Figure 6J through 6L), suggesting that the displacement of PKARIα from the lysosomal microdomain in KI cardiomyocytes may be leading to dysregulated lysosomal Ca 2+ release sufficient to directly trigger ryanodine receptor (RyR) opening. For this to occur, close physical proximity between the 2 structures would have to take place. Stimulation emission depletion imaging confirmed that LAMP2-positive lysosomes were closely coupled with RyRs ( Figure 7A and 7B), with nearest-neighbor distance histograms in WT and KI cardiomyocytes showing the majority of the lysosomes lying in close (ie, <200 nm) proximity to RyRs, with no significant difference between genotypes ( Figure 7B).
We therefore assessed the dynamics of intracellular Ca 2+ release from RyR by perfusing WT or KI cardiomyocytes with a 0Na + /0Ca 2+ extracellular solution (which prevents triggering of RyR opening from extracellular sources) and included the use of the reversible RyR inhibitor, tetracaine, to allow for simultaneous quantification of intrinsic RyR Ca 2+ leak. Cardiomyocytes showed stable Ca 2+ transient recordings under tetracaine perfusion, with no spontaneous events occurring in either genotype under these conditions. However, following tetracaine washout,  Figure 7C and 7D).
No differences in the RyR leak/load relationship (an indirect assessment of RyR opening probability 29 ; Figure 7E) or PKA-mediated RyR phosphorylation ( Figure VIIH in the Data Supplement) were found between genotypes, indicating that the Ca 2+ oscillations were unlikely to be driven by inherent changes in RyR opening. By contrast, depletion of lysosomal Ca 2+ stores using acute bafilomycin A1 treatment completely abolished Ca 2+ oscillations under 0Na + /0Ca 2+ conditions ( Figure 7F), whereas competitive inhibition of Ca 2+ -permeable lysosomal TPCs using Ned-19 significantly attenuated the incidence of Ca 2+ oscillations ( Figure 7G). Measurement of SR Ca 2+ load (in nonoscillating cells) indicated that the ability of either drug to prevent global SR Ca 2+ oscillations was not a consequence of reduced SR Ca 2+ content (Figure VII I and VIIJ in the Data Supplement), supporting the conclusion that these events were a direct result of spontaneous lysosomal Ca 2+ release from TPCs, occurring when PKARIα was no longer localized to the lysosome. Despite the presence of spontaneous SR Ca 2+ release in KI cardiomyocytes, we did not observe an increase in pacing-induced ventricular arrhythmias in these mice ( Figure VIIIA in the Data Supplement). Likewise, there was no evidence for induction of Ca 2+ -activated stress responses in KI hearts. Specifically, transcript levels for multiple markers of the unfolded protein responsea conserved system of endoplasmic reticulum stress signaling cascades activated in response to protein misfolding or altered SR/endoplasmic reticulum Ca 2+ content-showed no evidence of increased transcriptional activation in KI LVs ( Figure VIIIB in the Data Supplement). KI LVs also showed no marked difference in the conversion of LC3-I to LC3-II or degradation of p62 ( Figure VIIIC in the Data Supplement), which, together, indicated that activation of the autophagosome-lysosome pathway was not altered in these mice.

Redox-Dependent Regulation of Lysosomal Ca 2+ Release by PKARIα Is Cardioprotective Against I/R Injury
SR Ca 2+ oscillations are known to occur in the initial period of myocardial reperfusion, leading to cell death and LV dysfunction. 30 Given our observation that PKARIα disulfide formation is induced shortly after myocardial reperfusion in humans and mice, we posited that inhibition of global Ca 2+ release by oxidized, lysosomally targeted PKARIα may confer cardioprotection in the postischemic heart. To test this hypothesis, hearts from WT and KI mice were subjected to ex vivo I/R, with LV function measured throughout and infarct size assessed following the 60 minutes reperfusion (Figure 8A). Furthermore, to determine the contribution of TPC-dependent lysosomal Ca 2+ release, hearts of either genotype were administered Ned-19 (or dimethyl sulfoxide vehicle) at the time of reperfusion.
Although no difference in LV hemodynamic measurements were seen during the baseline stabilization period ( Table III in the Data Supplement), KI hearts administered vehicle at reperfusion showed significantly lower LV developed pressures throughout the reperfusion period ( Figure 8B) and displayed 2-fold larger infarcts compared with WTs ( Figure 8C and 8D). Absence of differences in PKA-dependent RyR phosphorylation during I/R ( Figure  IX in the Data Supplement) ruled out the possibility that direct alterations in RyR accounted for the poorer outcome in vehicle-treated KI mice. Instead, inhibition of lysosomal Ca 2+ release, by addition of Ned-19 at the time of reperfusion, was sufficient to restore both contractile function and infarct size in KI hearts to levels comparable with WT, with no further protective effects observed in WT hearts. These findings are consistent, therefore, with a model in which disulfide-modified PKARIα limits I/R-induced Ca 2+ overload by decreasing lysosomal triggering of global SR Ca 2+ release.

DISCUSSION
Our findings led us to 3 major conclusions: (1) PKARIα disulfide formation is a consistent and conserved response to myocardial I/R injury in vivo, occurring both in humans and mice; (2) oxidation of PKARIα serves as a means to compartmentalize PKARIα within the lysosomal microdomain, where it acts as an inhibitor of TPC-dependent Ca 2+ release; and (3) this regulatory mechanism is an adaptive response to I/R, which allows the heart to limit the extent of injury and aid functional recovery. Figure 5 Continued. distance from the lysosome. Normalized data for WT cells were calculated from those included in Figure 4I. Data shown as mean±SEM; repeated measures 2-way ANOVA with Bonferroni correction, P=0.005 for significant interaction between genotypes and distance; n=35 to 37 cardiomyocytes, each from 3 mice/genotype. C, Neonatal rat ventricular myocytes were transfected for 24 hours with GFP only (transfection control), GFP + the RI α anchoring disruptor, (RIAD) or GFP + the RII anchoring disruptor, super-AKAP-IS, and then fixed and coimmunostained for PKARIα (Atto-647, yellow) and LAMP2 (AF-594, magenta). STED images were acquired only in GFP-positive cells. D, PKARIα fluorescent intensity (arbitrary units [a.u.]) was quantified within a given distance from each lysosome as in Figure 4F (binned by Atto-647 resolution; 80 nm). P<0.0001 for significant interaction, P<0.001 for difference between foci at 80, 160, and 240 nm distance determined by post-hoc testing. E, Comparison of normalized intensities within 80 nm of LAMP2-positive vesicles was made between cells transfected with GFP only and those transfected with GFP + RIAD or GFP + super-AKAP-IS. Normalized data for the GFP group were calculated from cells used in Figure 5D. Data shown as mean±SEM; repeated measures 2-way ANOVA with Bonferroni correction, *P<0.001 for GFP versus GFP + RIAD and #P<0.01 for GFP + RIAD versus GFP + super-AKAP-IS; n=28 to 38 cardiomyocytes, each from 3 independent isolations/conditions. For A through C, scale bar=1 μm and 500 nm for all magnified images. AKAP indicates A-kinase anchoring protein; GFP, green fluorescent protein; KI, knock-in; PKA, protein kinase A; PKA cat , protein kinase A catalytic subunit; PKARIα, regulatory subunit Iα-containing protein kinase A; and WT, wild-type.

Functional Impact of PKARIα Disulfide Formation on Kinase Function and Localization
Although redox modification of several kinases has been shown to promote catalytic activation, 21 our data rule out the possibility that PKARIα disulfide formation has the same effect. Instead, we provide strong evidence that the principal regulatory function of PKARIα disulfide formation is to promote localization of the holoenzyme complex to distinct subcellular compartments through enhanced AKAP binding. This  ORIGINAL RESEARCH ARTICLE conclusion is at odds with some previous studies, which report increased PKA activity in response to elevations in ROS 31,32 and reactive nitrogen species. 33 However, the observed changes in PKA catalytic activity were inferred from downstream functional readouts-some of which have important limitations 34 -or increased substrate phosphorylation, which make it difficult to distinguish between genuine increases in PKA catalytic activity versus focused subcellular targeting of the enzyme. The use of a genetically encoded FRET biosensor here, in conjunction with PKARIα KI cardiomyocytes, provided a robust means to directly assess changes in intrinsic catalytic activity with varying degrees of PKARIα disulfide formation, and showed no correlation between the 2, suggesting that previous findings may instead be a consequence of substrate-induced activation within specific microdomains, 35 a characteristic unique to RIα-containing PKA. The observation that physically restricted pools of PKARIα are lost when disulfide formation is prevented (C17S mutation) argues strongly for a concomitant loss of AKAP-mediated anchoring of PKARIα that is dependent, at least in part, on the structural stability afforded by the disulfide bond. 22 In support of this hypothesis, disruption of RIα-AKAP interaction in fluorescence recovery after photobleaching experiments caused significant liberation of PKARIα from the immobile pool, an effect that was not observed in C17S-expressing cells. Likewise, in cardiomyocytes both the C17S "redox dead" mutation and disruption of RIα-AKAP binding using RIAD resulted in displacement of the PKARIα holoenzyme from lysosomes. For the latter, lysosomal

Regulation of Lysosomal Ca 2+ Release by PKARIα
Oxidation-dependent localization of PKARIα to the lysosome represents a significant, and entirely novel, mechanism through which PKA regulates Ca 2+ release in the heart. Our classic understanding of Ca 2+ regulation by PKA involves the rapid phosphorylation of key Ca 2+ handling proteins, including RyR, phospholamban, and the L-type Ca 2+ channel, which act concordantly to enhance contraction and relaxation. However, several studies suggest that these substrates are uniquely targeted by RIIcontaining pools of PKA, whereas activation of PKARIα has little effect on excitation-contraction coupling. [36][37][38] Consistent with this, we find no evidence for differential phosphorylation of "classic" PKA targets (eg, in RyR or phospholamban), nor do we see differences in cardiomyocyte contractility or LV systolic function in vivo or in isolated hearts between WT and KI mice. Instead, we find that the striking Ca 2+ oscillations and spontaneous Ca 2+ release events observed in KI cardiomyocytes are a result of dysregulated Ca 2+ release from lysosomal TPCs when PKARIα is absent from this microdomain. Lysosomal Ca 2+ efflux can promote SR Ca 2+ release either by triggering RyRs Ca 2+ release directly 39 or by enhancing SR Ca 2+ loading. 28 We observed minimal enhancement of SERCA-mediated Ca 2+ uptake in KI cardiomyocytes, with no obvious difference in SR Ca 2+ load or RyR leak compared with WT. Instead, we found spontaneous triggering of SR Ca 2+ release (in the absence of sarcolemmal Na + /Ca 2+ flux), which could be prevented by inhibiting RyR opening or lysosomal Ca 2+ release through TPC channels. These findings, and the fact that we found lysosomes lying in close physical proximity to the RyR, indicate that PKARIα directly modulates the crosstalk between lysosomal TPCs and RyRs. Of note, the prevention of SR Ca 2+ release by these drugs was not driven by a reduction in SR Ca 2+ load. In fact, in KI cardiomyocytes, Ned-19 significantly increased load. Because this would be expected to promote Ca 2+ oscillations by increasing RyR opening probability, it is not likely that the increased load had a direct effect on the inhibitory effects of Ned-19.

PKARIα Disulfide Formation as an Adaptive Response to I/R
Although early bursts of ROS are thought to be the primary mediators of reperfusion-induced injury, 1,2 evidence indicates that some degree of ROS are needed at the time of reperfusion to protect the heart. 4,5 In particular, I/R-induced elevations in NADPH oxidase (NOX)derived ROS activate redox signaling pathways that promote cell survival. 3,40,41 In this regard, compartmentation of oxidant sources and their downstream targets is suggested to be a discriminating factor between adaptive versus maladaptive signaling. 3 Consistent with this, we observed that disulfide-dependent compartmentation of PKARIα was a necessary event to confer cardioprotection. When this response was lost in KI mice, cardiomyocytes exhibited dysregulated lysosomal Ca 2+ release, which ultimately led to exacerbated I/R injury.
The fact that pharmacological inhibition of TPCs at the time of reperfusion reduced infarct size and improved functional recovery in KI hearts highlights a causal role for lysosomal Ca 2+ release in mediating PKARIα's adaptive response to I/R. Inhibition of TPCs, either pharmacologically or by genetic knockdown, has been shown to protect the heart against I/R injury, both in vitro and in vivo. 42 It is interesting that pharmacological inhibition of TPC conferred cardioprotection in WT mice only when a more potent derivative of Ned-19, called Ned-K, was used. 42 This may explain why, in keeping with the same report, 42 we did not see an added benefit in WT hearts perfused with Ned-19.
Although our observations provide strong evidence for a cardioprotective role of oxidized PKARIα independent of increases in cAMP, it is conceivable that further activation of RIα-containing pools of PKA by cAMP may provide additional protection from I/R injury. Glucagonlike peptide-1 and prostaglandin E1, which are currently being tested in clinical trials for their use in treating myocardial infarction 43,44 and reperfusion injury, 45,46 have both been found to promote selective activation of RI-but not RII-containing PKA in cardiomyocytes. 36,38 Preclinical studies have already shown that the beneficial effects of these hormones rely on PKA, 47,48 although the downstream mechanisms have yet to be fully elucidated. Our findings indicate that inhibition of lysosomal Ca 2+ release by PKARIα may contribute to the cardioprotective effects of these hormones and further suggest that targeted enhancement of lysosomal PKARIα, or inhibition of TPCs, offers a novel adaptive signaling pathway to exploit for the prevention of I/R injury.

Potential Limitations
Although we showed that disulfide formation is important for restricting PKARIα to lysosomal regions, it is not clear from our data whether regulation of lysosomal function occurs through a direct interaction between PKARIα and TPCs or through more distal PKA-dependent signaling events in this microdomain. In vitro PKA can directly phosphorylate TPCs and alter channel opening 49 50 neither of which was found to be materially different between KI and WT mice. It should also be noted that the contribution of enhanced SR Ca 2+ oscillations to the exacerbated I/R injury seen in KI mice was inferred from studies in isolated cardiomyocytes, as opposed to a direct assessment during I/R. Nevertheless, our data in KI mice showing that Ned-19 prevents SR Ca 2+ oscillations in cardiomyocytes and limits myocardial I/R injury strongly support a link between them.

CONCLUSIONS
Our work identifies, for the first time, oxidation-dependent compartmentation of the PKARIα holoenzyme to the lysosomal microdomain, where it acts as a potent inhibitor of intracellular Ca 2+ release. In the setting of I/R, where PKARIα disulfide formation is induced, this regulatory mechanism is critical for limiting infarct size and offers a novel target for the design of cardioprotective therapeutics.