Introduction

Sudden cardiac arrest is a major cause of death in the Western world,¹ and mortality after resuscitation remains >50%.^2,3 This excess mortality is reflected in the post–cardiac arrest syndrome,² a condition that comprises postarrest central nervous system dysfunction, postarrest ischemia/reperfusion, post–myocardial arrest dysfunction (PMAD), and continuation of the factors precipitating sudden cardiac arrest. PMAD is believed to be a form of myocardial stunning^4–6 and has been observed for hypoxic arrest and after ventricular fibrillation.⁷ Recognizing that mortality remains high after resumption of circulation, the International Liaison Committee on Resuscitation has recommended more detailed resuscitation research, including the phenomenon of PMAD.¹

PMAD may result from oxidative stress–induced cytosolic calcium overload. Calcium overload has traditionally been explained by 2 processes: reverse-mode operation of the sodium-calcium exchanger related to intracellular sodium overload during ischemia and sarcoplasmic reticulum ATPase dysfunction related to ATP depletion.⁸ These processes have been suggested to lead to reversible myofilament dysfunction and subsequently to irreversible cell death partly through Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) signaling pathways.^8–10 The relationship between acute calcium overload and long-term augmentation of calcium-induced calcium release (CICR) is critical but understudied. In particular, the biochemical pathways identified would be predicted to potentiate CICR. In addition, because normal CICR generates, on a beat-to-beat basis, systolic calcium levels in localized domains 10-fold higher than predicted to occur during postarrest calcium overload,¹¹ any CICR augmentation might be positioned to play a dominant signaling role. Moreover, as it relates to excitation-contraction coupling, potentiated CICR transients would be expected to result in more forceful contractions, not the opposite. For example, in a post–cardiac arrest study, ryanodine receptor (RyR)–dependent calcium release was shown to support reversible injury.¹² However, despite data in support of such a mechanism, no long-lasting changes in CICR have been demonstrated after reperfusion¹³ except for 1 early report.¹⁴ Alternatively, the increased CICR transients could be a consequence of a mismatch between cell contractility and external mechanical demand whereby reduced shortening, or diastolic stretch, of cells may raise intracellular Ca²⁺ levels¹⁵ or Ca²⁺ releasability from the sarcoplasmic reticulum.¹⁶ Two limitations of previous experimental models are that there is an absence of in vivo data and that much of the in vitro mechanistic data have come from the buffered saline–perfused Langendorff preparations that suffer from significant edema.¹⁷ In addition, CICR and CaMKII signaling have been linked to β-adrenergic signaling under pathological conditions,¹⁵ and cardiac arrest has been associated with dramatic elevations in catecholaminergic signaling, which is not present in the isolated heart model.¹⁸ We believed that a model preserving the neurohumoral axis might better reflect the in vivo environment and developed an in vivo model of cardiac arrest with resuscitation using extracorporeal membrane oxygenation (ECMO).¹⁹ Here, we report findings from this in vivo cardiac arrest–ECMO model relating to PMAD. Two novel approaches were introduced in this study. First, we examined cellular force in vitro using microcarbon fiber assessment simultaneously with CICR measurements. Second, we developed an approach to measuring calcium dynamics in vivo using a genetically engineered calcium indicator. Our data from this in vivo model support the contribution of calcium potentiation, CaMKII signaling, and oxidative stress to PMAD and suggest a potential new therapeutic candidate, Alda-1.

Methods

Studies were approved by the Stanford Administrative Panel on Laboratory Animal Care and conform to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH).

Invasive Hemodynamics and Development of ECMO

Male Sprague-Dawley rats (weight, 250–350 g) were anesthetized and oxygenated/ventilated with 1% to 2% isoflurane, maintained at 37°C, and monitored throughout for depth of anesthesia. An ECMO blood circuit was built to support reanimation after cardiac arrest (Figure 1A).²⁰ A conductance catheter (Millar Instruments, Houston, TX) was introduced into the left ventricle for pressure-volume loop measurements as described previously.^21,22

• Download figure
• Download PowerPoint

Animal Models of Disease

After institution of the ECMO circuit (Figure 1A), sudden cardiac arrest was initiated either by cessation of ventilation or by rapid ventricular pacing to induce ventricular fibrillation (Figure 1B–1E).¹⁹ Both processes led to rapid hypotension, followed by pulseless electric activity or asystole (Figure 1C). The arrest period was 15 minutes, after which ECMO was initiated and the heart resumed spontaneous beating (Figure 1F). We saw no differences between these 2 modes of cardiac arrest. Data are presented from the hypoxic cardiac arrest model because of the similarity to existing descriptions of isolated Langendorff heart ischemia/reperfusion. For selected experiments, rat hearts were explanted and retrogradely perfused on a Langendorff system. After stabilization (pre-arrest) for 10 minutes, hearts were subjected to 15 minutes of no-flow ischemia, and coronary perfusion was restored for 30 minutes.¹³ During explanted ischemia/reperfusion experiments, HEPES-buffered saline (See “Isolation of Left Ventricular Cardiac Myocytes”) was bubbled with 98% oxygen/2% isoflurane. In select experiments, heparinized blood was used to perfuse Langendorff hearts.

Isolation of Left Ventricular Cardiac Myocytes

Within 5 minutes of heart explantation from arrested animals weaned from ECMO (typically within 20 minutes after initiation of ECMO) or Langendorff hearts when specified, left ventricular cardiac myocytes were isolated according to our published protocols.²³ Experiments were performed with HEPES-buffered solution containing (in mmol/L) 1 CaCl₂, 137 NaCl, 5.4 KCl, 15 dextrose, 1.3 MgSO₄, 1.2 NaH₂PO₄, and 20 HEPES (pH 7.4) maintained at 37.7°C with a feedback control system (IonOptix, Milton, MA).

Myocyte Unloaded Shortening and Relaxation

Myocyte contraction properties, including sarcomeric length, were evaluated with a sarcomeric imaging acquisition system (SarcLen; IonOptix) as previously described.²³

Two-Carbon Fiber Technique

Carbon fibers (CFs; 12-µm diameter, kindly provided by Professor Jean-Yves LeGuennec)²² affixed to miniature hydraulic manipulators (SM-28; Narishige, Tokyo, Japan), computer controlled with a piezoelectric translator (P-621.1CL; Physik Instruments, Karlsruhe/Palmbach, Germany) and mounted on a custom-made railing system (IonOptix), were attached to single isolated ventricular myocytes. CFs were stretched axially by the piezoelectric translator controlled with custom software (Matlab; MathWorks, Natick, MA). CF bending and force were measured and analyzed with IonWizard.²⁴ Individual systolic and diastolic cell results were plotted and fitted with a linear regression (Origin; OriginLab, Northampton, MA) to give both end-systolic and end-diastolic length-tension relationships. Data were normalized to the starting systolic force or as the Frank-Starling gain when specified.²⁵

Calcium Transient Measurements

In vivo CICR measurement was performed with a GCaMP6f genetically encoded calcium indicator delivered by an AAV9 vector that was directly injected into the left ventricular apex of anesthetized rats (described in the “Delivery of AAV9-GCaMP6f” below). After 3 or 4 weeks, fluorescence transients were imaged with a custom-built fiberoptic excitation imaging system (AUST Development, LLC) attached to a camera system described previously^26,27 and analyzed with ImageJ (NIH). For cell experiments, cardiomyocytes were loaded with 0.2 µmol/L Fluo-5f-AM (Molecular Probes, now part of Thermo Fisher, Waltham, MA) for 30 minutes and then allowed to incubate in dye-free HEPES-buffered saline for an additional 30 minutes to allow de-esterification of the calcium dye. Spatially averaged electrically evoked calcium transients were measured with a standard FITC cube (Chroma, Bella Falls, VT) using the HyperSwitch system (IonOptix). Fluorescence transients were normalized to ΔF/F units.²⁶ Myocyte calcium transient properties, including ΔF/F, rise time, and decay time, were evaluated with a calcium imaging acquisition system (IonOptix) as previously described.²³

Delivery of AAV9-GCaMP6f by Direct Intramyocardial Injection

Adult male Sprague-Dawley rats (weight, 275–300 g) were anesthetized with 2% isoflurane in oxygen. A left thoracotomy was performed after intubation and ventilation, and 150 μL normal saline containing AAV9-GCaMP6f (total titer 2e12, Penn Vector Core) was injected into 3 spots at equal volume at the apex of the left ventricular wall. After the successful injection, the chest was closed, and the animal was extubated and allowed to recover. GCaMP6f signal was detected at 3 to 4 weeks after delivery.

Pharmacological Reagents

Epinephrine infusion was performed with 23 µg·kg⁻¹·min⁻¹ for 30 minutes through venous access. In select experiments, control cardiomyocytes were exposed to 100 nmol/L epinephrine in the extracellular solution. For ryanodine experiments, 1 mg dissolved ryanodine (Sigma, St. Louis, MO) was injected into the femoral vein empirically to stop the heart. To inhibit CaMKII, 1 mg KN92 or KN93 (Sigma) or 100 µg myristoylated autocamtide-2–related inhibitory peptide (AIP; Calbiochem or Anaspec) was used. A myristoylated AC3-C peptide (KKALHAQERVDCL, synthesized by Anaspec), an inactive peptide inhibitor of CaMKII, was used as a control.²⁸ To inhibit protein kinase A (PKA), 1 mg H89 (Sigma) was raised in 1 mL of 0.9% normal saline and infused over 10 minutes. Experimental protocols were initiated 10 minutes after infusion. Alda-1, 32.4 mg Alda-1 dissolved in 2 mL 50% polyethylene glycol/50% PBS, was injected intraperitoneally 30 to 60 minutes before the experimental protocol.

Sham Experiments

Sham experiments with animals on ECMO for 30 minutes yielded normal values in isolated cells (data not shown). Control animals refer to no ECMO unless otherwise stated.

Lucigenin-Enhanced Chemiluminescence

Total O₂^•− was measured by lucigenin-enhanced chemiluminescence as described previously.²⁹ Chemiluminescence was recorded by luminometer (Bio-Orbit, Turku, Finland) with 5 μmol/L lucigenin. In parallel experiments, ventricular tissue was incubated in N^G-monomethyl-l-arginine (1 mmol/L), diphenylene iodonium (10 µmol/L), or Alda-1 (20 µmol/L). After measurement of baseline readings for 4 minutes, samples were equilibrated and dark adapted for 5 minutes, and chemiluminescence was recorded for 10 minutes. Recordings were performed by researchers blinded to sample identity. Results were expressed as counts per second per milligram of tissue dry weight.

Oxidative Fluorescent Microtopography

O₂^•− was detected in the ventricle with the fluorescent probe dihydroethidium (Molecular Probes) as described previously.²⁹ Cryosections (30 μm) were incubated in physiological buffer for 30 minutes at 37°C with either Alda-1 (20 µmol/L) or mito-TEMPO (10 µmol/L) followed by 5 minutes of dark incubation with 2 μmol/L dihydroethidium. Images were obtained on a confocal microscope at ×60 (Bio-Rad MRC-1024 laser; filter settings: excitation filter, 488 nm; emission filter, 550 nm) and quantified (red intensity times area) with Image-Pro Plus software (Media Cybernetics, Rockville, MD). Analysis was performed by researchers blinded to sample identity. Mean fluorescence was calculated from 4 separate, high-power fields from each quadrant to produce n=1.

Protein Extracts and Western Blot

Heart tissues were homogenized in lysis buffer containing 50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 1% NP-40, 0.5% sodium deoxycholate, 1% SDS, and Halt protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific). Heart tissue homogenates were separated on 4% to 20% TGX gel (Bio-Rad) and transferred to Immobilon-FL polyvinylidene fluoride membrane (Millipore). Membranes were blocked in Odyssey TBS blocking buffer and incubated with respective primary antibodies: CaMKII (Santa Cruz Biotechnology), phospholamban (2D12), RyR (under nonreducing conditions; 34C; Thermo Fisher Scientific), phospho-CaMKII Thr286/287 (D21E4; Cell Signaling), phospholamban pThr17, phospholamban pSer16, RyR2 pSer2814 (Badrilla), and GAPDH (Sigma). IRDye 800CW goat anti-mouse IgG and IRDye 680LT goat anti-rabbit IgG were used as secondary antibodies (LI-COR). Quantitative Western images were acquired with an Odyssey Fc Imager, and images were analyzed with Image Studio Software (LI-COR).

Statistics

For comparisons between independent samples, the t test using the Welch-Satterthwaite correction for unequal variances was used. One-way Kruskal-Wallis tests were performed for multiple-group comparisons. For significant differences, post hoc testing was performed. P values from post hoc testing were corrected for multiple comparisons with the Bonferroni correction. Exact P values are stated in most cases, and a value of P<0.05 was taken to indicate a statistically significant difference between means. All data are presented as mean±SEM.

For comparisons with nested designs (multiple cells from different numbers of rats), a mixed-effects model was used to account for the correlation between cells from the same rat. The treatment (PMAD versus control) was modeled as a fixed effect, and the animal was modeled as a random effect. We fitted the mixed models using maximum likelihood and tested the significance of the treatment fixed effect.

Results

Myocardial Dysfunction Is Found in a Rodent Model of Cardiac Arrest and Resuscitation

Pressure-volume hemodynamic measurements were made before (Figure 1G) and after (Figure 1H) cardiac arrest. Analysis showed significant decrement in contraction and relaxation after cardiac arrest (end-systolic pressure-volume relationship, −44%; cardiac input, 48%; preload recruitable stroke work, −31%; τ +500%; see Table). These data establish PMAD in our model.

We next examined unloaded sarcomere shortening (%SL) and peak contraction (v_c) and relaxation (v_r) velocity in isolated adult cells from control and PMAD hearts. For control versus PMAD cells, the %SL was 13±0.3 versus 9±0.3, respectively (P=1.61×10⁻⁶), a 23% decrease in the %SL in unloaded cells (data not shown; n=143 cells from 25 animals and 187 cells from 26 animals for control and PMAD, respectively, for data in the Table), whereas v_c was 4.5±0.2 versus 2.6±0.1 µ/ms (P=6.78e−7) and v_r was 2.7±0. versus 2.0±0.1 1 µ/ms (P=0.0078; data not shown; n=136 cells from 11 animals and 164 cells from 12 animals for control and PMAD, respectively, for above data).

We further characterized this contractile dysfunction in PMAD by measuring single-cell force and contractility using the 2-CF technique.³⁰Figure 2A shows an adult left ventricular myocyte before and during contraction in response to electric stimulation. Figure 2B and 2C shows the stepwise stretch protocol used to assess myocyte function. Figure 2C shows CF bending for the same stretched cell.³⁰ Corresponding to the derivation by regression of load-independent values in vivo, the end-systolic and end-diastolic length-tension relationships are defined in single cells as the slopes of the linear regressions of systolic and diastolic force (proportional to CF bending; see Methods) and plotted versus sarcomere length (Figure 2D and 2E). In the Table, pooled analysis of the slopes of the ESTLR and EDTLR for both control and PMAD cells is shown, demonstrating both to be depressed (60% and 36% of control, respectively) in PMAD cells (P<0.05). In addition, the Frank-Starling gain was used to compare the change in linear slopes and at each stretched length by dividing the active force by the passive force (Figure 2F).²⁵ These cellular data recapitulated load-varying curves obtained with pressure-volume loops and demonstrated impaired force development. We also explored the force-frequency relationship for both control and PMAD cells (Figure 2G). This demonstrated that the force-frequency relationship was depressed in PMAD cells at all pacing frequencies investigated (0.5–4.0 Hz).

• Download figure
• Download PowerPoint

Cardiac Myocytes From Postarrest Animals Exhibit Markedly Enhanced Peak Calcium Transients

Force development is critically regulated by CICR. Previous reports examining postarrest CICR have found little change but did so using high-affinity calcium indicators,^13,31 whereas rapid calcium transients were subject to dye kinetic filtering that could both mask important phenotypic differences and alter cellular contractility.²⁶ Our findings were consistent with this effect for Fluo-4 but not for the lower affinity dye Fluo-5f (online-only Data Supplement Figure I). Therefore, we combined measurements of mechanical load and CICR simultaneously in dynamically contracting cardiomyocytes (Figure 2H). It has been reported that CICR does not change with acute stretch in either amphibian cells²⁹ or mammalian heart cells. We observed this to be the case (Figure 2H).

We next explored CICR after cardiac arrest using the approach outlined earlier in Figure 2H. In Figure 3A, a single sweep of a steady-state pacing train is shown for a PMAD and a control cell. Consistent with one report,¹⁴ we found postarrest cells to have markedly enhanced peak calcium transients. In pooled analysis (Figure 3B), a significant 290% increase in CICR amplitude was present (ΔF/F_peak from control [n=97 cells from 11 animals] and PMAD [n=72 cells from 12 animals] were 99±8 and 272±13, respectively; P=1.53×10⁻⁶). We found no difference in the time to peak (t_p) of the calcium transient (52±4 and 46±10 milliseconds; P=0.11; same n as above). The time to 90% decay (t_d) of the CICR in PMAD was prolonged (340±9 versus 456±9 milliseconds; P=4e−15), as was the time constant of decay (τ; 152±5 versus 216±7 milliseconds). We cannot exclude that this effect is from dye kinetic filtering as the dye nears saturation in the PMAD group, rather than a true finding. The augmented CICR was present in isolated cells long after in vivo cardiac arrest (at least 8 hours, which was the longest time point investigated); hence, we labeled this phenomenon calcium long-term potentiation. In contrast, consistent with predicted effects of high-affinity indicators,²⁵ we found that Fluo-4 minimized this difference in CICR (online-only Data Supplement Figure I). At times, the calcium potentiation led to spontaneous delayed calcium increases, a cellular surrogate candidate for delayed afterdepolarizations. In Figure 3C, the top panel shows sarcomere shortening under load, and the bottom panel displays CICR for the same fiber. After stretch (arrow), a delayed calcium increase occurred and induced chaotic contraction without coordinated force generation. In this case, with rapid pacing, the fibrillatory behavior converted to organized contraction. All cells investigated from all PMAD hearts exposed to load displayed this phenomenon (n=12 from 4 rat hearts). Rarely, load-independent spontaneous delayed calcium increases were seen, as shown in Figure 3D. Such behavior was never seen in control cells.

• Download figure
• Download PowerPoint

Simultaneous measurements of CICR and force generation allowed us to test whether length-dependent activation, a putative mechanism for the Frank-Starling effect,³⁰ was impaired in PMAD by comparing contraction velocity immediately before and after stretch (Figure 3E). Care was taken to equalize baseline and stretched sarcomere lengths. The poststretch change in velocity was similar between control and PMAD, but the absolute velocity was also lower in PMAD (ratio of poststretch to prestretch v_c, 1.7±0.2 versus 1.8±0.2 for control versus PMAD; P=0.5). This is important because, although PMAD is associated with reduced baseline myofilament calcium sensitivity, the response to stretch (length-dependent activation) is preserved.

To examine the multidimensional relationship between force, frequency, and CICR, we plotted these together in 3-dimensional space. This demonstrated the reduced myofilament sensitivity seen at all frequencies and the extent to which the potentiation of the calcium transient abrogates the reduced contractile force (Figure 4A and 4B).

• Download figure
• Download PowerPoint

Because previous studies in vitro have not found an augmented CICR after arrest, we compared our data with data from cells isolated from saline-perfused bubble-oxygenated Langendorff resuscitated hearts. Cells from these hearts showed a more minor depression of %SL and no difference in CICR compared with controls (Figure 3B), in agreement with most reported results.^6,13 However, when the control Langendorff heart was perfused with oxygenated and pH-controlled blood from a cardiac arrest animal, augmented CICR was demonstrated (PMAD-Lan-Blood, Figure 3B), suggesting a dependency of effect not on neural innervation of the heart but on an intact circulation. Because cardiac arrest is known to be associated with marked increases in both catecholamines^18,32 and reactive oxygen species, we hypothesized these to be key mechanisms.

Potentiation of CICR Begins In Vivo

Although responses in isolated cells are often used as a surrogate for in vivo effects and the control animals gave us confidence that this effect was related to the cardiac arrest, the possibility remained that this was an artifact of that isolation process for those cells. To assess whether that CICR was augmented in vivo, we transduced rat hearts with AAV9-GCaMP6f vectors and measured in vivo CICR using a custom-built fiberoptic imaging system (AUST Development, LLC) tunneled under the skin to the heart (Figure 5A and 5B). Using this system, we were able to make sequential measurements in the same rat (eg, a week apart in Figure 5C). An example fluorescent image (acquired through the fiberoptic placed on the heart) from baseline and peak calcium transient is shown in Figure 5B. We then performed a cardiac arrest experiment as described above with the additional measurement of CICR by way of genetically engineered calcium indicator fluorescence. CICR after arrest was potentiated in this in vivo measurement, similar to our findings at the cellular level from this same model and the blood-perfused preparation described above, but in contrast to the traditional saline-perfused Langendorff preparation. Furthermore, the proarrhythmic potential of delayed calcium afterdepolarizations in this postarrest setting was seen, in agreement with our in vitro data (Figure 5E). We found the peak ΔF/F to be ≈300% increased after cardiac arrest compared with the internal baseline prearrest control (n=2 hearts). This increase, when corrected for heart rate by atrial pacing to similar heart rates, is less dramatic and approximates ≈200%.

• Download figure
• Download PowerPoint

Catecholamine Excess Reproduces Calcium Potentiation

It has been demonstrated that catecholamines are dramatically elevated with in vivo cardiac arrest.¹⁸ To test the hypothesis directly in vivo that catecholamine signaling was responsible for the augmented CICR, we blunted catecholamine signaling with β- and α-blockade (labetalol). However, elevated catecholamines are also required for survival, and the presence of signaling antagonists in vivo prevented adequate resuscitation, as might be expected. Therefore, to examine the role of catecholaminergic activation in the promotion of cardiac myocyte calcium potentiation in vivo, a stable, ventilated animal was infused with epinephrine for 30 minutes demonstrating its profound elevation after arrest. Infusion induced a typical adrenergic response, characterized by an increase in heart rate and mean arterial pressure. On termination of epinephrine infusion before cell isolation, blood pressure and heart rate normalized (blood pH was also normal) within 10 minutes. In isolated cardiomyocytes from this model, a pronounced increase in CICR was seen similar to after cardiac arrest (Epi, Figure 3A and B). We further explored whether cardiac myocyte calcium potentiation was sustained after brainstem herniation, a condition known to induce a physiological endogenous catecholamine surge.³³ This led to a similar dramatic increase in heart rate and blood pressure. In cells isolated from these hearts, a similar increase in ΔF/F_peak was seen (271±21; n=110 cells from n=9 animals; P=1e−14 versus control; Figure 3B, brainstem herniation, and online-only Data Supplement Figure 2II).

We also added epinephrine directly to the cell bath of control cells, which, as expected, acutely manifested a markedly increased contractile performance (%SL, 20±0.5) in association with a more modest increase CICR (ΔF/F_peak, 141±0.53%). This was in marked contrast to the prolonged administration in vivo as described earlier.

Cardiac Myocyte Calcium Potentiation Is Dependent on CaMKII and Ryanodine but Not on PKA

Because a link between subacute and chronic β-adrenergic signaling– and CaMKII-, rather than the acute PKA-, dependent pathways has been observed,^34–36 we tested whether CaMKII inhibition could block the induction of CICR potentiation in our in vivo cardiac arrest model. As shown in Figure 3B, AIP, a CaMKII inhibitor, given before and during cardiac arrest, abolished the CICR increase in both the PMAD and epinephrine-infusion models. The lower-affinity inhibitor KN93 blunted cardiac myocyte calcium potentiation in PMAD (Figure 3A and 3B) but not in epinephrine infusion. In the presence of AIP or KN93, no delayed afterdepolarizations or stretch-induced cellular fibrillatory behavior was seen. Because CaMKII activation is partly mediated by calcium-dependent pathways and CaMKII is known to be localized to the dyad, we tested whether inhibiting CICR during cardiac arrest would blunt cardiac myocyte calcium potentiation. We found that both for epinephrine infusion and PMAD, simultaneous ryanodine application blocked the augmentation in CICR (Figure 3B), suggesting that calcium release by RyR was a critical contributor (In the presence of ryanodine, animals were supported with ECMO). In contrast, the PKA inhibitor H89 did not prevent the increase in CICR either before or during cardiac arrest (Figure 3B), suggesting that CaMKII activation was a critical component. The sham inhibitor KN92 had no effect on CICR in PMAD (Figure 3B).

Cardiac Myocyte Calcium Potentiation Supports Impaired Cardiac Function

We measured both %SL (Figure 3F) and force (Figure 3G and 3H) in cells isolated from PMAD hearts that were exposed to CaMKII inhibitors, KN93, AIP, or ryanodine, during the initial cardiac arrest to inhibit CICR augmentation. We found that both %SL and baseline force were lower under these conditions, suggesting that the augmented CICR is necessary to support cardiac contractility after cardiac arrest.

CaMKII-Dependent Phosphorylation in PMAD

In the brain, CaMKII is a mediator of long-term potentiation, a form of synaptic plasticity, and is associated with enhanced calcium signaling.^37,38 CaMKII activity is important for calcium-regulated/mediated activities such as excitation-contraction coupling and excitation-transcription coupling.39 CaMKII activity may also become Ca/CaM autonomous, and such activity is increased in myocardial disease, contributing to apoptosis, arrhythmia, and disrupted excitation-contraction coupling and excitation-transcription coupling, leading to pathological hypertrophy.³⁹ CaMKII and its targets, RyR2 and phospholamban, have also been shown to be critical in cardiac ischemia/reperfusion injury.^40,41 To assess CaMKII activation/autophosphorylation as a potential mediator of cardiomyocyte calcium potentiation in our in vivo cardiac arrest model, we examined the phosphorylation status of CaMKII Thr287 (PT287) and total CaMKII protein in ventricular tissue lysates after cardiac arrest (Figure 6A). After 30 minutes of cardiac arrest and 30 minutes of ECMO reperfusion, the ratio of pCaMKII PT287 to total CaMKII nearly doubled in the PMAD group compared with the control group (Figure 6C), whereas the total CaMKII protein remained unchanged (Figure 6D), indicating an activation of CaMKII. This activation was reduced by 30% in the presence of the CaMKII inhibitor AIP given 30 to 40 minutes before the onset of cardiac arrest (Figure 6C). We also examined the phosphorylation state of RyR2 and phospholamban, the targets of both CaMKII and PKA, at the sarcoplasmic reticulum. CaMKII-dependent phosphorylation of phospholamban at Thr17 (PT17) showed similar patterns (Figure 6E and 6F). The ratio of PT17 to total phospholamban increased nearly 2-fold in the cardiac arrest group compared with the control group, and PT17 phosphorylation was abolished by AIP treatment before cardiac arrest. In contrast, PKA-dependent phosphorylation of phospholamban at Ser16 is not different between the cardiac arrest group and the control group (Figure 6J).

• Download figure
• Download PowerPoint

The most striking change during PMAD was found for RyR2, on the detection of both total protein and phosphorylation at Ser2814 (PS2814; Figure 6A). However, the ratio of PS2814 to total RyR2 did not vary significantly (Figure 6G). Signals for RyR2 protein and PS2814 surged ≈7-fold in PMAD (Figure 6H and 6I) and were not significantly altered by AIP or Alda-1 treatment. In addition, a single band of RyR2 protein was detected in hearts from the control group, whereas a doublet of bands became prominent after cardiac arrest, suggesting some form of oligomerization or protein modification (see Discussion). Because similar calcium potentiation was observed in cardiomyocytes isolated from rats treated with epinephrine and in PMAD (Figure 3B), the phosphorylation status of CaMKII was examined in heart tissues collected after 30 minutes of epinephrine infusion. CaMKII phosphorylation was also increased in this model, but to a lesser extent than seen during cardiac arrest (Figure 6K), leading us to consider the contributory role of oxidative stress (reactive oxygen species) in the effects seen in vivo.

Cardiac Arrest Increased Myocardial O₂^•− Generation Predominantly From Mitochondria

We measured myocardial superoxide (O₂^•−) production using lucigenin-enhanced chemiluminescence. We found it to be elevated after epinephrine infusion but significantly more elevated after PMAD (70% increase in total chemiluminescence compared with control; P<0.001; Figure 7A). When ventricular tissue was incubated with the nitric oxide synthase inhibitor N^G-monomethyl-l-arginine or the NADPH oxidase inhibitor diphenylene iodonium, total O₂^•− levels were not significantly different, suggesting that the major contribution of O₂^•− was mitochondrial. To confirm the chemiluminescence findings and to investigate further the cellular source of O₂^•−, we measured its production using dihydroethidium oxidative fluorescence microtopography (Figure 7B), again finding significant O₂^•− production in PMAD. Finally, to confirm the source of O₂^•−, we used the mitochondria-targeted antioxidant with O₂^•− and alkyl scavenging properties Mito-TEMPO. Mito-TEMPO led to a significant reduction in O₂^•− generation (P<0.01; Figure 7B), confirming the dominant mitochondrial source.

• Download figure
• Download PowerPoint

Alda-1 Reduces Calcium Potentiation and Improves Cardiac Performance in PMAD

Because the augmented CICR triggered by catecholamines and maintained by CaMKII was supportive of cardiac function and survival in the early stages, we looked toward downstream effects of oxygen radicals as a potential therapeutic intervention. Recently, activation of aldehyde dehydrogenase type 2 by Alda-1 has been shown to play a pivotal role in cardioprotection in the setting of myocardial infarction through detoxification of O₂^•−-induced reactive aldehydes (such as 4-hydroxynonenal acetaldehyde) to less reactive acids (such as 4-hydroxynon-2-enoic acid and acetic acid).⁴² We hypothesized the reactive aldehydes were playing an important role in PMAD. In keeping with this hypothesis, we found 4-hydroxynonenal adducts to be significantly increased after cardiac arrest, and administration of Alda-1 in vivo during cardiac arrest significantly decreased both 4-hydroxynonenal adducts (online-only Data Supplement Figure III) and O₂^•− to levels seen after Mito-TEMPO inhibition (P<0.01 versus control; Figure 7A and 7B). Alda-1 also had a dramatic beneficial impact on PMAD both in vitro and in vivo. At the cellular level, Alda-1 treatment blunted the increase in CICR seen in PMAD by 30%, similar to CaMKII inhibition (Figure 3B). However, although myofilament calcium sensitivity was unchanged with CaMKII inhibition, it was enhanced with Alda-1 (Figure 3F). In addition, immunoblotting of CaMKII and autophosphorylated CaMKII revealed that in hearts exposed to Alda-1 during cardiac arrest, T287 autophosphorylation was reduced by 30%, similar to the effect of AIP (Figure 6B and 6C). As expected, CaMKII inhibition had no impact on 4-hydroxynonenal adduct formation, whereas Alda-1 reduced 4-hydroxynonenal adduct formation after cardiac arrest (online-only Data Supplement Figure III). We also found that Alda-1 reduced malondialdehyde fluorescence after myocardial infarction in a separate pig model (data not shown).

Alda-1 Improved Outcome After Cardiac Arrest

We found weaning from resuscitation (defined as reaching 20% of the prearrest mean arterial pressure) was successfully achieved in 5 of 8 animals (62.5%) treated with Alda-1 compared with only 4 of 30 animals (13%) in the absence of Alda-1 (Figure 8D). Moreover, with Alda-1 administration, we found that whole-animal cardiac contractility, as indexed by the preload recruitable stroke work, contractility index, and τ, returned to near control levels (Figure 8A-8C and theTable). Alda-1 had no effect on these same parameters before arrest (data not shown).

• Download figure
• Download PowerPoint

Discussion

For >3 decades, PMAD has been recognized as the critical factor in recovery and survival after cardiac arrest.^5,6 To date, most research focused on ischemia/reperfusion has used variants of a Langendorff isolated heart model. Here, we present an in vivo model of PMAD, which, in leaving the neurohumoral axis intact, uncovered a significant increase in CICR in vivo that continued in isolated cardiac myocytes. A similar increase in CICR was reproduced when blood from an arrested animal was used as perfusate in the Langendorff system. We did not see this potentiation of CICR in the saline-perfused Langendorff model. This cardiac myocyte calcium “memory” seems to be triggered by a combination of catecholamines and reactive oxygen species and maintained by autophosphorylation of CaMKII with subsequent phosphorylation of RyR2 and phospholamban.

The role of CAMKII in cardiomyopathy has been described in seminal investigations.^34,43–45 Its activation is important in the generation of electrical storm^46,47 and in reperfusion-related arrhythmias,⁴⁸ whereas its inhibition can prevent the occurrence of ventricular arrhythmias.⁴⁹ In our study, we found that inhibition of CaMKII also reduced cellular stretch–induced arrhythmias. We propose that oxidative stress and aldehydic adduct formation contribute to acute myofilament dysfunction and impaired myofilament calcium sensitivity, which is partially compensated for by this increase in CICR. Consistent with this hypothesis, the small-molecule activator of aldehyde dehydrogenase type 2, Alda-1, dramatically improved cardiac myofilament calcium sensitivity at the whole-heart and cellular levels by acting upstream of calcium signaling to reduce both mitochondrial O₂^•− production and toxic aldehydic adduct formation. In so doing, myofilament calcium sensitivity after arrest was improved, and autophosphorylation of CaMKII and CaMKII augmentation of CICR were simultaneously reduced. This proposed mechanism (Figure 8E) has similarities to neuronal plasticity in which CaMKII senses phasic changes and then augments the same postsynaptic signals.³⁷ We speculate that, just as in the brain, CaMKII induces long-term potentiation³⁸ in the heart. The unique features of CaMKII autophosphorylation and its ability to follow calcium changes phasically³⁷ allow integration of calcium signals in the “postsynaptic” dyadic cleft.^{34,35,41,43,50–52}

In an attempt to track the timing of signaling events that accompanied the calcium potentiation during cardiac arrest and ECMO reperfusion, we traced the dynamics of CaMKII signaling in our in vivo PMAD model. We varied the time of cardiac arrest and reperfusion/resuscitation and indeed observed variations in the phosphorylation status of CaMKII (PT287) and phospholamban (PT17; online-only Data Supplement Figure IV). It is worth noting that the surge of total RyR2 protein and PS2814 detection were prominent in almost all the heart samples collected after cardiac arrest/reperfusion despite variation in CaMKII activation. It is likely that in addition to CaMKII, RyR2 is responsive to other signaling pathways involved in oxygen homeostasis and return of spontaneous contraction in this cardiac arrest rat model. The overall elevation of RyR2 in its active form (PS2814) would independently increase sarcoplasmic reticulum calcium release and contribute to augmented CICR. The elevation of RyR2 signal seen in the Western blots could indicate a rapid increase in RyR2 protein after cardiac arrest and reperfusion, although it is unclear which signals may lead to such induction. Alternatively, RyR2 proteins may be posttranslationally modified in response to cardiac arrest/reperfusion and become stabilized against degradation. It is also possible that cardiac arrest and oxidative stress resulted in an RyR2 conformational change, which leads to increased antigenicity. The Western blots of RyR2 and PS2814 were performed under nonreducing conditions in which the RyR2 protein could retain some folded structure. It has been reported that oxidative stress can induce conformational changes in RyR2.⁵³ Such conformational changes could make the epitope more accessible to the antibodies and result in higher detectable signals.

The sarcoplasmic reticulum Ca²⁺-ATPase regulator phospholamban is a common target of both CaMKII (PT17) and PKA (phosphorylation at Ser16). Examination of these phosphorylation sites after in vivo cardiac arrest/reperfusion demonstrated clearly that the CaMKII but not PKA pathway was activated after cardiac arrest because the ratio of PT17 to phospholamban surged while the ratio of phosphorylation at Ser16 to phospholamban remained unchanged. Inhibition of CaMKII with AIP before cardiac arrest reduced CaMKII activation and abolished phospholamban PT17 phosphorylation and calcium potentiation. However, in vivo and in vitro contractility was not improved. It is possible that CaMKII-mediated phospholamban PT17 phosphorylation is not contributing to PMAD. A previous study on the phospholamban S16A/T17A (nonphosphorylatable phospholamban) mutant mice showed that a CaMKII-dependent increase of phospholamban PT17 opposes rather than contributes to ischemia/reperfusion damage.⁴⁰ Ideally, genetic models such as phospholamban S16A/S17A mutant mice, CaMKII mutant mice (overexpression and knockouts), and RyR2 S2814D, S2814A mutant mice would be studied in this in vivo cardiac arrest model to define the molecular pathway and to clarify the roles of individual contributors. However, the technical challenge of reproducing cardiac arrest and ECMO resuscitation in the mouse has not yet been met.

Our work has limitations. First, these studies have not yet been performed in humans. Second, we did not explore the role of the action potential because of technical limitations related to photostability of fluorescence voltage dyes. Finally, quantifying CaMKII oxidation was technically challenging and is not reported here.

Conclusions

Using new approaches for simultaneous force and CICR measurement in vitro and GECI enabled calcium fluorescence measurements in vivo, we present a novel mechanism of PMAD (Figure 8E). We propose that hypoxia triggers a surge in both catecholamines and reactive oxygen species that triggers an increase in CICR that is further maintained by autophosphorylation of CaMKII. In addition, stretch of cardiac cells may increase calcium uptake and sarcoplasmic reticulum releasability, aiding CICR.^16,54 This increased CICR supports contraction in the face of reduced myofilament sensitivity. We show that inhibiting aldehydic adduct formation during arrest/reperfusion increases myofilament sensitivity, reduces CaMKII activation, and improves survival. These mechanisms should be explored further as potential targets for human therapeutic trials.

Acknowledgments

The authors thank Dr Peter Lee for his construction of the MATLAB control device for CF stretch. The authors also thank Dr Kathia Zaleta for assisting in experiments.

Footnotes