Myocardial death and dysfunction after ischemia-reperfusion injury require CaMKIIδ oxidation

Reactive oxygen species (ROS) contribute to myocardial death during ischemia-reperfusion (I/R) injury, but detailed knowledge of molecular pathways connecting ROS to cardiac injury is lacking. Activation of the Ca2+/calmodulin-dependent protein kinase II (CaMKIIδ) is implicated in myocardial death, and CaMKII can be activated by ROS (ox-CaMKII) through oxidation of regulatory domain methionines (Met281/282). We examined I/R injury in mice where CaMKIIδ was made resistant to ROS activation by knock-in replacement of regulatory domain methionines with valines (MMVV). We found reduced myocardial death, and improved left ventricular function 24 hours after I/R injury in MMVV in vivo and in vitro compared to WT controls. Loss of ATP sensitive K+ channel (KATP) current contributes to I/R injury, and CaMKII promotes sequestration of KATP from myocardial cell membranes. KATP current density was significantly reduced by H2O2 in WT ventricular myocytes, but not in MMVV, showing ox-CaMKII decreases KATP availability. Taken together, these findings support a view that ox-CaMKII and KATP are components of a signaling axis promoting I/R injury by ROS.

because of greater I KATP in the face of I/R or H 2 O 2 treatment. We and others previously identified threonine 224 on Kir6.2, the KATP pore forming subunit, as a site where CaMKII catalyzed phosphorylation promotes KATP trafficking away from the sarcolemma, causing reduced I KATP , and augmenting myocardial injury and death after I/R injury 18,19 . We developed a new knock-in mouse where Kir6.2 threonine 224 was replaced by alanine (T224A) to explore the potential for this mutation to protect against ox-CaMKII mediated myocardial injury. I KATP in T224A ventricular myocytes was resistant to decreases after H 2 O 2 treatment, as predicted, but unexpectedly I KATP was reduced at baseline compared to WT and MMVV ventricular myocytes, and viability of T224A ventricular myocytes exposed to simulated I/R was reduced. Collectively, our findings support a view that CaMKII is a pathological ROS sensor, and ox-CaMKII reduces I KATP contributing to myocardial death and dysfunction after I/R injury. It is likely that ox-CaMKII affects KATP by actions at more than one site.

ox-CaMKII is essential for myocardial I/R injury in vivo and in vitro.
Excessive activation of CaMKII contributes to I/R injury in animal models [20][21][22] , but the upstream pathways responsible for CaMKII activation in I/R are unknown. Increased oxidant stress follows myocardial reperfusion 4 , so we hypothesized that ox-CaM-KII 7 is important for adverse outcomes in response to I/R. Ox-CaMKII can contribute to cardiovascular 13,14 , pulmonary 23 , and other 24 diseases linked to excessive ROS. We used mice, developed in our laboratory 13,14 , where knock-in replacement of oxidizable regulatory domain methionines (281/282) with valines (MMVV) in CaMKIIδ, the predominant myocardial CaMKII isoform, prevents ox-CaMKII formation 13,14 .
As a first step, we measured area at risk, necrotic, and viable tissue in left ventricular slices 24 hours after I/R surgery. MMVV hearts exhibited less myocardial death compared to WT controls after I/R injury (Fig. 1A,B), despite similar area at risk (Fig. 1C). Twenty-four hours after I/R surgery MMVV mice showed significantly improved left ventricular ejection fractions and fractional shortening, and reduced left ventricular dilation compared to WT mice ( Fig. 1D-F, Supplementary Fig. 1A,B). In contrast, MMVV and WT mice had similar echocardiographic measurements at baseline, prior to surgery ( Fig. 1D-F, Supplementary Fig. 1A-C). We interpreted these data to support the hypothesis that ox-CaMKII was an upstream signal promoting I/R injury by increasing myocardial death, and leading to worsened myocardial function. Although CaMKIIδ is highly represented in myocardium 25 , it is also expressed in other cells and tissues, raising the possibility that the benefits of the MMVV mutation in I/R injury could arise outside of myocardium. In order to focus on the presumed role of ox-CaMKII in myocardium, we challenged isolated ventricular myocytes using a validated in vitro I/R model 26 , by imposing hypoxia with a gas permeability resistant lipid layer followed by removal of the lipid layer and reoxygenation. At baseline, in the absence of I/R conditions, MMVV and WT ventricular myocyte isolation yielded a similar percentage of viable cells (Fig. 1G,H). We found that, similar to our in vivo results, isolated ventricular myocytes from MMVV mice were relatively resistant to I/R injury compared to cardiomyocytes isolated from WT mice (Fig. 1G,H) (Fig. 1I,J). However, MMVV myocytes were protected against death after exposure to 1 mM H 2 O 2 compared to WT (Fig. 1I,J) to a similar extent as was measured in response to in vitro simulated I/R (Fig. 1H). Taken together, the data up to this point were consistent with a model where ROS contributed to myocardial death and dysfunction by a pathway involving ox-CaMKIIδ.
KAtp current is decreased by ox-CaMKII. CaMKII reduces cell membrane expression of KATP channels in cardiomyocytes 17,18 , and preventing loss of KATP channels has been proposed as a mechanism for the beneficial actions of CaMKII inhibition in I/R injury 17,27 . Based on these concepts, we asked if ox-CaMKII could decrease I KATP density in ventricular myocytes. We found that I KATP density was reduced by H 2 O 2 (0.4 mM) in cardiomyocytes isolated from WT ( Fig Fig. 2A inset), and applied pinacidil (0.1 mM) and 2, 4 dinitrophenol (DNP, 0.1 mM) to maximize baseline I KATP (red lines in Fig. 2A for both WT and MMVV representative traces) 28 , and glibenclamide (3-6 µM) to eliminate I KATP (blue lines in Fig. 2A for both WT and MMVV representative traces) 29 . We found that pretreatment of H 2 O 2 significantly reduced I KATP density in ventricular myocytes isolated from WT mice (black line in the left panel of Fig. 2A and summary data in Fig. 2B), but that H 2 O 2 application had almost no effect on I KATP density in MMVV ventricular myocytes (black line in the right panel of Fig. 2A and summary data in Fig. 2B). Data acquired under these experimental conditions showed that increased oxidant stress reduced I KATP and that ox-CaMKII was essential for this action.
We next asked if KATP channel opening probability (Po) was different between WT and MMVV cardiomyocytes. On cell-attached mode voltage clamp recordings showed similar Po and KATP channel density at baseline in ventricular myocytes isolated from WT and MMVV mice (Fig. 2C-E). H 2 O 2 (1 mM) significantly reduced the number of available KATP channels in WT myocyte cell membranes (Fig. 2C,D), however changes in Po after H 2 O 2 treatment were not significant in WT membrane patches (P = 0.23, Fig. 2E). In contrast, H 2 O 2 did not reduce KATP density nor Po in MMVV ventricular myocyte sarcolemma patches ( Fig. 2C-E). We repeated these studies in excised, inside out, cell membrane patches. In contrast to our findings in whole cell mode perforated patch clamp configuration ( Fig. 2A,B) and single channel cell-attached mode recordings (Fig. 2C,D), where cytoplasmic contents were preserved, the KATP channel density and Po were not changed by H 2 O 2 treatment in excised WT and MMVV ventricular myocyte membrane patches ( Fig. 2F-H). This lack of difference in KATP currents in the excised membrane patch recordings before and after H 2 O 2 suggested that KATP channels had similar intrinsic responses to H 2 O 2 , and that ox-CaMKII actions on I KATP in WT ventricular myocytes arose from a cellular function that was lost during membrane patch excision. We next tested the hypothesis that www.nature.com/scientificreports www.nature.com/scientificreports/ H 2 O 2 treatment reduced sarcolemmal KATP channels by endocytosis, using Dynasore, an endocytosis inhibitor. Dynasore (10 μM) eliminated loss of KATP channels by H 2 O 2 in WT myocytes (Fig. 2I), without affecting KATP channel Po (Fig. 2K). These findings support the hypothesis that ox-CaMKII contributes to reduced I KATP by augmenting KATP channel endocytosis.
Knock-in mice lacking Kir6.2 T224. The results up to this point showed that ox-CaMKIIδ was important for reducing KATP channel activity and cell membrane expression. We next turned our attention to threonine 224 on Kir6.2 (T224), the pore forming subunit of KATP, because T224 phosphorylation by CaMKII decreases I KATP by facilitating KATP sequestration from the sarcolemma 18,19 . We used CRISPR/Cas9 technology (see Methods and Supplementary Fig. 3A) to generate a new mouse model with knock-in replacement of T224 with alanine (T224A) (Fig. 3A,B). T224A mice were viable and born in predicted Mendelian ratios. Despite the known role of KATP in insulin secretion 30 3L). We interpreted these data to suggest that constitutive knock-in replacement of T224 protected against ROS induced loss of KATP from sarcolemma, but unexpectedly led to a reduction in basal sarcolemmal KATP expression.

pinacidil protects against cell death after I/R injury in Wt but not MMVV ventricular myocytes.
We next compared survival, before and after I/R injury, in isolated ventricular myocytes at baseline with pinacidil, or glibenclamide (Fig. 4). Pinacidil is a KATP agonist (i.e. 'opener') while glibenclamide is a KATP antagonist. Pinacidil is known to improve while glibenclamide worsens myocardial survival in response to I/R injury 31,32 . In order to test if benefits of KATP activation added to the increased survival in the MMVV cardiomyocytes after I/R injury, we next measured cardiomyocyte viability in response to I/R injury in the presence of pinacidil (10 μM). Pinacidil did not significantly affect cell survival in any of the groups under mock I/R control conditions ( Fig. 4A-D, left panels, middle data set). However, pinacidil significantly increased survival in ventricular myocytes isolated from WT mice after I/R injury (Fig. 4E,G, compare middle and left data sets in each panel). Pinacidil did not significantly change survival after I/R injury in MMVV (Fig. 4F) or T224A ventricular myocytes (Fig. 4H). These data suggested that MMVV improved survival, at least in part, by increasing I KATP because the benefits of MMVV and pinacidil were not additive. Glibenclamide (2-4 μM), a KATP antagonist, tended to reduce the viability of isolated ventricular myocytes compared to pinacidil treated groups in the absence (Fig. 4A-D) or presence ( Fig. 4E-H) of I/R injury (compare left and right data sets in each panel). Taken together, these data supported a concept where ox-CaMKII reduces I KATP leading to cardiomyocyte death after I/R injury (Fig. 5).

Discussion
The role of ROS in promoting I/R injury is widely accepted 33 . However, the complexity of ROS mediated injury presents a major barrier to unambiguous, and precise identification of downstream targets affected by ROS that contribute to pathological consequences of I/R injury. CaMKII has emerged as a ROS-activated signal with the potential to activate pro-death pathways in myocardium in response to myocardial infarction, angiotensin II 7 , and aldosterone 34 . The MMVV CaMKII mutant is resistant to ROS-triggered activation, but retains other wild type attributes, including activation by calcified calmodulin and the capacity to demonstrate Ca 2+ and calmodulin independent activity by threonine 287 autophosphorylation 7 . Thus, the MMVV knock-in mouse model represents an important tool to test for a potential contribution of ox-CaMKII to I/R mediated injury. The protection response of MMVV mice to I/R injury provides new, direct evidence that ox-CaMKII is a critical transduction element, coupling ROS to cardiomyocyte death and myocardial dysfunction. Our data also confirm that ox-CaMKII couples ROS to KATP, and that under pathological conditions ox-CaMKII contributes to cardiomyocyte death, at least in part, by reducing I KATP .
The scale bar is 100 µm. (J) Summary data for cell viability based on percentage of live cells, under different concentrations of H 2 O 2 . Cells are from 2 mice from each group. Data were analyzed with a one way ANOVA test (P < 0.0001). Sidak's multiple comparisons test was used for comparisons between groups. All comparisons between the 1 mM H 2 O 2 and other groups were significant (P < 0.0001). Significantly more MMVV compared to WT ventricular myocytes were viable after 1 mM H 2 O 2 (*p < 0.05). (2019) 9:9291 | https://doi.org/10.1038/s41598-019-45743-6 www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ I/R injury is a pathological insult that activates diverse 'upstream' signals, including increased ROS, with the potential to contribute to myocardial injury, dysfunction, and death 35 . The MMVV mice are incompletely protected from myocardial death and dysfunction in our study, showing that other, ox-CaMKII independent, pathways contribute to myocardial responses to I/R injury. CaMKIIδ is enriched in myocardium, but is also widely expressed in other tissues so ox-CaMKIIδ may contribute to I/R injury by extramyocardial actions, including by activation of inflammatory signaling cascades 36 . Thus, we do not interpret our results to rule out the possibility that ox-CaMKII contributes to I/R survival responses by actions outside of cardiomyocytes. However, given the complexity of I/R injury, the apparent magnitude of the ox-CaMKII contribution to I/R injury is remarkable, but nevertheless consistent with other findings showing MMVV mice resist complex, ROS-associated, triggers for myocardial disease. MMVV knock-in mice are known to harbor disease resistance to alcoholic cardiomyopathy 37 , cardiomyopathy in Duchenne's muscular dystrophy 38 , diabetic cardiomyopathy 13 , and angiotensin II-primed atrial fibrillation 14 . The MMVV cardiomyocytes showed a similar resistance to H 2 O 2 and I/R injury, suggesting the component of I/R injury protected by the MMVV mutation was related to ROS. The MMVV knock-in mutation was made on the CaMKIIδ background, leaving CaMKIIγ, a minor myocardial isoform capable of generating ox-CaMKII. The resistance of MMVV mice to I/R injury appears to confirm earlier findings in CaMKIIδ knock-out mice [39][40][41] , indicating that CaMKIIδ is the major CaMKII isoform contributing to disease in mouse models.
CaMKII is a multifunctional serine threonine kinase with a diverse array of molecular substrates. Regulation of intracellular Ca 2+ is a major focus for CaMKII actions in the cytoplasmic compartment. CaMKII catalyzed phosphorylation of ryanodine receptors 42,43 , voltage-gated Ca 2+ 44,45 , and Na + 46 channels all contribute to increasing cytoplasmic Ca 2+ . CaMKII actions at KATP can likely be considered within the framework of augmenting cytoplasmic Ca 2+ concentration because reduced I KATP has the potential to prolong action potential repolarization, and thereby promote cellular Ca 2+ entry 15 . We speculate that CaMKII mediated increases in intracellular Ca 2+ contribute to myocardial performance under physiological stress, but become maladaptive during pathological stress, leading to Ca 2+ overload and myocardial death.
KATP channels are mostly closed under normal physiological conditions, but KATP channels open under ischemic conditions in response to a local reduction in ATP. There are at least 3 mechanisms by which opening of KATP protects cardiac myocytes from injury and death 15 : 1. Increased K + conductance stabilizes the resting membrane potential, preventing triggered depolarizations that could augment Ca 2+ channel opening; 2. I KATP shortens the action potential plateau, resulting in reduced voltage gated Ca 2+ channel current; 3. KATP opening conserves intracellular energy stores by reducing intracellular Ca 2+ and contraction. Our cellular I/R experimental findings are consistent with in vivo 47,48 and in vitro [49][50][51] studies demonstrating that KATP channel opening protects against ischemic myocyte death.
MMVV and WT Dynasore treated ventricular myocytes were protected against loss of I KATP density after H 2 O 2 , consistent with previous reports that Ca 2+ activation of CaMKII triggered dynamin-dependent internalization of KATP channels. This process required phosphorylation of threonine at 180 and 224 and an intact (330)YSKF(333) endocytosis motif of the KATP channel Kir6.2 pore-forming subunit 18 . In pancreatic β cells, CaMKII-dependent phosphorylation at Kir6.2 T224 reduces KATP cell membrane expression, and is enabled by β IV -spectrin targeting 19 . Loss of KATP in skeletal muscle caused mice to be energetically inefficient, lean, and exhibit poor exercise tolerance 52 . These and other studies 27 , suggested to us that the T224A mutation, which biased against dynamic changes in KATP, should have broad-ranging consequences. However, despite these considerations, the T224A mice showed normal glucose-stimulated insulin secretion, normal glucose, and exercise tolerance similar to WT littermate controls. Isolated cardiomyocytes from T224A mice had reduced basal I KATP , and were not protected from I/R injury or H 2 O 2 . We interpret these data to suggest either that congenital absence of Kir6.2 T224 results in yet unknown compensatory changes, and/or that CaMKII actions at KATP involve multiple sites. I/R injury is a clinically important issue 4,53 . However, I/R models are notoriously misleading, and in some cases have provided proof of concept evidence for various pathways manipulated to reduce myocardial death that ultimately failed to provide anticipated benefits in clinical trials [53][54][55] . Here we used an in vivo I/R model, as a starting point, to avoid many potential pitfalls of less reliable, and potentially less clinically relevant in vitro and ex vivo studies 56 . We also focused on a highly validated pathway for concisely regulating ROS induced tissue injury by replacing CaMKIIδ regulatory domain methionines with valines. Thus, our finding that ox-CaMKII is required for a substantial amount of myocardial death after I/R injury adds to evidence that ox-CaMKII plays an important role in myocardial survival. Because of the consistently disappointing results of clinical trials of anti-oxidant agents for cardiovascular disease 57 , and the apparent protection exhibited by MMVV mice we hypothesize that therapeutic modulation of ROS will require precision targeting of pathological signals downstream to ROS. Ox-CaMKII may be such a signal, but to our knowledge, there have been no clinical trials with a CaMKII inhibitor drug, so our findings await translational testing in patients. openings in each membrane patch under conditions with and without H 2 O 2 , as in Fig. 2D. One way ANOVA and Tukey's multiple comparisons test were used for intergroup comparisons (*p < 0.05, **p < 0.01). (K) Summary data of open probability (NPo) in KATP channels analyzed from cell membrane patches shown in (J). One way ANOVA was used for comparison between all groups (P > 0.05). WT (4-7 cells, 2 mice), T224A (13-23 cells, 4 mice). (L) Expression of Kir6.2 in the heart was measured by RT-qPCR (Normalized against Gapdh, n = 5 for each genotype). www.nature.com/scientificreports www.nature.com/scientificreports/ Methods Animal use. All animal handling procedures were in accordance with National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committees of Johns Hopkins University School of Medicine. Two kinds of Knock-in mice: MMVV and T224A were used in this study. The MMVV mice have been described 13,14 . We developed the T224A mice for this study. All comparisons between knock-in MMVV and T224A and WT used WT littermate controls. Based on our preliminary experimental findings, and documented gender differences in KATP 58 , we used only male mice in this study.  In vivo I/R surgery. Myocardial I/R was performed in vivo 59 by 45 min of ligation of the left anterior descending coronary artery followed by 24 h of reperfusion. Mice were anesthetized with 4% isoflurane for ~2 min and maintained at 1.5-2% isoflurane. Anesthetized mice were intubated and mechanically ventilated at 120 breaths per minute with a tidal volume of 200 µl. Mice are given buprenorphine (0.03-0.07 mg/kg) or Rimadyl (4-5 mg/ kg) subcutaneously for post-operative analgesia. Through a sterile incision, a left thoracotomy and pericardiotomy was performed in the 5 th to 6 th intercostal space. The mice underwent coronary ligation using 7-0 prolene and PE10 tubing as a splint. After 45 minutes ischemia, the splint was removed and the suture released to reperfuse the heart. The chest and skin were closed with 5-0 silk suture. Following I/R, and prior to sacrifice, 1% solution of Evans Blue was injected into the cardiac apex to determine the area at risk or in vitro Langendorff perfusion of the aorta was used to perfuse Evans Blue 59,60 . Hearts were harvested and 1-mm sections of the hearts www.nature.com/scientificreports www.nature.com/scientificreports/ were stained with 1% triphenyl tetrazolium chloride (TTC) to measure the area of necrosis. The area at risk, area of necrosis, and left ventricular area were stored as digital images, and analyzed using ImageJ software. simulated cellular I/R injury. Simulated ischemia was induced by layering mineral oil (0.5 ml for 40 min) over a thin film of media covering the cells followed by 45 min of 'reperfusion' in normal media in a 1.5 ml tube, as described 26 . Ventricular myocytes were enriched in a 25 µl pellet by gently centrifuging the cell suspension of 1.5 ml (30 s, 100 g). After removing the supernatant, we added 0.5 ml mineral oil to the tube to limit atmospheric diffusion to the cells. A control group was treated to the same centrifugation protocol and supernatant removal, but 0.5 ml of Tyrode's solution was replaced without addition of oil. After 40 min the oil was removed and the cells were suspended in Tyrode's solution (mM: NaCl 137, KCl 5.4, CaCl 2 1.2, MgCl 2 2, NaH 2 PO 4 0.33, Glucose 10, HEPES-NaOH 10, pH 7.4) for 45 min. The entire process was conducted in a 37 °C water bath.

Generation of
Cell viability measurements. Dissociated ventricular myocytes were stored in Tyrode's solution at room temperature or at 37 °C, as indicated. We used morphology and Trypan blue (Corning or Gibco) staining to analyze cell viability, using 100 µl of cell suspension and 20 µl 0.4% Trypan blue for 5 min. The number of stained and unstained (round or square) cells were counted as dead cells; viable cells were counted as cells with a rod-shaped morphology and a length/width ratio of more than 3. Cell viability was expressed as the percentage of viable cells amongst total cells. The total count (for one sample point) ranged between 200 ->1000 cells (from counting at least 7 fields of view), and was completed within 15 min to preclude nonspecific uptake of Trypan blue. Cell imaging was performed with a Nikon Eclipse Ti inverted phase contrast microscope at 100X magnification. The images were stored in a computer and cell counting was performed by NIS Elements imaging software (Nikon) and validated by visual inspection of each image. Whole-cell current was recorded at a holding potential of −45 mV. Current-voltage relationships were obtained using a ramp protocol (5 to −100 mV at 25 mV/S, applied every 20 s). Membrane currents were filtered (low-pass Bessel response with a cut-off frequency of -3 dB at 1-2 kHz), digitized at 5 kHz, and stored on a computer hard disk with pCLAMP software (Clampex 10.07, Molecular Devices, Sunnyvale, CA, USA).
Single channel KATP currents were mostly recorded in cell-attached mode using a recording chamber (RC26, Warner, Hamden, CT, USA), bath solution (mM: KCl 150; EGTA 5; HEPES, 10; pH adjusted to 7.2 with KOH), and pipette solution (mM: KCl 150; CaCl 2 , 2; and HEPES 10; pH adjusted to 7.2 with KOH). The use of symmetrical recording solutions (150 mM K + ) resulted in an equilibrium potential for potassium (EK) and a resting membrane potential (Vm) around 0 mV, as determined from the I-V relationship of the KATP channel. All recordings were carried out at room temperature, and all patches were voltage clamped at −60 mV (i.e. with +60 mV pipette potentials). Single-channel currents were recorded with an Axopatch 200B patch-clamp amplifier (Molecular Devices, Sunnyvale, CA, USA), low-pass filtered (3 dB, 2 kHz) and digitized at 20 kHz online using Clampex 10.07 software (Molecular Devices, Sunnyvale, CA, USA) via a 16 bit A/D converter (Digidata 1440 A; Molecular Devices, Sunnyvale, CA, USA). Some single channel recordings were performed in excised cell membrane patches in the same solutions and conditions described for cell-attached mode recordings. echocardiography. In vivo cardiac morphology was assessed by trans-thoracic echocardiography (Vevo 2100, 40 MHz transducer; VisualSonics Inc, Toronto, Canada) in conscious mice. As previously described 61 , the M-mode echocardiogram was acquired from the parasternal long axes view of the left ventricle at the mid-papillary muscles level and at sweep speed of 200 mm/sec. The end-diastolic and end-systolic ventricular volumes (EDV, ESV), were obtained from the two chamber view of the heart in long axis view, using Simpson's method. The stroke volume (SV), and the percent ejection fraction (EF) were automatically calculated by the Vevo 2100 ultrasound system built in software. The studies and analysis were performed by an operator blinded as to the experimental group.
Data analysis and statistics. Comparisons of multiple groups were performed using one way ANOVA followed by Tukey's post-hoc correction or Sidak's multiple comparisons test (GraphPad Prism). Two-group analysis used an unpaired 2-tailed Student's t test. Statistically significant differences (defined as P ≤ 0.05) between genotypes (WT vs. MMVV or T224A) and respective treatment groups are indicated.