Stress Signaling JNK2 Crosstalk with CaMKII Underlies Enhanced Atrial Arrhythmogenesis

Abstract

Rationale

Atrial fibrillation (AF) is the most common arrhythmia and advanced age is an inevitable and predominant AF risk factor. However, the mechanisms that couple aging and AF propensity remain unclear, making targeted therapeutic interventions unattainable.

Objective

To explore the functional role of an important stress-response c-Jun N-terminal kinase (JNK) in sarcoplasmic reticulum (SR) Ca²⁺ handling and consequently Ca^2+-mediated atrial arrhythmias.

Methods and Results

We employed a series of cutting-edge electrophysiological and molecular techniques, exploited the power of transgenic mouse models to detail the molecular mechanism, and verified its clinical applicability in parallel studies on donor human hearts. We discovered that significantly increased activity of the stress-response kinase JNK isoform 2 (JNK2) in the aged atria is involved in arrhythmic remodeling. The JNK-driven atrial pro-arrhythmic mechanism is supported by a pathway linking JNK, CaMKII, and SR Ca²⁺ release ryanodine receptor (RyR) channels. JNK2 activates CaMKII, a critical pro-arrhythmic molecule in cardiac muscle. In turn, activated CaMKII up-regulates diastolic SR Ca²⁺ leak mediated by RyR channels. This leads to aberrant intracellular Ca²⁺ waves and enhanced AF propensity. In contrast, this mechanism is absent in young atria. In JNK challenged animal models, this is eliminated by JNK2 ablation or CaMKII inhibition.

Conclusions

We have identified JNK2-driven CaMKII activation as a novel mode of kinase crosstalk and a causal factor in atrial arrhythmic remodeling, making JNK2 a compelling new therapeutic target for AF prevention and/or treatment.

Introduction

Atrial fibrillation (AF) brings a high risk of mortality and associated morbidities, including stroke and heart failure (HF).^1-3 Considering our growing elderly population, AF is becoming an enormous public health challenge and up to 10-15% of 70-80 year olds will develop AF.^2-4 Current pharmacological AF treatment and prevention approaches are ineffective and our understanding of the underlying mechanisms is in complete, limiting the development of new therapeutic strategies.

c-Jun N-terminal kinase (JNK) is a well-characterized stress-response kinase that is activated in response to various cellular stresses such as UV light, ischemia, inflammatory cytokines, and aging.^5-9 Interestingly, many of these stresses are also well-established cardiovascular risk factors and JNK activation has been observed in cardiovascular conditions like ischemia, hypertrophy and HF. All these conditions are associated with increased AF risk.^{3, 5-8, 10-14} Not all hearts will experience a particular stress but all hearts will inevitably age. And, the aged heart is more susceptible to some of the stress stimuli.¹⁵ Thus, age-associated JNK activation is a compelling experimental focus here. In the human atria, AF often involves complex pathological remodeling due to coexisting cardiovascular diseases with increasing age. Here, our focus is to explore the role of JNK activation in enhanced AF susceptibility in the aged heart lacking comorbid conditions (i.e. with normal cardiac function, no AF history, no history of major cardiovascular diseases, and lacking structural remodeling). Once one specific pathogenic pathway is established, defining and understanding the contributions of all the other AF-associated circumstances, stresses and factors becomes possible.

JNK2 is a major isoform in the heart.¹⁶ Our results reveal that JNK2 activation is markedly elevated in aged human, rabbit and mouse atria. Further, this age-associated JNK2 activation causes abnormal intracellular calcium (Ca²⁺) waves and diastolic sarcoplasmic reticulum (SR) Ca²⁺ leak. It is well known that the action of CaMKII on SR Ca²⁺ mishandling (i.e. diastolic Ca²⁺ leak and waves) is pro-arrhythmic.^17-22 Indeed, CaMKII inhibition has been considered as a potential anti-arrhythmic intervention for HF.²³ Here, we discovered the JNK2-CaMKII crosstalk as a previously unknown molecular mechanism of CaMKII activation. Activated JNK2 directly phosphorylates CaMKII proteins, driving CaMKII's pro-arrhythmic effects on diastolic SR Ca²⁺ handling. In aged mice, JNK2 inhibition eliminated age-related CaMKII hyper-activation and the associated aberrant diastolic SR Ca²⁺ leak, Ca²⁺ waves, and AF susceptibility. These results identify JNK2 inhibition as a potential target in developing new therapeutic strategies to prevent or treat AF.

Methods

All data and supporting materials have been provided with the published article. An expanded Methods Section is available in the Supplemental Material.

Animal and cell models

Wild-type (WT) C57B/6j male mice (Jackson Laboratory, ME) at 24-32 months (aged) and 2-2.5 months (young) were studied. Three mouse models were used to assess the contribution of JNK and CaMKII on Ca²⁺ dynamics and AF genesis. They were: 1) a cardiac specific inducible MKK7D Tg mouse strain (a generous gift from Dr. Yibin Wang, UCLA) that can express cardiac MKK7D to robustly activate JNK with tamoxifen treatment;²⁴ 2) homozygous JNK2 knockout mice²⁵ (JNK2KO [Mapk9tm1Flv/J]; Jackson Laboratory); and 3) AC3-I mice,²¹ with cardiac-specific transgenic expression of a CaMKII peptide inhibitor. JNK2KO and AC3-I mice were treated with anisomycin as previously described.^{8, 26} Young (6 months) and aged (60 months) New Zealand White male rabbits were also used. Four young rabbits were treated with a JNK activator anisomycin.²⁷ Rabbit atrial myocytes were isolated as previously described with modification.¹⁷ All animal studies followed the Guide for the Care and Use of Laboratory Animals (NIH Publication, 8th Edition, 2011) and were approved by the Institutional Animal Care and Use Committees of Rush University Medical Center (RUMC), Loyola University Chicago (LUC), University of Alabama at Birmingham (UAB), and University of Iowa.

A well-characterized cultured myocyte line (HL-1, from Dr. William Claycomb, Louisiana State University) was used for our studies as previously described.^{7, 8, 26} Cells were treated with JNK activator anisomycin or infected with adenoviral JNK activator MKK7D to activate JNK in the presence or absence of JNK or CaMKII inhibition as described in the Supplement Methods.

Human atrial specimens

Human right atrial tissues were obtained from donor hearts (Online Table I) provided by Illinois Gift of Hope Organ & Tissue Donor Network (GOH) and Alabama Organ Center (AOC). The studies were approved by the Human Study Committees of RUMC, LUC, UAB, AOC and Illinois GOH.

Atrial arrhythmia induction in mice in vivo and human atrial preparation ex vivo

In vivo AF induction was conducted in sedated mice and electrogram data were recorded using a 1.1F octapolar catheter inserted into the right atrium as previously described.¹⁹ ex vivo AF induction in human donor atria was performed as described in the Supplemental Methods.

Confocal Ca²⁺ imaging

Confocal line scanning Ca²⁺ images were obtained from intact atrium or atrial myocytes as previously described.²⁸ Frequencies of Ca²⁺ waves/sparks and time constant (τ) of Ca²⁺ decay were analyzed from intrinsic sinus rhythm and recovered beats after the burst pacing (5-10Hz) as previously described.^28-30 Tetracaine-sensitive SR Ca²⁺ leak was measured using our well-established protocols as previously described.^{17, 29, 31}

Intact atrial optical mapping

Intact mouse hearts were pre-loaded with Rhod2-AM (5μM) followed by Rh237 (10mM; Invitrogen). V_m and Ca²⁺ signals were simultaneously recorded using a dual channel optical imaging system as previously described.^{7, 26} The standard deviation of the difference between the activation time of action potential and calcium transient (Δt_Vm-Ca) for a total of 400 channels within the mapping field was calculated to reflect the heterogeneity of the relationship between V_m and Ca²⁺ signals as previously described.³²

Single RyR2 recording

ingle RyR channel function was measured by fusing isolated WT mouse SR vesicles into lipid bilayers as previously described.³³ Anisomycin, alkaline phosphatase,¹⁷ the CaMKII inhibitors KN93¹⁷ & AIP and/or the JNK2 specific inhibitor JNK2I-IX were applied to the cytosolic side of the reconstituted RyR2 channels.

CaMKII activity biosensor imaging

The adenoviral mutant variant CaMKII sensor, Camui-vv (lacking the oxidation M280/M281 site but containing the intact autophosphorylation Thr286 site), was used to quantify the contribution of CaMKII phosphorylation and oxidation on CaMKII activation in anisomycin-treated isolated rabbit myocytes as previously described.³⁴

Biochemical assays

Immunoblotting was performed as previously described.^{8, 17} Protein expressions were detected using specific antibodies and JNK2 immunoprecipitation (IP) was also conducted using a JNK2 specific antibody as previously described.¹⁷ HA-tagged WT CaMKII-WT and mutant CaMKII-T286A vectors were constructed as previously described to determine the direct action of JNK2 on CaMKII phosphorylation, detected by immunoblotting and ADP-Glo™ Kinase assay (Promega), per manufacturer's instructions.³⁴

Statistical analysis

All data are presented as mean ± SEM. Differences between multiple groups or any two groups were evaluated using one-way ANOVA with post-hoc Tukey's test or Student's t-test. When heterogeneity of variance was evidenced, a nonparametric Mann-Whitney test or nonparametric one-way ANOVA was performed. A p-value <0.05 was considered to be significant.

Results

JNK activation and AF susceptibility

As age increased, human atria showed markedly enhanced activation of JNK (phosphorylated JNK, JNK-P), as assessed by immunoblotting (Fig. 1A). This trend persisted when JNK-P was normalized to either unchanged total JNK2 proteins (Fig. 1B, & Online_Figs. Ia-Ib) or JNK1 (data not shown). While we found unchanged total JNK proteins, JNK1 and JNK2 mRNA expression were also unchanged in human atria with increasing age (Online_Figs. Ic-Id). The human atria were obtained from hearts of donors without a history of AF or major cardiovascular disease (Online_Table I). We also measured AF inducibility in Langendorff-perfused human donor heart atrial preparations challenged with electrical burst pacing. Three out of four aged donor hearts showed pacing induced AF events after a train of electrical stimulation. Fig. 1C shows pacing induced AF in two aged atrial preps subjected to 1Hz and 3 Hz pacing, respectively (30s, 2× diastolic threshold; Fig. 1C, long red arrows; n=4). In contrast, no pacing induced AF events were found in any of the four young controls (even at a higher pacing frequency 4Hz; Fig. 1D; n=4).

Figure 1. Activated JNK in human atria is associated with increasing age and arrhythmogenicity.

A ) Representative immunoblotting images and plot showing enhanced phosphorylated JNK (JNK-P) in aged human atria. B) Similar increase in the ratio of JNK-P to total JNK2 proteins. C) Representative electrogram (EG) traces of burst pacing induced AF (after 1Hz or 3 Hz burst pacing) in two normal aged human hearts without a history of AF or coexisting cardiac diseases. D) Representative EG trace after 4Hz burst pacing in a young healthy human donor heart.

Like human atria, aged WT mouse atria also had elevated JNK-P (Fig. 2A) and this was associated with a dramatically increased propensity for pacing-induced AF¹⁹ (Figs. 2B-2D) compared to that of young controls. These aged mice showed preserved cardiac function and an unchanged amount of atrial interstitial fibrosis (quantified using Trichrome staining as previously described;³⁰ Online_Figs. IIa-IIb) compared to young controls. A JNK2 specific inhibitor (JNK2I-IX with no action on JNK1 or other MAP kinases³⁵) abolished pacing-induced AF when applied to WT aged mice (Fig. 2B, right bar). When induced with tamoxifen, cardiac-specific MKK7D transgenic mice²⁴ expressed constitutively active MKK7, an upstream JNK activator.⁵ In these mice, JNK activation was significantly increased by tamoxifen treatment (Fig. 2E). The cardiac function of the MKK7D mice was the same before and 5 days after treatment (Online_Fig. IIc). In young MKK7D mice, induction of JNK activation resulted in an increased incidence of pacing induced AF (n=4/5 vs 0/6 tamoxifen-treated WT-littermates; Figs. 2F-2G). Thus, JNK activation promotes AF initiation in young mice similar to the action of age-associated JNK activation in aged WT mice.

Figure 2. Activated JNK enhances atrial arrhythmogenicity in aged mice and cardiac-specific JNK activated young mice.

A ) Immunoblotting images and quantitative data showing increased phosphorylated JNK (JNK-P; activated JNK) in aged mouse atria. B) Summarized data of increased AF inducibility in aged WT mice compared to that of young (Yg) controls. Summarized data showing JNK2 specific inhibitor (JNK2I-XI) in vivo treatment in aged WT mice strikingly attenuated pacing induced AF events (far right bar) as seen in untreated aged WT mice. C-D) Representative intra-cardiac electrogram (EG) traces of burst pacing induced AF in an aged mouse (C), while no arrhythmia was induced in a wild-type (WT) young mouse (D). E) Representative images and summarized quantitative immunoblotting results showing increased JNK-P in cardiac specific tamoxifen-treated (Tamx; one dose per day for 5 days) MKK7D mouse atria (MKK7D⁺) compared to tamoxifen-treated WT littermates (MKK7D^-). F-G) Summarized AF inducibility and representative EG trace of burst pacing induced AF in a tamoxifen-treated MKK7D mouse. Lower panel shows a trace of simultaneously recorded surface ECG. SRh=sinus rhythm.

JNK activation and abnormal Ca²⁺ activities

Confocal Ca²⁺ imaging in Langendorff-perfused intact aged WT mouse atria revealed that there were frequent Ca²⁺ waves during the intrinsic sinus rhythm, but even more waves were evidenced after a bout of rapid electrical pacing (Figs. 3A-3B). The Ca²⁺ transient decay time-constant (τ) was also significantly prolonged in aged WT atria (Fig. 3B, lower panel) compared to young controls. Simultaneous optical recordings of intracellular Ca²⁺ and V_m showed significantly increased spatiotemporal heterogeneity between the two signals (Δt_Vm-Ca) in aged WT atria compared to young controls (Online_Figs. IIIa-IIIb). This increased heterogeneity of Δt_Vm-Ca in the aged atrium is arrhythmogenic³⁶ and aligned with the higher frequency of diastolic Ca²⁺ waves (Figs. 3A-3B).

Figure 3. Activated JNK enhances abnormal Ca²⁺ activities in intact aged mouse atria and cardiac-specific JNK activated young mouse atria.

A ) Representative confocal images of Ca²⁺ waves and Ca²⁺ transients after burst pacing in Langendorff-perfused aged intact mouse atria and restored sinus Ca²⁺ transients after burst-pacing in sham WT intact mouse atria. B) Summarized bar graphs showing significantly increased frequency of spontaneous (sinus rhythm (SRh) before bust pacing) and pacing-induced Ca²⁺ waves along with prolonged relaxation time of Ca²⁺ transients during the recovery period (when burst pacing was stopped) in aged mouse atria compared to that of WT young controls. C) Representative electrogram traces of burst pacing induced AF in anisomycin-treated (Aniso) WT mice and no arrhythmia induced in anisomycin-treated JNK2 knockout (JNK2KO) mice. D) Summarized data of AF inducibility in anisomycin-treated WT and JNK2KO mice vs sham controls. E) Example confocal images of increased Ca²⁺ sparks & waves in anisomycin(Aniso)-treated WT mouse atria after burst pacing (upper), and restored sinus Ca²⁺ transients after burst pacing in anisomycin-treated JNK2KO mouse atria (lower). F) Summarized data showing significantly increased Ca²⁺ waves and prolonged τ of Ca²⁺ decay during sinus rhythm (SRh) and in response to burst-pacing in anisomycin(A)-treated WT young mouse atria compared to anisomycin-treated JNK2KO young mouse atria.

The contribution of JNK activation to the abnormal arrhythmogenic Ca²⁺ handling in the absence of aging was explored by treating young WT mice with a JNK activator, anisomycin.^{7, 8} In young WT hearts, anisomycin treatment dramatically increased pacing-induced AF (Figs. 3C-3D). It also increased spontaneous and/or post-pacing Ca²⁺ waves, while Ca²⁺ transient decay was prolonged (vs sham-controls; Figs. 3E-3F). This is likely the same action of age-associated JNK activation in WT aged atria (Figs. 2A-2D & 3A-3B). To further assess JNK specific action of anisomycin, we applied this agent to JNK2 knockout (JNK2KO) mice. We found that anisomycin-treated JNK2KO mice did not show aberrant Ca²⁺ handling or pacing-induced AF (Figs. 3C-3F). Note that aged and anisomycin-treated young mouse atria showed unchanged amounts of JNK2 and JNK1 proteins (Online_Figs. IIIc-IIId). The JNK2KO atria had the normal JNK1 but only trace amounts of JNK2 (Online_Figs. IIIc-IIId). These results indicate that there is a JNK2 specific action on aberrant Ca²⁺ activities and AF, suggesting JNK2 is a critical determinant of abnormal events.

JNK activation and abnormal diastolic Ca²⁺ handling

We and others have previously shown that increased diastolic SR Ca²⁺ release causes abnormal ectopic activities, which can lead to arrhythmogenesis in diseased hearts.^{17, 37, 38} We monitored diastolic SR Ca²⁺ release using the tetracaine-sensitive SR Ca²⁺ leak protocol.¹⁷ SR Ca²⁺ leak was elevated in aged mouse atrial myocytes (Fig. 4A) compared to young controls. The JNK2 inhibitor JNK2I-IX completely abolished this elevation. Confluent monolayers of HL-1 myocytes recapitulate in situ many genotypic and electrical cardiac phenotypes, including the cell-cell interactions present in the heart.^{8, 26} Intracellular Ca²⁺ handling in HL-1 myocytes was measured (Online_Fig. IVb-IVc). Anisomycin-treated HL-1 myocytes had significantly increased tetracaine-sensitive SR Ca²⁺ leak that was eliminated by JNK2 inhibition (Figs. 4B-4C). This is consistent with our results from freshly isolated aged mouse myocytes (Fig. 4A). For confirmation in a larger animal model, JNK2 action on SR Ca²⁺ leak was also demonstrated in anisomycin-treated (24hr) atrial myocytes isolated from young rabbits (Online_Fig. IVd). Together, our results indicate JNK activation drives abnormal diastolic SR Ca²⁺ leak.

Figure 4. Activated JNK2 causes markedly increased diastolic SR Ca²⁺ leak via increased probability of RyR single channel opening (P_o).

A ) Summarized data showing increased diastolic SR Ca²⁺ leak in freshly isolated aged mouse myocytes; JNK2 specific inhibitor JNK2I-IX (JNK2I) treatment completely reversed the Ca²⁺ leak. B) anisomycin-treated (Aniso or A) HL-1 myocytes also showed dramatically increased diastolic SR Ca²⁺ leak; JNK2 specific inhibition completely prevented these anisomycin actions. C) Example traces of Aniso-treated vs sham control (Ctl) HL-1 myocytes in the tetracaine-sensitive leak confocal measurement protocol. D-E) Summarized data showing increased SR Ca²⁺ content (examined with caffeine-induced Ca²⁺-released) in aged mouse myocytes and anisomycin-treated HL-1 myocytes, which is reversed by JNK2I treatment. F-G) Unchanged Ca²⁺ transient amplitude (F) but prolonged τ of Ca²⁺ decay (G; which is also reversed by JNK2I treatment) in Aniso (A)-treated HL-1 myocytes. H-I) Summarized data shows that overexpression of inactivated JNK2dn proteins attenuates anisomycin (A)-induced SR diastolic Ca²⁺ leak (H) and overload (I), while inactivated JNK1dn has no such rescue effects. J) Ca²⁺ transient amplitude is unchanged in all groups. K) Sample single WT mouse RyR2 channel recordings before (control) and after cytosolic addition of 50ng/ml anisomycin (without and with pre-treatment of JNK2I-IX). The zero current levels are indicated by a dash. L) The mean single RyR2 P_o after anisomycin treatment alone (open square), and with KN93 present (filled triangle) or with alkaline phosphatase present (Alk.Ph., filled square) or with JNK2I-IX present (open triangle) are shown in the inset. Filled circle is WT RyR2 data. Anisomycin action on RyR2 cytosolic Ca²⁺ sensitivity is also illustrated (inset)

In addition, JNK activation also increased SR Ca²⁺ content in both freshly isolated aged mouse atrial myocytes and anisomycin-treated cultured HL-1 myocytes compared to controls. This SR Ca²⁺ overload was eliminated by JNK2 inhibition using JNK2I-IX (Figs. 4D-4E). The systolic Ca²⁺ transient amplitude was not changed by JNK activation (Fig. 4F). However, JNK activation prolonged the Ca²⁺ transient decay time-constant, but not if JNK2 was inhibited (Fig. 4G). We also assessed the contribution of NCX by measuring the Ca²⁺ decay rate of caffeine-induced Ca²⁺ transients. This decay rate was not altered by JNK activation (Online_Fig. IVe). To test JNK2 isoform specificity further, we applied an adenoviral dominant negative approach to HL-1 myocytes (Online_Figs. Va-Vd). JNK2dn abolished the anisomycin-evoked Ca²⁺ mishandling phenotype while JNK1dn did not (Figs. 4H-4J). This confirms that the JNK action on diastolic SR Ca²⁺ leak and overload is indeed JNK2 specific.

Elevated SR Ca²⁺ diastolic leak implies dysfunction of the SR Ca²⁺ release channels (ryanodine receptor, RyR2).³⁹ Single RyR2 channel function was assessed by fusing heavy mouse SR microsomes into artificial lipid bilayers.⁴⁰ Anisomycin application significantly increased RyR2 open probability (P_o; RyR2 cytosolic Ca²⁺ sensitivity) but not when a JNK2 inhibitor was present (Figs. 4K-4L). The anisomycin action on P_o was also prevented when either alkaline phosphatases or CaMKII inhibitor KN93 were present (Fig. 4L & Online_Fig. VIa). The anisomycin action on RyR2 P_o was also confirmed in human RyR channels (Online_Fig. VIb). The JNK associated action on RyR2 P_o explains the increased diastolic SR Ca²⁺ leak observed in atrial myocytes following JNK2 activation. The KN93 result simply CaMKII has a role in linking JNK activation and RyR dysfunction. Interestingly, no JNK or CaMKII were added, so the isolated RyR2s must have been associated with endogenous JNK and CaMKII. This was verified by measuring the component proteins of heavy SR microsomes (Online_Fig. VIc). An intriguing possibility is that JNK activation may promote CaMKII-dependent phosphorylation of RyR2.

JNK2 and CaMKII crosstalk in abnormal Ca²⁺ activities and AF susceptibility

CaMKII activation results in the phosphorylation of various SR Ca²⁺ handling proteins (e.g. RyR2 and PLB) and is pro-arrhythmic in diseased hearts.¹⁷ Blocking of CaMKII function appears to interrupt JNK-to-RyR2 signaling (see above results). Immunoblotting revealed that activation of both CaMKII and JNK (i.e. CaMKII-P & JNK-P) increased with age in human atria (Fig. 5A). This trend persisted when CaMKII-P was normalized to total CaMKIIδ proteins (Online_Fig. Ib). This JNK-CaMKII relationship was also present in aged rabbit atria and anisomycin-treated young rabbit atria (Figs. 5B-5C). The CaMKII-dependent phosphorylation of the RyR2 (RyR2815-P) and PLB (PLB17-P) proteins were assessed. Aged rabbit atria & young anisomycin treated rabbit atria had elevated CaMKII-P and JNK-P levels as well as increased RyR2815-P (normalized to total RyR2) and PLB17-P levels (normalized to SERCA2; Figs. 5B-5C). In contrast, NCX and SERCA were unaltered and there was no change in protein kinase A (PKA)-dependent RyR2 and PLB phosphorylation (RyR2809-P and PLB16-P; Online_Figs. VIIb-VIIc). We did analogous testing in mouse models. A similar outcome was obtained in tamoxifen-induced MKK7D mouse atria with its constitutive JNK activation (Online_Fig. VIIa). In JNK2KO young atria, anisomycin-treatment did not enhance CaMKII-P or increase the levels of RyR2815-P and PLB17-P (Figs. 5D-5F). Also, the PKA-mediated phosphorylation of RyR2809 and PLB16 remained at a level comparable to sham-controls and anisomycin-treated WT mouse atria (Figs. 5E-5F). These control atria had similar JNK2 protein levels. JNK2KO atria had a normal JNK1 level but only trace amounts of JNK2 (Online_Figs. IIIc-IIId). In aged JNK2KO mouse atria with a long-term JNK2 ablation, there was dramatically reduced CaMKII activation (Fig. 6A). Likewise, there was dramatically reduced CaMKII activation in aged WT atria that were treated with JNK2I-IX for ten days (Fig. 6B). These results indicate age with its associated atrial JNK2 activation, results in phosphorylation of the CaMKII protein and in turn CaMKII-dependent phosphorylation of RyR2 and PLB.

Figure 5. Activated JNK2 leads to enhanced CaMKII activation and promotes CaMKII-dependent phosphorylation of SR Ca²⁺ handling proteins.

A ) Example images and pooled immunoblotting data showing that significantly increased CaMKII-P is positively correlated with enhanced JNK activation in human atria with increasing age. B-C) Representative images and summarized data showing that markedly increased CaMKII-P in JNK activated atria from aged rabbits and anisomycin-treated young rabbits. And, this is linked to enhanced CaMKII-dependent phosphorylation of RyR2815 (RyR2815-P) and PLB17 (PLB17-P). D) Immunoblotting results suggest that enhanced JNK-P is associated with enhanced CaMKII, but CaMKII activation in JNK2KO mice treated with anisomycin (Aniso) is significantly reduced compared to WT mice treated with Aniso. E-F) JNK2KO mouse atria attenuated CaMKII-dependent phosphorylation of RyR2815 and PLB17 (RyR2815-P, PLB17-P) compared to that of anisomycin-treated WT mice, while RyR2809 and PLB16 phosphorylation levels (RyR2809-P, PLB16-P) were unchanged.

Figure 6. JNK2 enhances CaMKII activation and CaMKII inhibition prevented JNK-induced arrhythmic activities.

A ) Representative images and summarized immunoblotting data showing increased CaMKII-P along with enhanced activation of JNK (JNK-P) in aged WT mouse atria. Dramatically reduced cardiac CaMKII activation in aged JNK2KO mice compared to that of WT aged mice. B) Summarized data showing JNK2 inhibition (in vivo treatment) reversed the CaMKII activation to the baseline of young hearts compared to markedly increased CaMKII-P in untreated aged mice. C) Representative electrograms of burst-pacing followed by self-reversion to sinus rhythm (no arrhythmia induced) in anisomycin-treated (Aniso) AC3-I mice and AC3-I-sham control mice (n=0/6, 0/6). This suggests that CaMKII inhibition in AC3-I mice prevents Aniso-induced atrial arrhythmias. D) Summarized data suggest that CaMKII inhibition in AC3-I mice completely abolished Aniso-induced aberrant atrial Ca²⁺ waves and prolonged τ of Ca²⁺ decay. E) Immunoblotting images of attenuated activated CaMKII in AC3-I mice, while Aniso-induced JNK activation remains increased. F) Summarized data of diastolic SR Ca²⁺ leak in KN93+Aniso (KN93+A) treated as well as KN92+Aniso (KN92+A) treated HL-1 myocytes compared to sham controls (sham).

We also explored the JNK/CaMKII interaction in AC3-I mice, which have cardiac-specific expression of a CaMKII peptide inhibitor.²¹ In young AC3-I atria, anisomycin-treatment did not enhance AF susceptibility or abnormal Ca²⁺ waves in AC3-I atria (Figs. 6C-6D). These AC3-I atria had substantially activated JNK but little (if any) CaMKII activation (Fig. 6E). Together, these results imply CaMKII activation may be required for JNK-driven arrhythmogenic activities. To test this, we pretreated HL-1 myocytes with the CaMKII inhibitor KN93 or its inactive congener (KN92). The CaMKII inhibition (KN93) eliminated the expected increase in anisomycin-evoked diastolic Ca²⁺ leak and Ca²⁺ transient decay time constant (Fig. 6F and Online_Fig. VIIIa). In contrast, the KN92 pre-treatment (leaving CaMKII function intact) did neither. In HL-1 myocytes, JNK2 inhibition, using either a dominant negative approach or the JNK2 inhibitor JNKK2I-IX, substantially reduced anisomycin-driven CaMKII activation in HL-1 myocytes (Figs. 7A-7C). Moreover, co-infection of HL-1 myocytes with AdJNK2 and constitutively activated AdMKK7D to increase JNK2 activation levels, resulted in significantly increased CaMKII-P (Online_Fig. VIIIb). In contrast, co-infected AdJNK1 and AdMKK7D did not alter CaMKII-P levels. Evidence of successful adenoviral infection is shown in Online_Fig. Ve. The JNK2 isoform-specificaction on CaMKII activation was also confirmed in isolated young rabbit atrial myocytes (Online_Fig. VIIIc).

Figure 7. JNK2 activates CaMKII via direct phosphorylation of CaMKII proteins.

A-B) Immunoblotting images of CaMKII-P and JNK-P in Aniso-treated HL-1 myocytes with and without overexpressed inactivated JNK2dn or JNK2 inhibitor pretreatment. Bottom panel shows positive HA signals in HA-tagged AdJNK2dn-infected cells as evidence of successfully overexpressed JNK2dn proteins. C) Summarized data showing increased CaMKII activation in Aniso-treated HL-1 cells while overexpressing AdJNK2dn or JNK2I-treatment reversed CaMKII activation to the control level. D) Immunoblotting images of co-immunoprecipitated CaMKII with JNK2 specific antibody in human atrial homogenates. E-F) Blotting images of active human JNK2 (hJNK2) in phosphorylation of CaMKII in both human atrial tissue homogenates (E; in contrast to the even amount of alpha-actinin and GAPDH proteins) and pure human CaMKIIδ (hCaMKIIδ) proteins (F). Each assay was repeated at least three times. G) Immunoblotting images showing increased phosphorylation of HA-IPed CaMKII-WT proteins but not HA-IPed CaMKII-T286A mutant (Mu) proteins compared to empty vector (V) controls (three experimental repeats). Ponceau staining shows equal expression between CaMKII-WT and mutant CaMKII-T286A samples. H) Summarized data of increased ADP production from CaMKII phosphorylation by pure active hJNK2 proteins in HA-IPed CaMKII-WT samples but not HA-IPed mutant CaMKII-T286A compared to CaMKII-WT sham-controls without pure hJNK2 incubation. I) Summarized data showing increased ratio of CFP-camui-vv to FRET-camui-vv fluorescence signals in anisomycin-treated isolated rabbit atrial myocytes vs sham-controls.

JNK2 activates CaMKII

Next, we found that a JNK2 specific antibody pulled down the CaMKII protein from human atrial homogenates (Fig. 7D), where the CaMKII was phosphorylated in a JNK2 dose-dependent manner (Fig. 7E). The added JNK2 was the pure active human full-length JNK2 (hJNK) protein. These hJNK2 proteins were incubated with pure full-length human CaMKII proteins (hCaMKII; without Ca²⁺/calmodulin present) and this also resulted in JNK2 dose-dependent phosphorylation of hCaMKII (Fig. 7F). To further explore the JNK2-CaMKII interaction, we transfected HEK293 cells with CaMKII-WT or CaMKII-T286A HA-tagged vectors. The expressed CaMKII proteins were IPed using an HA antibody and then incubated with hJNK2. The CaMKII-WT, not the CaMKII-T286A, protein was phosphorylated by the hJNK2 (Fig. 7G). The ADP-Glo™ kinase phosphorylation assay was used to assess ATP consumption during the phosphorylation reaction. A significant level of hJNK2 phosphorylation of CaMKII-WT, not CaMKII-T286A, was detected (Fig. 7H). While we cannot exclude the possibility of JNK action on other phosphorylation sites of CaMKII, our results indicate that JNK2 directly activates the CaMKII protein.

Aging increases oxidative stress^{11, 41} and this can also activate CaMKII.⁴² To test the contribution of oxidation to CaMKII activation, we used a mutated CaMKII activity biosensor called camui-vv that lacks the oxidation sensitive Met280/281 sites but still has the key Thr286 autophosphorylation site.³⁴ We measured camui-vv FRET in anisomycin-treated (24hrs) rabbit atrial myocytes and found significantly increased CaMKII activity (Fig. 7I). To corroborate this result, we pre-treated HL-1 myocytes with anti-oxidant N-acetyl-L-cysteine (NAC) and found that this did not alter anisomycin enhanced CaMKII-P status (Online_Fig. VIIId). These results suggest the JNK2-drivenincrease in CaMKII activation can occur in the absence of CaMKII oxidation.

Discussion

We discovered that JNK signaling is associated with abnormal Ca²⁺ activities and enhanced AF propensity in the aged heart. Our finding reveals for the first time that the JNK2 isoform directly activates CaMKII. The activated CaMKII in turn enhances RyR2-mediated SR Ca²⁺ leak, enhancing AF propensity (Fig. 8). This is very likely an important pathway linking cellular stresses (e.g. aging) to abnormal diastolic Ca²⁺ activities.

Figure 8. Schematic diagram of proposed mechanism of JNK2-driven CaMKII activation that in turn promotes aberrant SR Ca²⁺ leak triggered arrhythmic activities.

JNK activation has been observed in various cardiovascular diseases associated with a dramatically increased AF propensity, including myocardial infarction (MI) and HF.^{5, 8, 10} AF, MI and HF frequently occur together in the aging population.^2-4 However, the underlying mechanisms of AF arrhythmogenic substrate development in aged (but otherwise normal) hearts remain incompletely understood. Our current study revealed a link between JNK activation and enhanced arrhythmic susceptibility in human atria with increasing age but preserved cardiac function and no history of AF or any major cardiovascular diseases. Moreover, the anti-arrhythmic action of JNK inhibition indicates JNK activation, abnormal diastolic Ca²⁺ handling and enhanced AF propensity are causally linked in both aged and young animal models.

One might argue that atrial structural remodeling could be another contributing pro-arrhythmic factor. Indeed, patchy fibrosis and generally more interstitial fibrosis is commonly associated with cardiovascular diseases like MI, HF, and diabetic cardiomyopathy. However, we and others have reported that there is no evidence of age-associated structural remodeling in the aged rabbit and human atria without a history of coexisting cardiovascular disease or AF.^{8, 30, 43} Here, all aged animals and JNK challenged young animals showed preserved cardiac function and normal interstitial fibrosis level. Multiple types of cells can be involved in arrhythmogenic remodeling and AF development. Our studies include cellular Ca²⁺ imaging in isolated atrial myocytes as well as RyR2 single channel recordings where cell identity is not an issue. And, our results in whole animals, isolated hearts, intact atria, atrial myocytes, and single RyR2 channels are consistent across this broad experimental spectrum. Thus, our conclusion that there is a critical pro-arrhythmogenic action of JNK mediated CaMKII activation in atrial myocytes is well supported.

In addition to JNK, p38 is another major member of the stress response kinase MAPK family.⁴⁴ In response to stress stimuli the actions of JNK and p38 are dependent on cellular context. For example JNK and p38 have opposite functions (activation or suppression) in cellular senescence.^{45, 46} This is in agreement with our previous findings of unchanged p38 in both aged atria^{7, 8} and in long-term anisomycin-treated atrial myocytes (data not shown). Although possible contributions of other MAPKs remain to be investigated, our results strongly point to JNK having a critical role in promoting atrial arrhythmias.

The JNK isoforms are generally targeted to distinct functions.⁴⁷ In the heart, JNK1 is linked to helping preserve cardiac function and promoting apoptosis during ischemia-reperfusion in MI hearts.⁴⁷ However, to our knowledge, the JNK2 contribution to normal or pathological heart function is unknown to date. Our data show that JNK2 activation is involved in driving SR Ca²⁺ mishandling and thus enhancing AF propensity. This JNK2 action was not only present in aged hearts but also occurred in young animals challenged by JNK activation. Genetic JNK2 depletion or JNK2 specific inhibition reduced arrhythmogenicity in aged and young animal models. This implies that this pathway is not a specific ramification of age, but instead is ever-present and thus available to respond to cellular stresses. Specifically, activated JNK2 boosts SR Ca²⁺ handling and this enhances AF propensity in the absence of coexisting cardiovascular disease.

Another novel finding here is our identification of a unique example of kinase pathway crosstalk as JNK2 directly activates CaMKII to drive the pathology. We and others have previously reported that CaMKII phosphorylation of Ca²⁺ handling proteins is a key pro-arrhythmic factor.^{17, 22, 38, 39} Here, we show JNK2-driven CaMKII activation results in CaMKII-dependent phosphorylation of RyR2815 and PLB17. The consequence is arrhythmogenic diastolic SR Ca²⁺ mishandling. Specifically, increased diastolic SR Ca²⁺ leak triggers aberrant Ca²⁺ activities. We present striking rescue results where either CaMKII or JNK2 inhibition eliminates this downstream diastolic Ca²⁺ handling dysfunction. Enhanced diastolic SR Ca²⁺ leak by itself will lower SR Ca²⁺ content. Interestingly, JNK activation increased both SR Ca²⁺ leak and content. This suggests JNK activation enhances SR Ca²⁺ uptake sufficiently to overcome the larger RyR2-mediated SR Ca²⁺ leak. Consistent with Guo et al³¹ who explored CaMKII action in PLB KO mice, the enhanced SERCA function here did not accelerate the Ca²⁺ transient decay rate because SR Ca²⁺ leak was also enhanced. Previous results⁴⁸ from cross-bred CaMKII-delta overexpression and PLB KO mice are also consistent with this outcome. We are aware that the CaMKII inhibitor KN93 used in the current studies may have off-target effects. However, we have also used the AC3-I mice (cardiac overexpression of CaMKII inhibitory peptide) and AIP in RyR channel measurement to confirm the action of CaMKII inhibition in JNK-driven abnormal Ca²⁺ activities in intact atria. The results were consistent with the rescue action of CaMKII inhibition using KN93 or JNK inhibition in diastolic SR Ca²⁺ leak in myocytes and RyR channel activities.

SR Ca²⁺ leak, uptake and load are increased in paroxysmal AF but CaMKII and RyR2 phosphorylation status have been reported to be unchanged.⁴⁹ The role of CaMKII-dependent SR dysfunction in AF is clearly complex. Our target here was to define JNK's contribution in arrhythmic Ca²⁺ handling in aged atrium in the absence of AF history. Age-associated atrial arrhythmogenicity may or may not be analogous to that in paroxysmal or chronic AF situations. However, Li et al have reported JNK activation in a tachypacing canine AF/HF model.⁵⁰ While the contribution of JNK in sustained AF clearly requires further investigation, we present compelling results that show age-associated JNK activation drives CaMKII-dependent atrial arrhythmogenicity via promotion of RyR2-mediated Ca²⁺ leak and this can be avoided by limiting JNK2 function. Suppression of CaMKII function is known to mitigate arrhythmias and various heart diseases in animal models provoking a great deal of interest in development of CaMKII inhibitors as possible anti-arrhythmic therapeutic agents.²³ We have revealed a novel example of a JNK2-CaMKII link where JNK2 directly activates CaMKII to drive arrhythmic remodeling.

Whether there are other mechanisms of JNK-driven Ca²⁺ mishandling and other ion channels that promote AF clearly warrants further investigation. For example, protein phosphatases are involved in regulating the CaMKII phosphorylation status. While our results of cell free JNK2 and CaMKII protein binding studies suggest that JNK2 directly phosphorylates CaMKII without the presence of protein phosphatases, the role of protein phosphatases in this JNK-CaMKII relationship requires further investigation. However, the preponderance of our current results does suggest a central (albeit not necessarily an exclusive) role of JNK in AF.

It is well-known that JNK activation is involved in the development of cancer, diabetes, and arthritis.^{5, 6, 10} To date, JNK inhibition has been explored as a possible anti-cancer and arthritis therapeutic target in clinical trials.^{10, 47} Our results make exploring therapeutic strategies of JNK inhibition to address cardiac arrhythmias an attractive and still unexplored option. Lastly, aging is just one of the cellular stresses associated with JNK activation.⁸ Other stresses like inflammation, obesity, alcohol abuse and HF are also linked with JNK activation and some of these are associated with greater AF propensity. Therefore, the JNK2-CaMKII crosstalk discovered here might have broad potential therapeutic benefits in anti-arrhythmic drug development. Further investigation is required to determine how JNK inhibition might influence other cellular properties in order to minimize/understand potential off-target effects of any JNK-targeted anti-arrhythmia therapy.

Novelty and Significance.

What Is Known?

• Atrial fibrillation (AF) is the most common sustained arrhythmia and a major public health problem, which currently lacks effective pharmacological therapies.

• Advanced age is a major risk factor for AF. Consequently, the burden of AF is growing exponentially as the mean age of many populations around the world grows.

What New Information Does This Article Contribute?

• Kinase-on-kinase pathogenic crosstalk is critical in governing intercellular Ca²⁺ signaling and consequently Ca²⁺-mediated atrial arrhythmias.

• Advancing age and other stresses (like alcohol, obesity, inflammation, etc.) drive JNK activation and the JNK-CaMKII crosstalk is likely a critical mechanism that couples arrhythmia and these stresses.

• JNK2 inhibition may be a potential target in developing new therapeutic strategies to prevent or treat AF.

Atrial fibrillation is associated with a high risk of mortality and associated morbidities, including stroke and heart failure. However, there is no clear molecular concept addressing the mechanism for enhanced atrial arrhythmogenicity in the aged heart. Here, we report that activation of the kinase JNK2 leads to arrhythmogenic diastolic Ca²⁺ mishandling. Further, we found that JNK2 is a critical activator of calmodulin kinase II (CaMKII), a highly validated proarrhythmic signal, through direct phosphorylation of CaMKII. The phosphorylated and hyper-activated CaMKII ultimately drives the diastolic Ca²⁺ dysfunction that triggers atrial arrhythmias. Our study reveals a previously unrecognized link between JNK2 activation and the age-related enhancement of AF propensity. This link involves a novel, previously unknown, form of pathogenic kinase-on-kinase crosstalk. Our studies reveal a new potential therapeutic target (JNK2) that can be leveraged to prevent the CaMKII hyper-activation and thus limit AF and potentially other cardiovascular diseases.

Nonstandard Abbreviations and Acronyms

Ad

adenovirus

AF

atrial fibrillation

Aniso

anisomycin

AV

atrial ventricular

Ca²⁺

calcium

[Ca²⁺]_SRT

total SR calcium

CaMKII

Ca²⁺/calmodulin-dependent kinase II

CaMKII286-P

phosphorylated CaMKII at Thr286

camui

FRET-based CaMKII sensor

camui-vv

FRET-based CaMKII sensor with mutated Met280Val & Met281Val

Δt_Vm-Ca

activation time difference of action potential and Ca²⁺ transient

hCaMKII

human CaMKII

HF

heart failure

hJNK

human JNK

IP/IPed

immunoprecipitation/immunoprecipitated

JNK

c-Jun N-terminal Kinase

JNK1dn

dominant negative JNK1

JNK2dn

dominant negative JNK2

JNK1/2dn

dominant negative JNK1 and JNK2

JNK2KO

JNK2 knockout

JNK2I-IX

JNK2 inhibitor

MI

myocardial infarction

NAC

N-acetyl-L-cysteine

NCX

Na⁺/Ca²⁺-exchanger

PLB

phospholamban

PLB16-P

phosphorylated PLB at Ser16

PLB17-P

phosphorylated PLB at Thr17

P_o

opening probability

RyR

ryanodine receptor

RyR2

ryanodine receptor isoform 2

PKA

protein kinase A

SR

sarcoplasmic reticulum

SRh

sinus rhythm

Tg

transgenic

τ

time constant

WT

wild-type