All laboratory mice and rats in this study were maintained in a pathogen-free facility at the University of Manchester. Animal studies were performed in accordance with the United Kingdom Animals (Scientific Procedures) Act 1986 and were approved by the University of Manchester Ethics Committee and all procedures were conducted under the Project License P3A97F3D1. C57BL/6J mice and Sprague Dawley rats were used. For biochemical, histological, hemodynamics, and echocardiographic studies 5 to 8 animals per group were used. All in vivo studies were blinded for both genotype and surgical procedure/treatment during the measurement and analysis stages. Genetically modified and wild-type litter-mates are analyzed in groups of the same age and sex. Only male mice aged between 9 and 10 weeks old were included in all in vivo experiments in this study. Cervical dislocation followed by removal of the heart prior to permanent cessation of the circulation was used for euthanasia in mice. Rats were sacrificed by intraperitoneal injection of Pentobarbitone Sodium 20% w/v and 250 μl Heparin Sodium (1,000 U/ml).

Mice were anesthetized with 2% isoflurane and ventilated for TAC surgery as previously described (24). Briefly, the transverse aorta between right innominate and left common carotid arteries was subjected to a 27-gauge constriction by a 7-0 Prolene suture to produce an approximate pressure difference of 30–40 mmHg between the two common carotid arteries. Buprenorphine (0.1 mg/kg) was administered for post-operative analgesia. Cardiac function was assessed by echocardiography and the hearts were collected for further analyses.

To examine the effect of ER stress in the heart, tunicamycin (2 mg/kg) (Sigma, T7765) or saline was applied to mice by intraperitoneal injection. Either 2 or 7 days after the injection, cardiac function was examined and the hearts were collected for further analyses.

Mice were anesthetized with 1.5% isoflurane and echocardiographic assessment of cardiac function was carried out. Transthoracic M-mode echocardiographic recordings were performed using an Acuson Sequoia C256 system (Siemens) ultrasound machine (24).

Five micrometer-thick paraffin embedded heart sections were stained with hematoxylin & eosin to measure cross-sectional areas by randomly selecting 200 cardiomyocytes. TUNEL assay was performed to detect apoptotic cardiomyocytes using the in situ Cell Death Detection kit (Roche). A total of 10,000 cardiomyocytes from random fields per heart were analyzed. The images were obtained by Zeiss Axioplan2 microscope and then analyzed by Image J software. In situ superoxide production was measured with the oxidative fluorescent dye dihydroethidium (DHE). Cryosections (10 μm) were incubated with DHE for 30 min at 37°C. Images were captured using an Olympus fluorescence microscope. Fluorescence intensities were quantified with ImageJ software.

Intracellular glutathione levels was determined using the Glutathione Assay Kit (#CS0260, Sigma-Aldrich Inc.), following the manufacturer's instructions. For oxidized glutathione (GSSG) determination, reduced glutathione (GSH) was inactivated by the addition of 2-vinylpyridine in the presence of triethanolamine. GSH content was calculated subtracting GSSG from total GSH content. Quantifications were performed by parallel measurements of GSH or GSSG standards, and results were expressed as nmol/mg protein.

iWT.D2 and WT1.Bld2 human iPSCs were derived from peripheral blood cells, and maintained with E8 medium (Thermo Fisher Scientific) on Geltrex-coated plates. Cardiomyocyte differentiation of iPSCs was performed as previously described (21). iPSC-derived cardiomyocytes (iPSC-CMs) were cultured in cardiac culture medium (RPMI1640 HEPES Glutamax, B27) up to 120 days for maturation.

H9C2 cells were seeded onto sterile coverslips. After treatment with tunicamycin (5 μg/ml), cells were fixed in 4% PFA. Cells were permeabilized in 0.2% triton X-solution and blocked with 10% normal donkey serum (Stratech). Fixed cells were incubated with primary antibody (1:200 dilution) against Nrf2 (Abcam, ab62352). Secondary anti-rabbit antibody (Stratech, 1:1,000) conjugated to Alexa Fluor^®488 was used to detect Nrf2. DAPI was used for nuclear visualization. Images were collected on an Olympus fluorescence microscope. To exclude the possibility of false positive by the primary antibodies, IgG (Cell Signaling, 2729) was used as the control antibody. IgG did not detect any positive staining, which was not represented.

To determine Hrd1 promoter activity, H9C2 cells were transfected with human Hrd1 promoter luciferase reporter (−3,324 to −1 relative to transcription start site, containing XBP1s specific UPRE sites, kindly provided by Professor Kazutoshi Mori, Kyoto University, Japan) for 48 h. The luciferase activity was analyzed using Dual-Luciferase Reporter Assay System (Promega, E1910) according to the manufacturer's instructions.

ARCMs were transfected with plasmids expressing Flag-Pak2(T402E), Flag-XBP1s or Myc-Hrd1 in combination with control or Hrd1 siRNA. Forty-eight hours after transfection, 100 μg/mL of cycloheximide was added, and chase experiments were performed in a time window of 1 h after addition of cycloheximide to examine the effect of active Pak2, Hrd1, and XBP1s in the degradation rate of Nrf2. Changes of Nrf2 levels were assessed by immunoblot.

Immunoprecipitations were performed with Protein G sepharose (Sigma) following the manufacturer's instruction. Briefly, H9C2 cells or heart tissue were lysed with immunoprecipitation buffer (Tris 50 mM, NaCl 250 mM, 0.25% v/v TritonX- 100, and 10% Glycerol; pH 7.4). Two milligram of the protein extract was immunoprecipitated with antibodies against Nrf2 (Ab137550, Abcam), HA (2367S, Cell Signaling), Hrd1 (Nb100-2526, Novus), or IgG (Cell Signaling, 2729), respectively. Immune complexes were eluted in Laemmli sample buffer (65.8 mM Tris-HCl, pH 6.8, 2.1% SDS, 26.3% glycerol, 0.01% bromophenol blue). Precipitated and input proteins were subjected to SDS-PAGE and immunoblotted using the respective antibodies.

Cardiomyocytes cell viability was determined by MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] reduction assay. After treatment, cells were incubated in culture media containing MTT solution (0.5 mg/mL) for 3 h at 37°C. The purple formazan product in the cells was solubilized by addition of acidic isopropanol (0.04 M HCl) and the absorbance was measured at 570 nm with background subtraction at 630 nm. Cell viability was expressed as the percentage of untreated controls.

Total protein from tissues or cells was obtained with Triton lysis buffer (Tris 20 mM, NaCl 137 mM, EDTA 2 mM, 1% Triton X-100, β-glycerophosphate 25 mM, Na₃VO₄ 1 mM, phenylmethanesulfonylfluoride 1 mM, aprotinin 1.54 μM, leupeptin 21.6 μM, 10% glycerol; pH 7.4). Protein concentration was determined by Bio-Rad protein assay. Protein extracts (20 μg) were subject to immunoblot analyses with antibodies listed in Supplementary Table 1. Immune-complexes were detected by enhanced chemiluminescence with anti-mouse or anti-rabbit immunoglobulin-G coupled with horseradish peroxidase.

Total RNA was extracted from ventricular tissues or cells using Trizol (Invitrogen) followed by conversion to cDNA. All primers were purchased from Qiagen. Real-time quantitative PCRs were performed using SYBR Select PCR Master Mix according to the manufacturer's instruction in the Step One Plus PCR System (Applied Biosystems). The fold change was analyzed using the 2-^ΔΔCT method (25). The level of expression was normalized to Gapdh.

The study complies with the Declaration of Helsinki and was reviewed and approved by the ethics committee of the Technische Universität Dresden (Az: EK 422092019 and EK EK 446122011. LV tissue was obtained from dilated heart failure patients (n = 5), and tissue from non-failing donor hearts (n = 5) served as reference.

Sample sizes were calculated based on available comparable data to achieve 90% power at a 5% significance level. For all in vivo and in vitro analyses, normal distribution (Gaussian distribution) was first determined by Shapiro–Wilk test. Data were then analyzed using Student's t-test for comparisons between two experimental groups or one- or two-way ANOVA with Bonferroni post-hoc tests for comparisons among multiple experimental groups. Data are expressed as scatter plots indicating mean ± S.E.M. P-values p < 0.05 were considered statistically significant.

In our recent study, we showed exacerbated hypertrophy in cardiac-specific Pak2 knockout mice (Pak2-CKO) under pressure overload by transverse aortic constriction (TAC) by an increase of deleterious ER stress response in the heart (21). Pak2-CKO mice show increased myocardial cell death (Supplementary Figure 1A), decreased cardiac function (Supplementary Figure 1B) and dysregulation of the IRE-XBP1 branch of the UPR (Supplementary Figure 1C) leading to substantial ER stress and aggregation of ubiquitinated proteins in TAC stressed hearts (Supplementary Figures 1C,D). Intriguingly, compared to their Pak2-Flox littermates, Pak2 deleted mice under TAC show a significant increase in reactive oxygen species (ROS) (Figure 1A) and imbalance in reduced glutathione content by determination of the GSH/GSSG ratio in hearts (Figure 1B; Supplementary Figures 2A,B). To directly investigate the effects of ER stress on cardiac function, acute stress was induced in mice with tunicamycin via intraperitoneal injection (2 mg/kg), which inhibits N-linked glycosylation of proteins in the ER (26). The use of tunicamycin, in combination to anti-cancer drugs, has been proposed to enhance the efficacy of chemotherapy (27). However, the systemic ER stress produced leads to increased cardiomyocyte apoptosis and development of cardiomyopathy (28). Similar to what we observed with TAC, Pak2 cardiac depletion in mice subjected to ER stress show increased ROS (Figure 1C), imbalance of GSH/GSSG ratio (Figure 1D; Supplementary Figures 2C,D) and increased cell death (Figure 1E). Moreover, cardiac deletion of Pak2 exacerbates ER stress, protein aggregation (Figure 1F; Supplementary Figure 2E) and increased cardiac dysfunction (Supplementary Figure 2F) in tunicamycin treated mice. Overall, this data suggests that Pak2 depletion, in addition to induce UPR imbalance, might impair the redox response in hearts under pathological stress.

Pak2 promotes a protective response to ER stress via activation of IRE1-XBP1 branch of the UPR in cardiomyocytes (21). Moreover, Pak2 can be activated in response to hydrogen peroxide in isolated cardiomyocytes (Supplementary Figure 3A). Pak2 depletion hinders IRE1-XBP1 activation in response to oxidative stress and leads to an increase in the apoptotic marker CHOP (Supplementary Figure 3B), suggesting that Pak2 could be protective against oxidative stress in cardiomyocytes via IRE-XBP1 activation in response to ROS-induced ER stress. The observation that oxidative stress is exacerbated by Pak2 depletion in the heart is supported by gene array analysis preformed in Pak2-CKO hearts under ER stress. Among those transcripts significantly altered compared to Pak2-FLOX control (P < 0.05), in addition to ER stress response, there is a significant downregulation in transcripts involved in oxido-reduction and glutathione metabolism according to functional annotation by DAVID version 6.2 (Figure 2A). Prompted by this data, we decided to investigate the interrelationship between Pak2 activation and key redox pathways. Under acute ER stress (2 days tunicamycin injection), both Pak2 phosphorylation and the expression of the redox regulator Nrf2 show significant increase in mouse hearts (Figure 2B). Prolonged ER stress (1 week tunicamycin injection) leading to cardiac dysfunction (Figure 2D; Supplementary Figure 4A) and increased hypertrophy (Figure 2E; Supplementary Figure 4B) results in decrease in cardiac Pak2 phosphorylation (Figure 2B). Intriguingly, these hearts, showed further increase in Nrf2 protein levels with no change in Nrf2 mRNA (Figures 2B,C). Similarly, using TAC to provoke cardiac disease-mimetic ER stress, we observe a significant reduction in Pak2 phosphorylation and further upregulation of Nrf2 in the hearts of mice after 5 weeks of pressure overload (Figures 2F,G), showing severe cardiac dysfunction (Supplementary Figure 4C) and CHOP upregulation (Figure 2F). Interestingly, Pak2 deleted (Pak2-CKO) hearts show enhanced Nrf2 upregulation as early as 2 weeks after TAC (Figure 2H), suggesting a potential role for Pak2 in Nrf2 expression. Importantly, the increase in Nrf2 protein levels under ER stress in wild type, Pak2-Flox and Pak2-CKO hearts occurred with no significant change in Nrf2 mRNA levels (Figures 2C,G,I). These results indicate that whereas under initial ER stress there is a simultaneous increased Pak2 phosphorylation and Nrf2 expression, prolonged ER stress or pressure overload show an inverse correlation between Pak2 activation and Nrf2 protein levels in the dysfunctional heart.

To examine the effects of Pak2 activation in the heart, cardiac-specific expression of constitutively active Pak2 was achieved via AAV9-mediated gene delivery in mice. C57BL/6 mice were injected via tail vein with TnT promoter-driven AAV9-Pak2-T402E (AAV9-Pak2) or AAV9-GFP prior to tunicamycin treatment for 1 week. Consistent with our previous findings, overexpression of active Pak2 prevents the ER stress induced progression into cardiac hypertrophy and dysfunction (Figures 3A–E; Supplementary Figures 5A,B). Pak2 overexpression also significantly alleviates cardiac cell death and fibrosis induced by tunicamycin (Figures 3F,G; Supplementary Figure 5C). Furthermore, in response to tunicamycin-induced ER stress, active Pak2 is able to activate the IRE1-XBP1s branch of the UPR (Figure 3H) with no significant effect over ATF6 or PERK. Interestingly, cardiac expression of active Pak2 prevents the increase in Nrf2 protein levels observed with no significant change in Nrf2 transcript level (Figure 3I). Overall, we further confirmed the protective role of cardiac specific activation of Pak2 under ER stress. Additionally, the inverse relation observed on the expression of Nrf2 and Pak2 in vivo might suggest a negative role for Pak2 in Nrf2 expression under cardiac ER stress.

To elucidate the link between the Pak2 activation and Nrf2 downregulation, a cellular cardiomyocyte model was used. Similar to our in vivo observations, acute tunicamycin treatment (5 μg/mL for 2 h) induced an increase in Pak2 phosphorylation and Nrf2 levels in isolated adult rat cardiomyocytes (ARCMs). Sustained ER stress for 24 h, leading to exacerbated protein aggregation (Supplementary Figure 6A) disrupts Pak2 activation and results in further increase in Nrf2 accumulation (Figure 4A; Supplementary Figure 6B). Adenovirus mediated shRNA-ablation of Pak2 in cardiomyocytes under ER stress greatly exacerbates ubiquitinated protein accumulation (Supplementary Figure 6C) and induces an increase in Nrf2 levels accompanied by an increase in the apoptotic markers CHOP and cleaved caspase 3 (Figure 4B). As Nrf2 is crucial in the maintenance of redox homeostasis, we tested whether the increase in Nrf2 expression observed is associated to an increase in Nrf2 nuclear translocation and antioxidant transcriptional activity. Tunicamycin induced an increase in Nrf2 nuclear localization after 2 h of treatment as detected by immunoblot after cell fractionation in rat cardiomyocytes (Figure 4C). Moreover, Pak2 deletion promotes an increase in nuclear Nrf2 following tunicamycin treatment (Figure 4D). Rat cardiomyocytes Immunofluorescence showed a retention and accumulation of Nrf2 in the nucleus following tunicamycin treatment in Pak2 depleted cells (Figure 4E). Functionally, qPCR analyses on the Nrf2 transcriptional targets Nqo1, Ho1 G6pd and Gpx1, showed initial increase at 2 h tunicamycin treatment whilst a significant decrease in their expression is observed at 12 h, consistent with the observed Nrf2 localization (Figures 4E,F). Whereas, Pak2 ablation did not significantly affect the mRNA expression of Nqo1, Ho1, and G6pd, induction of Gpx1 by sustained ER stress is exacerbated by Pak2 knockdown (Figure 4F). This suggests that the accumulated Nrf2 observed under Pak2 knockdown in cardiomyocytes could be active toward specific transcriptional targets, thus Nrf2 under Pak2 depletion and the subsequent ER stress environment could be triggering transcriptional responses different from its protective antioxidant role. Of note, early Nrf2 inhibition by ML385 (2 μM) in cardiomyocytes prior to treatment with tunicamycin show a significant reduction in cell survival (Figure 4G). Interestingly, when cardiomyocytes were subjected to ER stress induced by tunicamycin following Pak2 depletion, Nrf2 inhibition after the onset of ER stress appears to partially enhance cardiomyocytes survival (Figure 4H), indicating that in Pak2 depleted cardiomyocytes, excessive Nrf2 might play a detrimental role. Recent reports have linked aberrant Nrf2 activity with a pathological increase in RAAS gene expression (20). Promoter sequence analyses show conserved predicted Nrf2 binding sites for key components of the RAAS system (Supplementary Figure 7A). Thus, we wanted to test whether the activation of Nrf2 in the context of maladaptive ER stress is associated with dysregulation of angiotensin signaling. qPCR analyses showed upregulation of Agt, Atr1, and Ace1 in response to ER stress. Interestingly, whereas Pak2 depletion has little effect on the mRNA expression at 2 h of tunicamycin, following 12 h an exacerbated induction of Agt, Atr1, and Ace1 mRNA expression and downregulation in Mas1 expression was observed (Figure 4I; Supplementary Figure 7B). Furthermore, this aberrant overexpression of RAAS transcripts observed in Pak2 depleted cells, can be reverted by inhibition of Nrf2 (Figure 4J). Importantly, both the Angiotensin II receptor blocker Losartan and the angiotensin-converting enzyme (ACE) inhibitor Enalapril were able to significantly reduce apoptotic markers and increase cell survival after 24 h tunicamycin in Pak2 depleted cardiomyocytes (Supplementary Figures 8A–D). This suggests that the detrimental effects of Pak2 deletion under sustained ER stress is at least in part due to local dysregulation of angiotensin signaling. Overall, this data suggests that in conditions of Pak2 depletion and sustained ER stress, Nrf2 is retained in the nucleus and has a detrimental effect in cardiomyocytes by inducing an aberrant expression of RAAS genes.

Increased Nrf2 levels in Pak2 depleted hearts after pressure overload or ER stress occurred without significant changes in Nrf2 mRNA levels (Figures 2C,G,I). Similarly, cardiac overexpression of active Pak2 resulted in a decrease in Nrf2 protein (Figure 3I). This suggested that Pak2 might promote Nrf2 degradation. We had previously reported Pak2 as an ERAD regulator via XBP1-Hrd1 activation (21). Interestingly, Hrd1 expression follows a similar pattern to Pak2 activation following both direct ER stress and pressure overload (Supplementary Figures 9A,B). Moreover, Pak2 in vivo and in vitro depletion ablates Hrd1 expression in response to stress (Supplementary Figures 9C–E) and cardiac-specific overexpression of constitutively active Pak2 promotes cardiac Hrd1 upregulation in mice under systemic ER stress (Supplementary Figure 9F). Hrd1 targets misfolded proteins of the endoplasmic reticulum for ubiquitination and subsequent proteasomal degradation and it has been described as a novel Nrf2 ubiquitin E3 ligase (17). Led by the above results we decided to test whether the XBP1-Hrd1 axis of the UPR is responsible for the Pak2 mediated downregulation of Nrf2 in the heart. Consistent with this idea, the increased Nrf2 accumulation and Hrd1 downregulation observed in stressed cardiomyocytes under Pak2 knockdown can be reverted by overexpression of XBP1s in vitro in tunicamycin treated cardiomyocytes (Supplementary Figure 10A). Furthermore, restoring XBP1s activity via AAV9 injection, increases in Hrd1 levels in Pak2 knockout mice 4 weeks after TAC, and reverts the Nrf2 accumulation in Pak2-CKO hearts (Supplementary Figure 10B). Additionally, XBP1s overexpression can revert Pak2 knockdown mediated Nrf2 nuclear accumulation (Supplementary Figure 10C) and the aberrant transcriptional activity observed in RAAS components mRNA (Supplementary Figures 10D,E) resulting in enhanced cardiomyocyte survival (Supplementary Figure 10). Overall, these data indicate that Pak2 via XBP1 activation is able to relieve the aberrant activation of Nrf2 in cardiomyocytes. This further supports the role of Pak2 as a negative regulator of Nrf2 via activation of the IRE1-XBP1-Hrd1 axis.

Pak2 knockdown significantly decreases tunicamycin-induced Hrd1 transcription as measured by a luciferase reporter gene carrying human Hrd1 promoter region (−1 to −3,324), which can be restored by overexpression of XBP1s (Figure 5A; Supplementary Figure 11A). To address the role of Hrd1 on Nrf2 degradation, Nrf2 ubiquitination was assessed by immunoprecipitation of HA-tagged ubiquitin and detection of Nrf2 by immunoblot in cardiomyocytes. We observed that overexpression of active Pak2 enhances Nrf2 ubiquitination (Figure 5B; Supplementary Figure 11B). Importantly, Pak2 mediated Nrf2 ubiquitination is abolished by both, the Hrd1 inhibitor LS-102 (10 μM) (Supplementary Figure 11C) and by Hrd1 knockdown via siRNA (Figure 5C). Immunoprecipitation assays showed that Nrf2 and Hrd1, but not Pak2 directly interact in cardiomyocytes (Figure 5D; Supplementary Figure 11D), and this interaction is enhanced by ER stress (Figure 5E). To confirm that overexpression of Pak2 mediated Nrf2 degradation, we performed cycloheximide (CHX) chase experiments. We found that overexpression of active Pak2 enhances the clearance of Nrf2 within a window of an hour (Figure 5F). Importantly, Pak2 triggered Nrf2 downregulation can be prevented by inhibition of Hrd1 (Supplementary Figure 11E), and Hrd1 knockdown abolishes Pak2-induced Nrf2 clearance (Supplementary Figure 11F). Consistently, direct overexpression of Hrd1 significantly accelerates Nrf2 degradation in cardiomyocytes (Figure 5G). Nrf2 expression is mainly regulated by Keap1, which facilitates Cullin 3 mediated Nrf2 ubiquitination for proteasomal degradation. Importantly, neither Keap1 nor Cullin 3 expression are affected by Pak2 (Supplementary Figures 12A,B) and silencing Keap1 does not seem to impede Pak2 mediated Nrf2 downregulation (Figure 5H). Overall, we found that Pak2 prevents Nrf2 aberrant activation by enhancing its ubiquitination and degradation via the IRE1-XBP1-Hrd1 axis in response to ER stress.

Finally, to seek human-relevant data, we tested whether downregulation of Nrf2 via Pak2 activation is protective in human cardiomyocytes. Human-induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) were subjected to tunicamycin-induced ER stress. Tunicamycin treatment for 24 h resulted in Nrf2 accumulation accompanied by enhanced apoptosis markets (Figure 6A) and a significant reduction in cell survival (Figure 6B). Adenoviral overexpression of active Pak2 was able to restore Hrd1 expression with the subsequent Nrf2 downregulation (Figure 6A), alleviate cell death and enhance cell survival (Figure 6B). Conversely, Pak2 depletion in human cardiomyocytes dramatically increased apoptosis in response to ER stress (Figures 6C,D). Interestingly, these effects observed for Pak2 knockdown on ER stress related cell death can be partially restored by inhibition of Nrf2 (Figures 6C,D) highlighting the detrimental role of Nrf2 in Pak2 defective cardiomyocytes in response to sustained ER stress. Moreover, Pak2 depletion results in altered expression of Atr1 and Mas1 in human cardiomyocytes (Figure 6E) which can be prevented by inhibition of Nrf2. The above data suggest that Pak2 plays a protective role in human cardiomyocytes in response to stress by enhancing the protective UPR and preventing the accumulation of Nrf2 and the consequent aberrant expression of RAAS components leading to cardiomyocytes death. Importantly, these results are consistent with our findings in myocardium from transplantation patients with dilated heart failure showing a significant decrease in Pak2 expression at both protein and transcript level with an increase in Nrf2 protein expression compared to samples obtained from healthy donors (Figure 6F; Supplementary Figure 13). Loss of Pak2 could then contribute to the pathogenesis of heart failure by promoting myocardium damage from defective ER stress response and enhancing Nrf2 maladaptive regulation of RAAS activation.