Respiratory chain signalling is essential for adaptive remodelling following cardiac ischaemia

Abstract Cardiac ischaemia‐reperfusion (I/R) injury has been attributed to stress signals arising from an impaired mitochondrial electron transport chain (ETC), which include redox imbalance, metabolic stalling and excessive production of reactive oxygen species (ROS). The alternative oxidase (AOX) is a respiratory enzyme, absent in mammals, that accepts electrons from a reduced quinone pool to reduce oxygen to water, thereby restoring electron flux when impaired and, in the process, blunting ROS production. Hence, AOX represents a natural rescue mechanism from respiratory stress. This study aimed to determine how respiratory restoration through xenotopically expressed AOX affects the re‐perfused post‐ischaemic mouse heart. As expected, AOX supports ETC function and attenuates the ROS load in post‐anoxic heart mitochondria. However, post‐ischaemic cardiac remodelling over 3 and 9 weeks was not improved. AOX blunted transcript levels of factors known to be up‐regulated upon I/R such as the atrial natriuretic peptide (Anp) whilst expression of pro‐fibrotic and pro‐apoptotic transcripts were increased. Ex vivo analysis revealed contractile failure at nine but not 3 weeks after ischaemia whilst label‐free quantitative proteomics identified an increase in proteins promoting adverse extracellular matrix remodelling. Together, this indicates an essential role for ETC‐derived signals during cardiac adaptive remodelling and identified ROS as a possible effector.


| INTRODUC TI ON
The high susceptibility of cardiomyocytes to oxygen deprivation has been originally attributed to their almost exclusive dependence on oxidative metabolism for ATP production. 1 Indeed, approximately 30% of the cardiomyocyte's intracellular volume is occupied by mitochondria which harbour the electron transport chain (ETC) in their inner membrane. The ETC is composed of four oxidoreductase complexes (cI-cIV) that facilitate substrate oxidation and electron transfer with oxygen being the final acceptor. Importantly, electron transfer is coupled to the generation of a proton-electrochemical potential gradient across the inner mitochondrial membrane, which itself is the driving force for ATP production, a process described as oxidative phosphorylation (OXPHOS). Although the ETC thus couples ATP production to oxygen consumption, pioneering work in isolated rabbit heart mitochondria 2 as well as in intact rabbit hearts 3 revealed that cardiac ATP depletion alone is insufficient to account for all post-ischaemic tissue damage. 2 Instead, electron flow through the ETC upon reperfusion was described as a source of potentially detrimental reactive oxygen species (ROS), proposed to trigger ischaemia-reperfusion (I/R) injuries. 2 Conversely, the therapeutic use of untargeted antioxidants such as vitamin C for post-ischaemic cardioprotection has given contradictory results, [4][5][6][7][8][9] and itself produced detrimental side effects when therapeutically used. 10 These seemingly contradictory results were thought to be due to biphasic effects of ROS with high ROS concentrations leading to damage and lower concentrations eliciting adaptive responses. 11,12 The term 'mitohormesis' was proposed 13 to describe this phenomenon. Indeed, a recent study demonstrated that increased ROS levels specifically in the mitochondrial compartment have a cardioprotective effect upon I/R. 14 Of course, it could equally be possible that untargeted antioxidants cannot reach sufficient levels in the mitochondrial compartment to be effective, and therefore fail to detoxify mitochondrial ROS at the site of its production. Furthermore, ROS may generate positive or negative effects in a time-dependent manner.
To test this assumption, a number of mitochondrial-targeted antioxidants have been designed. MitoQ, one prominent example, consists of ubiquinol (reduced quinone) targeted specifically to mitochondria by a covalently fused lipophilic triphenyl phosphonium (TPP) cation. 15 Importantly, in rats that underwent cardiac I/R, MitoQ, but not the untargeted antioxidant or TPP alone, significantly decreased contractile dysfunction, cell death and mitochondrial damage. 15 Positive effects of MitoQ were also seen in other studies addressing post-transplantation injury 16 and pressure overload-induced heart failure. 17 Following a similar rationale, Szeto-Schiller (SS) peptides 18 were used, which accumulate in mitochondria in a membrane potential-independent manner. The use of SS peptides caused phenotypic improvements in a model of angiotensin II-induced mitochondrial ROS production and myocardial contractile failure 19 as well as upon transverse-aortic constriction. 20 Both, MitoQ and SS peptides appeared biologically safe and beneficial when long-term administered 21,22 arguing that intramitochondrial ROS are rather detrimental than physiologically relevant. Finally, using high-throughput chemical screening, small molecules that suppress superoxide and/or hydrogen peroxide production at the I Q site of cI were identified, S1QELs, 23 that act through a mechanism that does not affect regular OXPHOS. Most importantly, ROS generated at I Q was discussed to be instrumental for cellular stress signalling. Indeed, application of S1QELs protected the perfused mouse heart from I/R injuries. 23 A unifying mechanism by which the different mitochondrially targeted antioxidants may fulfil their beneficial effects was recently put forward. 24 According to this, succinate, an intermediate metabolite of the tricarboxylic acid cycle, accumulates during cardiac ischaemia. 25 Reperfusion evokes rapid succinate oxidation through reverse electron transport (RET) from ubiquinol to cI concomitant with a burst of mitochondrial ROS.
Scavenging RET-induced ROS may therefore be the mechanism by which cellular and organ damage is prevented although other sources of ROS, that are intrinsically linked, may play a signalling role as well. 26 Plants and many other organisms, but not insects or vertebrates, harbour a by-pass mechanism to protect from ETC-mediated respiratory stress conditions, the alternative oxidase (AOX). 27 AOX is a non-protonmotive single di-iron carboxylate redox transfer protein. 28 Where expressed, AOX oxidizes ubiquinol and by-passes cIII and cIV, directly transferring electrons to oxygen. In previous studies, we were However, post-ischaemic cardiac remodelling over 3 and 9 weeks was not improved.
AOX blunted transcript levels of factors known to be up-regulated upon I/R such as the atrial natriuretic peptide (Anp) whilst expression of pro-fibrotic and pro-apoptotic transcripts were increased. Ex vivo analysis revealed contractile failure at nine but not 3 weeks after ischaemia whilst label-free quantitative proteomics identified an increase in proteins promoting adverse extracellular matrix remodelling. Together, this indicates an essential role for ETC-derived signals during cardiac adaptive remodelling and identified ROS as a possible effector.  40 This suggests a role for the mitochondrial respiratory chain in both, energy homeostasis and stress signalling.
Here, we tested how maintaining electron flow impacts on remodelling in the post-ischaemic mouse heart. We show that AOX expression supports ETC function and decreases mitochondrial ROS levels, yet impairs adaptive cardiac remodelling. This indicates an essential role for ETC-derived stress signals in post-ischaemic cardiac adaptation.

| Animal models
Mice with ubiquitous expression of Ciona intestinalis AOX 33 were maintained on a C57BL/6J background in temperature-and humidity-controlled facilities at 12 hours of light-to-dark cycles with access to water and food ad libitum. Open-chest in situ I/R procedures were conducted at the University of Cambridge, Cambridge, UK upon approval by Home Office license 70/8238. In vivo I/R procedures were conducted at the Max Planck Institute for Heart and Lung Research in Bad Nauheim, Germany upon approval by the Regierungspräsidium Darmstadt (V54-19c20/15-B2/1014).

| Open-chest in situ I/R mouse model and estimation of infarct size and zone at risk
Details of the procedure have been published elsewhere. 14,41 Briefly, WT and AOX transgenic littermates were anesthetized and ventilated with ambient air supplemented with oxygen (peak terior descending artery (LAD). The LAD was then surrounded by a 7-0 nylon suture to form a snare. After a 15 minutes period of stabilization, the I/R procedure was started with 30 minutes of ischaemia followed by 2 hour reperfusion. After the I/R procedure, the LAD was re-occluded and Evans blue was injected retrogradely through the aortic root. Evans blue staining was used to demarcate the zone of ischaemia hereafter referred to as the region at risk zone. After staining, hearts were excised, perfused with 0.9% saline solution, frozen and transversely cut into 1 mm slices. The slices were incubated in 1% triphenyltetrazolium chloride (TTC) in sodium phosphate buffer (pH 7.4) and incubated at 38°C for 20 minutes. TTC stains the viable (non-infarcted myocardium) brick-red based on catalytic dehydrogenase activity. Finally, the slices were immersed in 10% formalin and areas of infarct and risk zone determined slice by slice using planimetry. The wound was closed using absorbable, synthetic 5-0 vicryl (polyglactin 910) sutures. Weaning from ventilation took several minutes and was ended when spontaneous respiration became evident.

| Langendorff perfusion experiments
Hearts were isolated from mice, and procedures to measure cardiac functions were performed as previously described. 42 Briefly, after deep anaesthesia with isoflurane (5%), hearts were excised from the chest cavity, transferred rapidly to ice-cold saline and immediately mounted on the cannula of Langendorff perfusion system.

| Respirometry and ROS production using isolated mitochondria
Mitochondria were isolated from cardiac tissue as described previously. 43 Mitochondrial oxygen consumption and ROS production were To assess hydrogen peroxide production as the consequence of RET, mitochondrial respiration was stimulated by the addition of succinate and respiration allowed to diminish all oxygen present within the chamber. Once anoxic, mitochondria were left for 20 or 30 minutes to simulate ischaemia, after which oxygen was allowed back into the chamber and oxygen consumption and ROS production measured. To assess sole AOX activity, cyanide was added at the start of the experiment.

| Western blot analysis
Western blots were performed essentially as described elsewhere. 44 Rabbit serum raised against two AOX peptides (anti-AOX, 1:20 000, 21st Century Biochemicals) 30 and mouse monoclonal antibody against the voltage-dependent anion channel (anti-VDAC1, 1:1000, Abcam ab14734) were used and protein bands were visualized using a LI-COR Odyssey flatbed scanner with anti-mouse and antirabbit secondary antibodies conjugated to IRDye 680RD and IRDye 800CW, respectively.

| Respirometry using skinned heart fibres and heart tissue homogenate
Mitochondrial respiratory activity upon in vivo I/R was measured using skinned fibre isolation (3-weeks I/R) as described in detail elsewhere 45,46 or after homogenization of heart tissue using a POLYTRON PT 1200 E Manual Disperser (Ecoline; 9-weeks I/R). For skinned fibre isolation, left-ventricular heart tissue was dissected on ice in a plastic cell-culture dish using extra sharp forceps to extract thin muscle fibres. These fibres were permeabilized for 30 minutes using saponin (50 µg/mL) as previously described. 45,46 Upon washing, fibres were dried on Whatman paper and weighted before transfer to the oxygraph chamber. For heart tissue homogenization, left-ventricular heart tissue was weighted and then minced on ice using scissors and a POLYTRON homogenizer before transfer to the oxygraph chamber.
Using either technique, mitochondria were found to be accessible for substrates and inhibitors, and exhibited coupled respiration albeit at different activity levels per mg of the respective sample. Oxygen consumption was measured using Oroboros O2k oxygraphs (Oroboros Instruments, Innsbruck, Austria) 47

| RNA extraction, reverse transcription (RT) and qPCR
Total RNA was extracted from left ventricles by RNeasy Micro Kit

Gene
Forward primer Reverse primer TCGCAGAGATGTCCAGTC CCCACCGAACTCAAAGAAG

| Transmission electron microscopy
A small piece of tissue from the heart apex was kept in storage buffer containing

| Nano LC-MS/MS and data analysis
Mouse heart samples were homogenized using a FastPrep-24 5G High Speed Homogenizer (MP Biomedicals). Tissue pieces were

| Statistical analyses
Statistical analyses, except for proteome, were performed using GraphPad Prism (GraphPad Software, version 7 for Mac OS X).

| AOX does not decrease acute I/R injuries
Succinate accumulation during cardiac ischaemia 25 is one proposed trigger for RET-mediated ROS production 24 and supposedly I/R-mediated injury. AOX was previously demonstrated to blunt RET-induced ROS production under various conditions. 33,34,36,37 We therefore first tested whether the extent of acute I/R injury (infarct size) would be decreased in the post-ischaemic heart of mice with excellent cardiac AOX protein expression and catalytic AOX activity. 33 Using the open-chest in situ I/R model, 41 30 minutes of ischaemia were applied by occluding the left anterior descending (LAD) coronary artery followed by 2 hours of reperfusion in wild-type (WT) and AOX littermates. The infarct size was estimated as a percentage scar tissue of the risk zone. 41,52 As shown in Figure 1A,B, AOX expression did not confer any benefit. We next sought to investigate whether the heart might nevertheless benefit during the early post-ischaemic phase through improved functional recovery of surviving myocytes and analysed heart contractile functions ex vivo in the isolated, perfused (Langendorff) heart.
Upon 45 minutes of ischaemia, WT and AOX hearts showed nearly indistinguishable systolic (P syst, Figure 1C) and diastolic (P diast, Figure 1D) pressure-response curves. Furthermore, calculated left ventricular developed pressures (LVDP [maximal systolic-diastolic pressure]) ( Figure 1E) and heart rate*pressure products (RPP, a surrogate value of cardiac work) 53 ( Figure 1F) revealed no difference. Of note, hearts typically develop a rigour during ischaemia.
This manifests as an increase in pressure during the ischaemic insult and, in our experience, correlates closely with the infarct size.
WT and AOX hearts developed the same degree of rigour during ischaemia ( Figure 1D) once again being indicative for a lack AOXmediated cardioprotection during I/R.

| AOX is catalytically active in post-anoxic heart mitochondria
One possible explanation for the observed failure of protection could be insufficient AOX expression or its catalytic impairment in the post-ischaemic heart. To test this possibility, we isolated mitochondria from WT and AOX mouse hearts and measured oxygen consumption and ROS production in vitro, and, importantly, confirmed a robust expression of AOX by Western blot ( Figure S1A). As expected, AOX supported succinate-driven respiration in a cyanide-resistant manner ( Figure 2A) at ambient oxygen concentrations. AOX-driven cyanide-resistant respiration was associated with the prevention of mitochondrial hydrogen peroxide production ( Figure 2B). Next, we tested whether AOX catalytic activity depends on a certain oxygen concentration. In other words, we sought to test if AOX loses its catalytic activity underneath a critical threshold of oxygen availability such as seen in the ischaemic heart. We thus measured oxygen consumption and hydrogen peroxide production under different oxygen concentrations ranging from ~21% (ambient air) to full anoxia ( Figure 2C,D). We found that the oxygen consumption in AOX heart mitochondria was slightly higher than in WT across a broad range of oxygen concentrations ( Figure 2C).
Hydrogen peroxide production showed a linear relationship with oxygen availability in WT and was lower in AOX mitochondria ( Figure 2D).
To simulate the situation in the post-ischaemic heart, we challenged heart mitochondria by a 20 minutes period of anoxia (simulated ischaemia, Figure 2E,F). Using this approach, oxygen consumption in WT heart mitochondria was slightly higher than in those from AOX hearts, at all oxygen concentrations ( Figure 2E). In contrast, hydrogen peroxide production, which again showed a linear relationship with oxygen availability, was significantly lower in AOX heart mitochondria ( Figure 2F). To rule out the possibility that the period of anoxia was not sufficient, we replicated the experiments with an extended period of anoxia (30 minutes), which essentially gave the same results

| AOX improves mitochondrial functions 3 weeks after an ischaemic insult
We hypothesized that the restoration of electron flux by AOX and supposed decrease in RET-induced ROS, despite the lack of a measurable advantage during the acute phase, might yet be beneficial for the post-ischaemic remodelling process. We tested this assumption in WT and AOX mice 3 weeks after a transient (45 minutes) ischaemic insult followed by restoration of blood flow (reperfusion). 54 Functional left-heart parameters (P syst and LVDP) measured ex vivo, however, revealed no significant difference between WT and AOX ( Figure 3A,B). Conversely, high-resolution respirometry using permeabilized heart fibres revealed that cI-linked respiration, driven by pyruvate, glutamate and malate (PGM) plus ADP ( Figure 3C), as well as cIV activity driven by ascorbate/TMPD ( Figure 3D I/R in WT hearts but, importantly, this increase was blunted in AOX animals ( Figure 4A). Since ANP has been described as profibrogenic 56 and since excess cardiac collagen synthesis and deposition is known to negatively affect contractile function, 57 we measured transcripts of collagen 1 (Col1a1) and 3 (Col3a1), which we found specifically increased in the AOX I/R group (Figures 4B,C). We also looked at the expression of mitochondrial genes involved in the response to oxidative stress. Superoxide dismutase 2 (Sod2) is a mitochondrial protein that detoxifies superoxide generated as by-product of OXPHOS, by converting it to hydrogen peroxide and molecular oxygen. Sod2 is induced as part of the cell-inherent ROS stress response, 58 which, when lacking, induces cardiac malfunctioning. 59 Transcript levels of Sod2 appeared down-regulated in AOX hearts compared with WT, irrespective of any intervention ( Figure 4D). Uncoupling proteins (UCPs), in contrast, are mitochondrial inner membrane proteins that act as proton leaks, thereby dissipating the proton gradient.
Uncoupling protein 2 (Ucp2) has previously been shown to be instrumental for protection against pressure overload-induced right heart failure. 60 Here, we found Ucp2 specifically up-regulated in the AOX I/R group ( Figure 4E). Finally, heart failure due to apoptotic loss of cardiomyocytes has been long discussed. 61 Bcl-2-associated X protein (BAX) is a pro-apoptotic member of the BCL2 protein family.
Indeed, Bax transcript expression has been negatively correlated with cardiac function whilst Bcl2 transcript expression has positive effects. 62 More specifically, an increased Bax/Bcl2 ratio has been seen in cardiac fibrosis. 63 Of note, we found a significant increase in the ratio of Bax to Bcl2 ( Figure 4F-H), which is indicative for a cellular shift towards a pro-apoptotic phenotype.

| AOX impairs cardiac contractility 9 weeks after an ischaemic insult
The

| AOX fosters extracellular matrix remodelling in the post-ischaemic heart
To shed light on the mechanisms that may underlie the post-ischaemic functional deterioration of AOX I/R hearts, we initiated an unbiased, label-free quantitative proteome approach (data submitted to PRIDE database, identifier PXD014061) and used GOrilla, an online tool designed to discover and visualize enriched gene ontologies in ranked lists 64,65 (Tables S1 and S2). We revealed a pronounced expression of proteins involved in the reorganization of the extracellular matrix as a major feature in the post-ischaemic AOX heart. This is best exemplified by the expression of the protein periostin (POSTN, osteoblast-specific factor). Periostin was up-regulated almost fivefold in post-ischaemic compared with SHAM-operated AOX hearts at 3 weeks, and 17-fold at 9 weeks, whilst the increase in the corresponding WT hearts was <2-and approximately sevenfold after 3 and 9 weeks, respectively.

| D ISCUSS I ON
In the present study, we tested how respiratory restoration by AOX affects the development of I/R injury and adaptive remodelling in the mouse heart. Our data demonstrate that AOX, despite being sufficiently expressed and catalytically active in the healthy and diseased mouse heart, 33,38,40   Despite its ability to maintain respiration and dampen RET-induced ROS in isolated mitochondria, we found that AOX not only fails to provide cardioprotection but actually promotes post-ischaemic maladap- tation. An interesting aspect of our study is that the presence of AOX in the post-ischaemic heart restores cI-driven respiration. In fact, cI may be both, a ROS-producing complex for instance under conditions as seen in the post-ischaemic heart 24,25 or itself a target for ROS damage. 67 Previously, however, it was shown that supplementing rat hearts undergoing ischaemic insults with Ndi1, a single-subunit protein that in Saccharomyces cerevisiae serves as an NADH dehydrogenase, greatly decreases I/R injury and infarct size. 68,69 In the light of this and of our  70 It is important to note, however, that whilst catalase facilitates the turnover of hydrogen peroxide to water and oxygen, AOX prevents the production of superoxide at the impaired ETC and thus acts far upstream.
To provide some indication of the mechanism whereby AOX leads eventually to a maladaptive cardiac remodelling, we studied the expression of a small set of marker genes. Although the findings do not provide a definitive explanation, they reveal valuable clues that should be followed up by a more detailed characterization of the physiological and molecular changes in the AOX I/R model, as now discussed.
First, AOX abrogated the up-regulation of Anp at the transcript level.
ANP was originally found to be up-regulated in the failing heart, irrespective of the underlying cause 71 and was assumed to be compensatory in nature. 72 This view has subsequently been challenged. 73 ANP is now considered a diagnostic marker of maladaptive cardiac remodelling. 55,[74][75][76][77] This notion may need further revision, in the light of our finding of increased expression of Col1a1 and Col3a1 in AOX hearts, despite the blunted Anp response. Whilst collagen decreases wall stress, it also impairs contractile function due to increased stiffness, and accompanies the development of heart failure. 57 One consequence of Anp signalling, shown previously in rat pulmonary arterial smooth muscle cells, is the inhibition of TGFβinduced extracellular matrix components. 78 We suggest that this might account for the observed up-regulation of periostin, consistent with another study where heart remodelling upon stress was associated with progressive extracellular matrix remodelling. 79 However, the role of periostin in cardiac repair remains unclear: whilst it has been found to initiate cell cycle re-entry of adult cardiomyocytes upon stress 80 it is also associated with myocardial fibrosis in some forms of heart failure. [81][82][83] Consistent with the idea that AOX interferes with a respiratory stress signal related to mitochondrial ROS, we observed downregulation of the oxidative stress marker Sod2, but up-regulation of Ucp2, previously observed in conditions of cardiovascular stress. 84 In conclusion, we infer that mitochondrial respiratory restoration and the presumed decrease of mitochondrial ROS by AOX in the post-ischaemic heart are not sufficient to confer cardioprotection.
Instead, AOX expression interferes with adaptive organ remodelling leading to contractile failure, implying an essential role for ETCderived stress signals in this process.

ACK N OWLED G EM ENTS
The authors thank Pierre Rustin for valuable discussions, and Jana Meisterknecht and Nadine Woitasky for excellent technical assis-

CO N FLI C T O F I NTE R E S T
MSz has financial interests in developing therapeutics based on Ciona intestinalis AOX. All other authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data needed to evaluate the conclusions in the paper are present in the paper and/or the supplements. Additional data related to this paper may be requested from the authors.