The histone H3K9 methyltransferase SUV39H links SIRT1 repression to myocardial infarction

Myocardial infarction (MI) dampens heart function and poses a great health risk. The class III deacetylase sirtuin 1 (SIRT1) is known to confer cardioprotection. SIRT1 expression is downregulated in the heart by a number of stress stimuli that collectively drive the pathogenesis of MI, although the underlying mechanism remains largely obscure. Here we show that in primary rat neonatal ventricular myocytes (NRVMs), ischaemic or oxidative stress leads to a rapid upregulation of SUV39H, the mammalian histone H3K9 methyltransferase, paralleling SIRT1 downregulation. Compared to wild-type littermates, SUV39H knockout mice are protected from MI. Likewise, suppression of SUV39H activity with chaetocin attenuates cardiac injury following MI. Mechanistically, SUV39H cooperates with heterochromatin protein 1 gamma (HP1γ) to catalyse H3K9 trimethylation on the SIRT1 promoter and represses SIRT1 transcription. SUV39H augments intracellular ROS levels in a SIRT1-dependent manner. Our data identify a previously unrecognized role for SUV39H linking SIRT1 trans-repression to myocardial infarction.

A number of factors, including senescence, unhealthy dietary choices and life styles, and surges in metabolic disorders, all contribute to the epidemic of ischaemic heart disease (IHD) across a wide spectrum of ages and ethnic groups worldwide. Despite decades of vigorous basic and clinical research, IHD remains the leading cause of death in major industrialized nations 1 . Myocardial infarction (MI), spontaneous or procedural, in patients with IHD, serves to further depress heart function, frequently induces deleterious cardiac remodelling, aggravates organ ischaemia, and is associated with increased risk of heart failure and mortality 2 . Clearly, deciphering the mechanisms that drive the pathogenesis of MI holds the key to stabilizing heart function and reducing MI-related deaths in IHD patients.
During MI pathogenesis, a host of stress stimuli inflict injuries in cardiomyocytes. Ischaemia depletes cellular ATP causing mitochondrial dysfunction and accelerating reactive oxygen species (ROS) generation. Elevated ROS levels, in turn, interfere with oxidative phosphorylation and dampen mitochondrial function creating a self-amplifying vicious cycle to drive the death of cardiomyocytes 3 . By comparison, anti-oxidant therapies have shown varied but promising effects in alleviating MI-induced heart injury in model animals and in IHD patients 4,5 . The class III protein deacetylase SIRT1 has been documented to confer cardioprotection in the context of IHD both in vivo and in vitro at least in part by suppressing ROS generation [6][7][8][9] . In keeping with these observations, SIRT1 expression levels and, by extension, promoter polymorphisms have been found to correlate with acute MI in model animals and in humans [10][11][12] . The underlying mechanism contributing to altered SIRT1 transcription in ischaemic cardiomyocytes is not understood.
The epigenetic machinery plays a pivotal role in regulating mammalian gene expression. Differential histone modifications represent a prototypical epigenetic mechanism wherein a combination of several modified residues on histone tails barcodes the promoter region and predicts the transcriptional outcome of a specific gene. A simplistic model has been proposed that, confirmed more recently through genome-wide sequencing analyses, trimethylation of histone H3 lysine 4 (H3K4Me3) heralds transcriptional activation, whereas trimethylation of H3 lysine 9 (H3K9Me3) leads to trans-repression. SUV39H, short for suppressor of variegation 3 (9) homologue, is the sole enzyme responsible for laying down the H3K9Me3 mark on the chromatin 13 . SUV39H was initially discovered as essential for silencing gene expression in the heterochromatin region. It remains obscure how SUV39H may contribute to gene regulation in the euchromatin region and more importantly, to the pathogenesis of MI.
We report here that upregulation of SUV39H accompanies SIRT1 downregulation in the heart during MI. Genetic deletion or pharmaceutical inhibition of SUV39H ameliorates cardiac injury following MI in mice. SUV39H modulates ROS levels in the heart in a SIRT1-dependent manner. Therefore, SUV39H links SIRT1 trans-repression to myocardial infarction providing rationale for targeting SUV39H1 in the development novel therapeutic strategies to treat ischaemic heart disease.

Results
SUV39H levels are upregulated in vivo and in vitro. We started off by examining the levels of SUV39H in the myocardium following ischaemia challenge in mice. As shown in Fig. 1a, there was a quick increase in Suv39h1 mRNA levels in ischaemic hearts in mice 12 h after the ligation of left anterior descending (LAD) artery procedure, which was sustained and further augmented at 4 and 7 days. Suv39h2mRNA levels were also upregulated in ischaemic hearts at 12 h but receded at 4 and 7 days. Western blotting showed a similar trend in the protein levels of Suv39h1 and Suv39h2 following MI (Fig. 1b). We also examined the expression levels of heterochromatin protein 1 (HP1), the binding partner for SUV39H during heterochromatinzation and gene trans-repression 13 . The Hp1a and Hp1b isoforms were not significantly altered in the heart throughout the observation window of 7 days following MI, but the Hp1g isoform was upregulated at day 4 and day 7 post MI in ischaemic heart tissues ( Supplementary Fig. 1).
In primary rat neonatal ventricular myocytes (NRVMs), ischaemic stress (Fig. 2a,b), simulated by the addition of antimycin A plus 2-deoxy glucose (AA þ 2-DG) to deplete intracellular ATP 14 , and oxidative stress (Fig. 2c,d), induced by the addition of hydrogen peroxide (H 2 O 2 ) (refs 15,16), both stimulated the expression of Suv39h1 and, to a lesser extent, Suv39h2. Similar observations were made in an immortalized rat myoblast cell line H9C2 ( Supplementary Fig. 2). Taken together, we conclude that SUV39H levels are responsive to MI-relevant stimuli in the heart and cultured cardiomyocytes indicative of a potential role for SUV39H in MI pathogenesis.
Since deficient ROS elimination plays a role in MI pathogenesis, we evaluated the effect of SUV39H deficiencies on ROS levels in vivo. As shown in Fig. 3f, dihydroethidium (DHE) staining demonstrated that ROS accumulation was much stronger in WT hearts than either Suv39h1 À / À or Suv39h2 À / À hearts. Furthermore, SUV39H deficiencies significantly attenuated the down-regulation of several anti-oxidant genes including superoxide dismutase 1 (Sod1), superoxide dismutase 2 (Sod2) and catalase (Cat) in the myocardium following MI (Fig. 3g,h).
To probe a myeloid lineage-specific role for SUV39H in myocardial infarction, we performed bone marrow transplantation assay. To this end, MI was induced in wild-type mice receiving bone marrow from either wild-type, Suv39h1 À / À or Suv39h2 À / À mice. Post-mortem autopsy revealed that there was no significant difference in terms of infarct size; nor was there any improvement in post-MI heart function ( Supplementary Fig. 5). Consistently, expression levels of anti-oxidant genes and cardiac   Suv39h1 h2 -/-MI (N=6)  ROS levels were not significantly altered among three different chimeric mice ( Supplementary Fig. 6). Taken together, these data strongly support a role for SUV39H in promoting MI in mice.
SUV39H inhibition attenuates MI in mice. Next, we sought additional evidence that SUV39H might play a role in the pathogenesis of MI. To this end, male C57/BL6 mice were injected peritoneally with chaetocin, a reported inhibitor of SUV39H activity 17 , or DMSO prior to the LAD procedure. Histone methyltransferase assays performed using heart tissue homogenates indicated that chaetocin injection markedly inhibited the activities of both SUV39H1 and SUV39H2 in mice ( Supplementary Fig. 7). Chaetocin administration significantly improved the survival rate of mice (Fig. 4a) and reduced infarct size (Fig. 4b,c). Accompanying these changes there was a reduction in Mmp9 expression, as measured by qPCR and western blotting ( Supplementary Fig. 8A,B), and cardiomyocyte apoptosis, as measured by TUNEL assay ( Supplementary Fig. 8C), in the heart following MI. DHE staining showed that cardiac ROS levels were suppressed by chaetocin administration (Fig. 4d). Furthermore, chaetocin administration was able to rescue the repression of anti-oxidant genes by MI (Fig. 4e,f). These protective effects afforded by chaetocin cumulatively contributed to improved post-MI heart function ( Fig. 4g,h). We also evaluated the effect of chaetocin on cardiac remodelling 4 weeks after infarction. Picrosirius red and Masson's trichrome staining showed that chaetocin ameliorated interstitial fibrosis ( Supplementary Fig. 9A). qPCR and western blotting measurements confirmed that chaetocin pre-empted the induction of pro-fibrogenic genes including type I collagen and alpha smooth muscle actin (a-SMA) in MI hearts ( Supplementary  Fig. 9B,C). Collectively, these data argue that SUV39H inhibition could not only reduce infarction but also block adverse cardiac remodelling in mice.
SUV39H represses SIRT1 transcription. Next, we tackled the mechanism whereby SUV39H may contribute to MI pathogenesis. SIRT1 has been assigned a cardiac-protective role and its expression is inversely correlated with acute MI in humans. Of interest, we observed an inverse correlation between SUV39H expression and SIRT1 expression in mice ( Fig. 1). Thus, we proposed that SUV39H might promote MI by targeting SIRT1. To verify this hypothesis, we performed the following experiments. Reporter assays ( Supplementary Fig. 10A) showed that SUV39H1 was able to repress the activity of the shortest SIRT1 promoter ( À 115/ þ 54). In a parallel experiment, we found that ecotopic (Myc-tagged) SUV39H could indeed bind to and consequently stimulate the accumulation of trimethylated H3K9 on the reporter construct transfected into H9C2 cells (Supplementary Fig. 10B). Chromatin immunoprecipitation (ChIP) assay showed that following AA þ 2-DG stimulation, SUV39H1 quickly assembled on this region of the Sirt1 promoter but not the Gapdh promoter in both NRVMs (Fig. 5a) and H9C2 cells ( Supplementary Fig. 11A); by comparison, SUV39H2 was recruited with much weaker affinity. Of interest, there was no significant enrichment of either SUV39H1 or SUV39H2 on the Sod1 promoter despite strong trans-repression by AA þ 2-DG treatment, indicating that SUV39H might regulate Sod1 transcription indirectly. Paralleling SUV39H binding, trimethylated histone H3K9 started to accumulate on the Sirt1 promoter while acetylated H3K9, an antagonistic marker to H3K9Me3, departed from the Sirt1 promoter consistent with its trans-repression. Similar observations were made with H 2 O 2 treatment (Fig. 5b, Supplementary Fig. 11B). We did not detect significant enrichment of H3K9Me2, another prominent histone modification associated with repressed chromatin, on the Sirt1 promoter suggesting that H3K9Me2 might not play a role in SIRT1 transrepression in the context of MI. Unlike HP1a and HP1b, HP1g exhibited enhanced binding to the Sirt1 promoter (but not to the Gapdh promoter or the Sod1 promoter) in NRVMs treated with either AA þ 2-DG or H 2 O 2 ( Supplementary Fig. 12). In keeping with this observation, HP1g also formed a complex with SUV39H1 and SUV39H2 (albeit with lower affinity than SUV39H1) on the Sirt1 promoter as evidenced by Re-ChIP assay ( Supplementary Fig. 13).
More importantly, we discovered that SUV39H1 and SUV39H2 were recruited to the Sirt1 promoter and interacted specifically with HP1g in the infarcted myocardium in mice (Fig. 5c, Supplementary Fig. 14), suggesting that SUV39H probably serves as a common link to SIRT1 trans-repression. Finally, reporter assay confirmed that SUV39H1 not only dosedependently repressed the Sirt1 promoter but also enhanced the trans-repression of Sirt1 promoter by either AA þ 2-DG ( Supplementary Fig. 15A) or H 2 O 2 ( Supplementary Fig. 15B). Again, of the three HP1 isoforms only HP1g repressed the Sirt1 promoter in the presence of either AA þ 2-DG or H 2 O 2 (Supplementary Fig. 16A-C). Using ChIP assay, we confirmed that ectopically introduced (HA-tagged) HP1g could indeed occupy the SIRT1 reporter construct in H9C2 cells ( Supplementary Fig. 16D). Therefore, we conclude that SUV39H represses Sirt1 transcription in cardiomyocytes.
SUV39H silencing or inhibition restores SIRT1 transcription. We next employed the following strategies to confirm that SUV39H is indispensible for SIRT1 trans-repression in the context of MI. siRNA-mediated knockdown of SUV39H1 or SUV39H2 abrogated H3K9Me3 assembly and restored H3K9Ac levels on the Sirt1 promoter despite the stimulation by AA þ 2-DG in cardiomyocytes (Fig. 6a, Supplementary Fig. 17A). Accordingly, SUV39H knockdown normalized Sirt1 expression at both mRNA ( In SUV39H1 and SUV39H2 deficient mice, we observed more H3K9Ac and less H3K9Me3 on the Sirt1 promoter as opposed to WT mice ( Fig. 6g), which probably contributed to higher SIRT1 levels in infarcted hearts (Fig. 6h,i). Consistent with the prior observations that HP1g interacted with SUV39H to repress Sirt1 transcription, silencing of HP1g, but not HP1a or HP1b, with siRNA led to partial restoration of Sirt1 expression in NRVMs exposed to either AA þ 2-DG or H 2 O 2 ( Supplementary Fig. 19). HP1g silencing also led to the normalization of anti-oxidant genes accompanying erasure of trimethyl H3K9 and restoration of acetyl H3K9 on the Sirt1 promoter in both AA þ 2-DG ( Supplementary Fig. 20) and H 2 O 2 -treated ( Supplementary Fig. 21) NRVMs.
SUV39H regulates ROS generation in a SIRT1-dependent manner.
Having established that SUV39H promotes ROS generation and SIRT1 trans-repression in the process of myocardial infarction, we asked whether the ability of SUV39H to regulate ROS levels in cardiomyocytes depends on SIRT1. As shown in Fig. 8a and Fig. 8b, depletion of SUV39H blocked the downregulation of Sod1, Sod2 and Cat by AA þ 2-DG treatment, which was preempted by siRNA-mediated SIRT1 silencing. Consistent with gene expression levels, DHE staining showed that SUV39H silencing prevented the induction of intracellular ROS generation by AA þ 2-DG, which was abrogated by SIRT1 inhibition (Fig. 8c). Similar results were obtained in cardiomyocytes treated with H 2 O 2 ( Supplementary Fig. 24). A specific SIRT1 inhibitor EX-527 achieved similar effects ( Supplementary Figs 25 and 26).
In the second set of experiments, inhibition of SUV39H activity by chaetocin alleviated the decrease of anti-oxidant gene expression by AA þ 2-DG (Fig. 8d,e) and the increase in intracellular ROS levels (Fig. 8f), both of which were neutralized by SIRT1 knockdown. Again, we were able to confirm that the anti-oxidative effects endowed by chaetocin were completely lost without endogenous SIRT1 activity in H 2 O 2 -treated NRVMs ( Supplementary Fig. 27). Likewise, EX-527 treatment also negated the effects of chaetocin in both AA þ 2-DG ( Supplementary  Fig. 28) and H 2 O 2 -treated ( Supplementary Fig. 29) NRVMs. Finally, we analysed whether this SUV39H-SIRT1 axis might be functional in mice. To this end, SUV39H knockout mice or wild-type littermates were injected via tail vein siRNA targeting SIRT1 or scrambled siRNA prior to the MI procedure. SIRT1 silencing almost completely blanketed the reduction in infarction size (Fig. 9a) and the improvement of post-MI heart function (Fig. 9b,c) as a result of SUV39H deficiency. qPCR and western blotting analyses indicated that in the absence of SIRT1, anti-   (a-c) Primary NRVMs were transfected with SUV39H siRNA or random siRNA (SCR) followed by treatment with AA þ 2-DG. ChIP assays were performed with indicated antibodies (a). SIRT1 expression was examined by qPCR (b) and western blotting (c). (d-f) Primary NRVMs were transfected with SUV39H siRNA or random siRNA (SCR) followed by treatment with H 2 O 2 . ChIP assays were performed with indicated antibodies (d). SIRT1 expression was examined by qPCR (e) and western blotting (f). (g-i) MI was induced in Suv39h1knockout (h1 À / À ) mice, Suv39h2 knockout (h2 À / À ) mice or wide-type (WT) littermates by LAD. ChIP assays were performed with indicated antibodies (g).        oxidant genes were significantly downregulated (Fig. 9d,e). In addition, Sirt1 depletion re-instated the intracellular ROS levels to those observed in wild-type hearts (Fig. 9f). Similarly, we found that the post-MI benefits associated with SUV39H deficiency in mice were largely gone by EX-6527 injection ( Supplementary Fig. 30). Of note, the cardioprotective effects of chaetocin was also dependent on SIRT1 since Sirt1 depletion (Fig. 10) or inhibition ( Supplementary Fig. 31) abrogated the reduction of infarction size, the normalization of heart function, the changes in anti-oxidant gene expression, and the suppression of ROS levels induced by chaetocin in post-MI mice. Collectively, these data demonstrate that SUV39H regulates ROS generation in a SIRT1-dependent manner.

Discussion
During MI pathogenesis, a host of insults collectively inflict damages on cardiomyocytes contributing to the loss of heart function, acute cardiac arrest and death. We present evidence that the histone H3K9 trimethyltransferase SUV39H links SIRT1 trans-repression to myocardial infarction. Using H3K9 trimethylation as a readout, Greiner et al. 17 have discovered chaetocin as a specific SUV39H inhibitor. Chaetocin has since been shown to confer protection against hepatoma 18 and leukaemia 19 in mice although the mechanisms seem to be multifold. It remains ambiguous whether the effects of chaetocin are entirely mediated through SUV39H and the specificity of chaetocin has been called into question 20 . Similar to our study, Schweizer et al. 21 have reported that chaetocin administration is associated with neuroprotection in an in vitro model of cerebral ischaemia. In addition, these authors demonstrated that chaetocin treatment activates the transcription of neurotrophin genes (BDNF) at least in part by stimulating H3K9 acetylation across the BDNF promoters. These data echo our own observation that chaetocin protects ischaemia-challenged cardiomyocytes by epigenetically activating Sirt1 transcription (Fig. 7) and suggest that at least in the context of ischaemia-induced cell damages, chaetocin may be equivalent to SUV39H silencing/depletion. More recently, Cotman and colleagues have identified SUV39H-mediated H3K9 trimethylation as a critical link between ageing and memory loss in mice 22 . Of importance, this group of investigators exploited a more selective SUV39H inhibitor called ETP69 to show that SUV39H inhibition significantly improved learning of complex task in aged mice. Not coincidently, MI occurs predominantly in the aged population. In the light of our finding that chaetocin injection ameliorates the loss of heart function following MI in mice (Fig. 4), these observations collectively allude to a scenario wherein SUV39H modulates ageing-related pathophysiological processes by epigenetically regulating key transcription events. Indeed, a report by Liu et al. 23 indicates that SUV39H1 deficiency circumvents premature ageing in Zmpste24 À / À mice that model the Hutchinson-Gilford progeria syndrome in humans, offering strong support for the argument that SUV39H activity promotes ageing and ageing-associated organ dysfunctions such as dementia and MI.
Curiously, Schweizer et al. and Liu et al. only examined systemic levels of H3K9Me3, which from a transcriptional perspective bears little, if any, relevance to the actual molecular proceedings underlying disease pathogenesis. A couple of investigations profiling genome-wide patterns of H3K9Me3 enrichment have offered conflict but insightful snapshots at the role of SUV39H in different pathobiological processes. Using hippocampus from acutely stressed Sprague-Dawley rats, Hunter et al. showed while the transposable and repeated regions saw a significant increase in H3K9Me3 binding upon stress, little change was found for H3K9Me3 in the gene body (for example, promoter regions) 24 , suggesting that SUV39H most likely plays a very limited role regulating transcription in the euchromatin region. In contrast, a group led by Casaccia has reported that during oligodendrocyte differentiation, a vast majority of H3K9Me3 can be mapped to the gene body encoding proteins involved in neuronogenesis 25 . We show here that SUV39H expression/activity is intimately connected to Sirt1 promoterspecific H3K9Me3 levels. In addition, we show that HP1g, but not HP1a or HP1b, co-regulates SIRT1 trans-repression with SUV39H. Unlike HP1a or HP1b, HP1g is exclusively situated in the euchromatin region 26 . These data suggest that SUV39H may regulate euchromatin transcription in a tissue-and contextspecific manner likely guided by HP1g. ChIP-seq analyses performed in healthy and infarcted myocardium may help clarify this issue and assign a more precise role for SUV39H in MI pathogenesis.
An intriguing observation here is that Suv39h1 deficiency seems to confer bigger advantage to mice in terms of survival, infarct size, post-MI heart function and ROS levels than Suv39h2 deficiency (Fig. 3), all of which could probably be attributed to the fact that SUV39H1 acts as a more potent repressor of Sirt1 transcription (Figs 5 and 6). In response to MI-relevant stress stimuli, the expression of SUV39H1 and SUV39H2 was altered with distinctive kinetics: SUV39H1 upregulation was more persistent while SUV39H2 upregulation was more transient (Figs 1 and 2). In addition, HP1g appears to preferentially interact with SUV39H1 on the Sirt1 promoter ( Supplementary Fig. 10), which may also explain why SUV39H1 played a more dominant role in SIRT1 transrepression. It has been previously shown that SUV39H1 and SUV39H2 redundantly maintain global H3K9Me3 levels, which are completely erased in SUV39H1/H2 double knockout (DN) mice 27 . It remains to be elucidated how SUV39H1 and SUV39H2 differentially bind to target genes to regulate MI pathogenesis. SUV39H2 only shares B60% amino-acid sequence identity with SUV39H1 with a highly variable N-terminal domain and the structural differences may determine target promoter selection by SUV39H1 and SUV39H2 (ref. 28). Alternatively, SUV39H1 and SUV39H2 may have non-overlapping lists of binding partners, which in turn may steer them to different chromatin regions to regulate transcription, because of distinct structures. A recent report by the Jenuwein group provides some support for this model: SUV39H-depednent heterochromatin formation relies on the sequence-specific transcription factors PAX3 and PAX9 that bind preferentially to the intergenic satellite repeats 29 . Since the PAX family of proteins including PAX3 and PAX9 possess well-defined functions in regulating a range of pathophysiological processes in the cardiovascular system 30 , we suspect a similar scheme may exist in the euchromatin region to control SUV39H target selection.
At the cellular level, we focused on the role of SUV39H in regulating ROS levels in cardiomyocytes, which is unlikely to reflect in full the realm that SUV39H reigns with regards to MI pathogenesis. For instance, lymphocyte mediated immune response plays a pivotal role in regulating MI pathogenesis encompassing injury, healing and remodelling 31 . By directly regulating the transcription of lineage specification genes, SUV39H1 has been show to strike a balance between different T-lymphocyte subsets 32 . In addition, SUV39H1 has also been shown to impact B-lymphocyte-dependent humoral immune response by modulating class switch recombination 33 . These observations highlight a potential role for SUV39H in shaping up the immune response during MI pathogenesis. Furthermore, a string of recent reports have put SUV39H at center stage in regulating stem cell phenotype [34][35][36] , an increasingly promising target for improving post-MI heart function. Of interest, bone marrow transplantation experiments ( Supplementary Figs 7 and 8) indicate that myeloid-derived SUV39H is unlikely to play a major role in MI pathogenesis. SUV39H has been reported to repress the transcription of p21 (ref. 37) and PML-RAR 38 target genes in macrophages although relatively little is known regarding the systemic regulatory role of SUV39H in myeloid cells in the context of MI. Further confounding the link between SUV39H and SIRT1 is the inherent weakness of some of our methodologies. In addition to the systemic, instead of lineagespecific, knockout models, we have used tail vein injection to deliver siRNAs, which may lodge in tissues and organs other than the heart such that a definitive relationship between SUV39H and SIRT1 cannot be taken for granted. Clearly, the cell-specific roles of SUV39H in MI pathogenesis should be revisited taking into account of the aforementioned possibilities when conditional knockout animals are made available.
Previous investigations have examined the functional relationship between SUV39H1 and SIRT1 in different contexts. For instance, Murayama et al. 39 have reported that SIRT1 forms a co-repressor complex with SUV39H1 to repress the transcription of rRNA in response to intracellular energy status. SIRT1 has also been found to cooperate with SUV39H1 to modulate the chromatin structure surrounding the leukaemic genes and prevent MLL-rearranged leukaemia 40 . Vaquero et al. have shown that SIRT1 could directly modify (deacetylate) and thus stabilize SUV39H1 to maintain the integrity of heterochromatin 41,42 . Our finding adds an additional layer to the intricate SIRT1-SUV39H1 interplay. It would be of key interest to see how these different modes of actions dynamically contribute to the regulation of ischaemic heart disease.
In summary, our data demonstrate the potential of SUV39H deletion/inhibition in alleviating damages associated with MI. A genomewide search for more SUV39H targets, using both RNA-seq and ChIP-seq techniques, in different types of cells in the context of MI would hopefully clear the lingering uncertainties and render a more rationalized decision to target SUV39H for drug development.
Mice. Suv39h1 knockout and Suv39h2 knockout mice have been previously described 27 . All protocols were approved by the intramural Committee on Ethic Conduct of Animal Studies of Nanjing Medical University. MI was induced by permanent ligation of the LAD coronary artery in male, 8-week-old mice. Briefly, the mice were anaesthetized with a mixture of ketamine (120 mg kg À 1 ) and xylazine (6 mg kg À 1 ), the hearts were exposed and the LAD was ligated with a 6-0 prolene suture. The control mice were sham operated wherein the ligature around the LAD was not tied. Post-MI heart functions were evaluated by echocardiography (GE Vivid 7 equipped with a 14-MHz phase array linear transducer, S12, allowing a 150 maximal sweep rate). Myocardial infarct size was determined by Evans blue/ triphenyltetrazolium chloride (TTC) staining 46 . Briefly, the hearts were perfused with saline on a Langendorff system to remove blood and then stained with with 1% Evans blue followed by incubation in 1% TTC-Tris solution) for 15 min at 37°C. For ROS measurements, frozen heart sections (4 mm) were stained with DHE (2 mmol l À 1 , Sigma) in a light-protected humidified chamber at 37°C for 15 min. The slides were visualized by co-focal fluorescence microscopy (Zeiss). To examine cardiomyocyte apoptosis, frozen heart sections (4-mm) were stained using a TUNEL kit (Roche) and the nuclei were counter-stained with DAPI (Sigma). The slides were visualized by co-focal fluorescence microscopy (Zeiss).
RNA isolation and real-time PCR. RNA was extracted with the RNeasy RNA isolation kit (Qiagen). Reverse transcriptase reactions were performed using a SuperScript First-strand Synthesis System (Invitrogen). Real-time PCR reactions were performed on an ABI Prism 7500 system. Primers and Taqman probes used for real-time reactions were purchased from Applied Biosystems.
Statistical analysis. Animal numbers and sample sizes reflected the minimal number needed for statistical significance based on power analysis and prior experience. No data were excluded from any of the experiments. Grouping was performed in a randomized manner. Blinding were not performed because it was not appropriate for the types of animal groups used here, or the types of comparisons used between groups. One-way analysis of variance with post hoc Scheffe analyses were performed using an SPSS package. Mouse survival data was analysed by the Kaplan-Meier log-rank test.
Data availability. All data generated or analysed during this study are included in this published article and its Supplementary Information files, or are available from the corresponding author on reasonable request.