Integrative analysis of differentially expressed genes and miRNAs predicts complex T3-mediated protective circuits in a rat model of cardiac ischemia reperfusion

Thyroid hormone (T3) dyshomeostasis in the cardiac ischemia-reperfusion (IR) setting negatively impacts on mitochondria function and extracellular matrix remodeling. The modulation of cardiac miRNAs may represent the underlying molecular mechanisms, but a systems biology perspective investigating this critical issue in depth is still lacking. A rat model of myocardial IR, with or without an early short-term T3-replacement, was used to predict putative T3-dependent miRNA-gene interactions targeted to mitochondria quality control and wound healing repair. As evidenced by mRNA and miRNA expression profiling, the T3 supplementation reverted the expression of 87 genes and 11 miRNAs that were dysregulated in the untreated group. In silico crossing and functional analysis of the T3-associated differentially expressed transcripts, identified a signature of interconnected miRNA-gene regulatory circuits that confer resistance to noxious cascades of acute stress. In this network the T3-down-regulated Tp53, Jun and Sp1 transcription factors emerge as critical nodes linking intrinsic cell death and oxidative stress pathways to adverse remodeling cascades. The data presented here provide a novel insight into the molecular basis of T3 cardioprotection in the early post-IR phase and highlight the contribution of a previously unappreciated complex T3-regulatory network that may be helpful in translating T3 replacement into clinical practice.

Heart disease following acute myocardial infarction represents a serious health problem and a major cause of death worldwide. Although timely reperfusion is mandatory to support cell function and to remove potentially noxious by-products of cellular metabolism, it also elicits pathogenetic processes that exacerbate injury due to ischemia per se 1,2 . Cardioprotective strategies targeting the activation of multiple pathologic pathways in the early stages of the post-IR wound healing process must be adopted to enhance cell resistance to death, to reduce adverse remodeling and to improve patient prognosis.
Thyroid hormones exert a fundamental role in cardiovascular system homeostasis 3 . Not surprisingly, even a small alteration of the thyroid function correlates positively with coronary risk factors and increased mortality for heart disease 4,5 . In patients with acute myocardial infarction (AMI) a condition of reduced T3 plasma level, known as low T3 state (lowT3S), represents a risk of cardiac disease progression and higher mortality 6,7 . Along this line, increasing clinical and experimental findings prompt for a cardioprotective role of lowT3S correction through T3 administration in cardiac disease 8,9 . However the clinical implementation of this therapeutic option is hampered by the residual concern of adverse effects. A comprehensive analysis of the cardioprotective networks triggered by T3 replacement in the context of post-IR signaling activation may help developing new therapeutic strategies and encourage pilot clinical trials.

Results
T3 replacement promotes post ischemic cardiac recovery without altering physiological parameters. We previously observed that about 30-40% of rats subjected to myocardial IR exhibited a phenomenon similar to the lowT3S observed in patients after AMI, i.e., an isolated reduction of serum T3 in the absence of overt thyroid disease. In these lowT3S rats we demonstrated that 48 h treatment (started 24 h after IR) with slightly supraphysiological dose of T3 (6 μg/kg/day) improved the post ischemic recovery of cardiac function at 3 days, although increasing heart rate 15,16 .
In the present study, to better mimic a physiological condition with a view to a translational perspective, we evaluated the cardioprotective effects of a new T3 dose that was sufficient to correct the post ischemic lowT3S without inducing other systemic changes, i.e. a T3 replacement dose. For this purpose, 24 h after induction of cardiac IR, rats were subjected to 48 h infusion with T3 at 3 μg/Kg day (IRT3 group) or its vehicle (IR group), functional and molecular results were compared with Sham operated rats used as control.
To validate our model as a preclinical study, we first needed to assess the safety and efficacy of the replacement dose. Sham, IR and IRT3 groups had similar plasma levels of free thriiodothyronine (FT3) at baseline (2.63 ± 0.22 pg/ml, 2.52 ± 0.21 pg/ml, 2.59 ± 0.24 pg/ml respectively). Consistent with the human disease, at 3 days post-IR, only the IR rats showed a significant reduction of FT3 indicative of lowT3S, while the IRT3 group maintained FT3 level comparable to those of Sham rats (Table 1). Moreover, T3 supplementation did not result in alteration of body weight, heart rate, respiration rate or body temperature that remained comparable among groups (Table 1). To assess the effect of the T3 replacement dose on heart function, we have evaluated parameters of heart geometry and contractile performance in diastole and systole by ultrasound imaging 3days after IR. As summarized in Table 1 no differences among group were observed in left ventricle (LV) end diastolic diameter or in the ratio of early to late ventricular filling velocities (E/A ratio). Instead T3 replacement resulted in better preserved LV end systolic diameter, ejection fraction and fractional shortening (Table 1). Taken together, these data attest the safety of the T3 replacement strategy and its effectiveness on the recovery of the post-ischemic LV contractile performance.

T3 replacement associates with rescue of the post ischemic gene transcriptional changes.
Cardiac mitochondria and interstitium remodeling seem emerging targets of T3 mediated cardioprotection 15, 16 . Here we analyze the expression of two focused panels of genes involved in the diverse cellular functions of mitochondrial biology and fibrosis signaling at 3 days after IR to gain a more thoroughly insight into the molecular effects of T3 early replacement on mitochondrial quality control and ECM homeostasis (see Supplementary Table S1 for gene list). With respect to Sham, the IR rats showed altered expression of 87 transcripts out of 150 total genes that gave an evaluable signal in the mitochondria profiler array (MPA) and fibrosis profiler array (FPA) ( Tables 2 and 3). This post-IR expression signature was characterized by 18 up-regulated and 19 down-regulated genes within the MPA (Table 2), and 50 up-regulated genes within the FPA (Table 3). As shown in Table 2, the up-regulated transcripts within the MPA includes genes of the intrinsic apoptosis pathway, regulators of mitochondrial membrane polarization and potential, as well as mediators of mitochondrial metabolites and small solute import. The IR down-regulated transcripts of the MPA includes genes essential for mitochondrial replication and function as well as for the transport of metabolite and ions ( Table 2). The up-regulated transcripts of the FPA comprise genes encoding ECM remodeling enzymes, Tgfb signaling molecules and inflammatory cytokines, along with additional genes pivotal in fibrosis (   FPA, demonstrates that a broad pattern of gene expression relevant to both mitochondrial function and wound healing processes is altered under a post-IR lowT3S. T3 replacement was associated to a significant differential modulation of 32 genes out of the 37 dysregulated transcripts within the MPA (Fig. 1A, Table 2), and 36 genes out of the 50 up-regulated transcripts within the FPA ( Fig. 2A, Table 3), accounting for 86% and 72% of genes altered by the IR procedure in the MPA and FPA respectively. These results suggested that T3 replacement rescued the post-IR transcriptional changes.
T3 replacement regulates the post ischemic myocardial gene expression signature to maintain mitochondrial function. To determine the biological and functional implications of these T3-dependent gene expression changes, we performed a functional enrichment analysis of the 32 plus 36 T3 differentially expressed genes (T3DE-genes) using DAVID annotation tool followed by data validation.
Within the MPA, the T3DE-genes were enriched into three main groups of biological processes implicated in (I) mitochondrial mediated cell death, (II) mitochondrial organization and quality control and (III) mitochondrial solute/metabolite transport (Fig. 1B). In line with the cardioprotective role of T3 replacement, almost all the genes involved in intrinsic apoptosis, p53-driven stress pathways, induction of mitochondrial depolarization and permeability transition were down-regulated by T3 (i.e Bak, Bid, Sfn, Pmaip1, Bbc3 and Tp53); while the T3 up-regulated genes were enriched in pathways that favor mitochondrial biogenesis and fusion (i.e. Pgc1α, Mfn1, Mfn2), antioxidant response to reactive oxygen species (i.e. Sod1 and Sod2), mitochondrial transport of metabolites necessary for ATP production and maintenance of mitochondrial membrane potential (i.e. Cpt1b, Cpt2; Slc25a12, Slc25a13) (Fig. 1B).
Data validation through qRT-PCR confirmed restored expression of relevant genes involved in p53-dependent proapoptotic cell death (Tp53, Bax, Bnip3), mitochondrial quality control (Ppargc1a, Sod2, Mfn2) and mitochondrial transport (Cpt2, Slc25a20) in the IRT3 group (Fig. 1C). To further support the array results, we showed that T3 infusion normalized the altered protein level of Tp53 and Sod2 that are key regulators of cell fate and superoxide detoxification (Fig. 1D). In line, we found an increased clearance of mitochondrial superoxide in the IRT3 group compared to IR (Fig. 1E). Globally, these findings suggest that T3 replacement is associated with activation of protective pathways that reduce cell death while enhancing mitochondrial quality control and function. Accordingly, spectrophotometric determination of oxidative phosphorilation (OxPhos) activities showed multiple defects of the mitochondrial respiratory chain in mitochondria isolated from myocardial tissue of IR rats compared to Sham, with significant reduction of CI (NADH:ubiquinone reductase) CI + CIII (NADH cytochrome c reductase) and CIV (cytochrome c oxidase) (Fig. 1E). A less impacted mitochondrial function was instead recorded in mitochondria from IRT3 group with a better preserved bioenergetics corroborating the protective role of T3 replacement (Fig. 1E).

T3 administration associates with rescue of the post-ischemic miRNA expression profiling.
Recently we and others have suggested miRNAs as intriguing novel players in the multi-level system of T3-dependent cardioprotection 15,16,19,20 . Here we exploited the miRNA profiling by next generation sequencing (NGS) to systematically characterize the miRNAs involved in the beneficial cardiac effects of T3 administered after IR (see material and method for sample details). In the IR group we identified a total of 102 dysregulated miRNAs (42 down-and 60 up-regulated) that met the criteria of (I) a 2-fold up or down-regulation compared to Sham, (II) a mean of reads per million (rpm) higher than 100 and (III) an adjusted p-value < 0.05 (Supplementary Table S2). As shown in Fig. 3A, with respect to the IR group, T3 treatment significantly reverted the expression of 3 up-regulated miRNAs (miR-222, miR-31a and miR-155) and 8 down-regulated miRNAs (miR-1/133a cluster, miR-133b, myomiRs-208a/208b/499, miR-29c and miR-338) toward values observed in the Sham group. Notably, T3 administration increased the level of miR-144/451 cluster, the former was already up-regulated in the IR group (Full length images are reported in Supplementary Fig. S1 and S2). (E) Superoxide clearance measured in isolated mitochondria 3 d post-IR in the presence or absence of T3 replacement. n = 5 animals per group.   with respect to the Sham while the latter showed similar level in IR and Sham groups (Fig. 3A). The results of NGS were confirmed by qRT-PCR analysis of all the differentially expressed miRNAs (Fig. 3B).
In accordance with previous reports 15, 16 , several members of the miR-30 family were down-regulated by IR. In the presence of T3 replacement, all the miR-30 members showed a weaker down-regulation as compared to IR alone (see Supplementary Table S2) although at not significant level, which is probably due to the intrinsic difficulties in the detection by NGS of miRNAs whose sequences differ in only 1 or 2 bases. Even though miRNA-30 family was excluded from the list of IR vs IRT3 differentially expressed miRNAs (T3DE-miRNAs) shown in Fig. 3A, we decided to include it in the subsequent analysis as we had already reported a significant role of T3 in the up-regulation of miR-30 after IR in previous works 15,16 . In any case, the miR-30 up-regulation after T3 treatment was confirmed in the present study through qRT-PCR analysis (Fig. 3B).
Integrative miRNA-mRNA analysis reveals the cardioprotective role of a complex T3 dependent network in the early post ischemia reperfusion setting. Next we aimed to evaluate whether the T3 differentially expressed miRNAs (T3DE-miRNAs) might be involved in the regulation of the biological processes emerged from the T3DE-genes functional enrichment analysis. To this purpose, we used MiRWalk to search for predicted target pathways of the T3DE-miRNAs restricting the analysis to the processes highlighted by the gene array profiling and that are critical for the evolution of the post ischemic cardiac disease evolution. As shown in Fig. 3C, and Supplementary Table S3, all the T3DE-miRNAs were predicted to have a role in the selected processes leading to adverse remodeling ("apoptosis", "p53 signaling", "Tgfb signaling" and "ECM receptor pathways"), and heart disease evolution ("hypertrophic cardiomyopathy" and "dilated cardiomyopathy").
Given that the T3DE-miRNAs and the T3DE-genes share common pathways, we used MiRWalk to investigate whether the T3DE-genes were predicted targets of the T3DE-miRNAs restricting the analysis to the anti-modulated miRNA/mRNA couples. The list of the T3DE-miRNAs and their target mRNAs that display inverse modulation (T3DE-targets) is reported in Tables 4 and 5 according to the gene functional grouping. It is of note that the prediction algorithms identified miRNA potential recognition elements (MREs) within the vast majority of T3DE-genes of both MPA (31 out of 33 T3DE-genes) ( Table 4) and FPA (31 out of 39 T3DE-genes) ( Table 5). This integrative analysis of T3DE-genes and T3DE-miRNAs identified the presence of T3-modulated miRNA-gene regulatory circuitries that are predicted to counteract noxious pathways leading to adverse remodeling such as cell death, mitochondrial dysfunction, extracellular matrix disarray, inflammation and pro-fibrotic cascades (Tables 4 and 5). Interestingly, the vast majority of the T3DE-targets within the MPA and FPA share some common T3DE-miRNAs, with miR-133, miR-208 and miR-30 families showing the highest number of overlapping mRNAs (Tables 4 and 5).
Taken together the in silico findings suggest the presence of functional cross-talking among the T3DE-targets of the MPA and FPA. Therefore, we used the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) software to visualize the molecular and functional interaction among the T3DE-targets. As shown in Fig. 3D, the T3DE-targets form an integrated regulatory network where Tp53 emerges as an important interconnecting node linking the pro-apoptotic pathways of the MPA to the pro-fibrotic and adverse remodeling signaling of the FPA. This analysis also evidenced two separated minor networks involved in carnitine shuttle (Cpt1b, Cpt2 and Slc25a23) and protein targeting to mitochondrion (Timm8b, Timm17a) with nodes corresponding to the T3-up-regulated mRNAs that are predicted targets of the T3-down-regulated miRNAs.

Discussion
Post ischemic cardiovascular disease, as well as cardioprotection, are multifactorial phenomena whose understanding requires systems-biology based approaches with all of the individual factors studied in relation with each other in an effort to unravel complex regulatory networks relevant to disease etiopathology and evolution. So far the beneficial cardiovascular effects of thyroid dyshomeostasis correction have been mainly inferred through the analysis of individual players in isolation 21,22 . This approach, although useful in unraveling several aspects of the T3-dependent cardiovascular effects, disregards the dynamic nature of the biological systems with all the synergistic or antagonistic actions of the involved factors.
The systems biology perspective adopted in this paper highlights for the first time the presence of integrated, T3-dependent miRNA-gene regulatory circuitries that are predicted to suppress cross-talking pathways involved in adverse cardiac remodeling and disease progression after an acute ischemic heart injury.
LV remodelling following acute IR is an early process in which mitochondrial disorganization, cell death, inflammatory response, fibroblast activation and ECM dyshomeostasis are the main myocardial triggers of adverse changes in tissue architecture leading to heart failure. As a consequence, an efficacious therapeutic strategy should counteract the initial activation of these detrimental pathways possibly without inducing adverse effects. In the present study we adopted a strategy of T3 replacement using a dose that resulted in effective restoration of FT3 plasma levels and preservation of mitochondrial function and cardiac performance without altering heart rate. We also verified that the new dose was appropriate to modulate the expression of well known TH responsive markers such as miR-208 and ppargc1 (pgc1α): the former is involved in the regulation of myosin isoform expression, the latter is a well-known regulator of mitochondrial biogenesis 14,23 . After checking safety and effectiveness of the new T3 dose, we assessed the role of the T3 replacement strategy in targeting the initial noxious processes induced by IR at the molecular level. To this purpose we analyzed the expression profile of the T3DE-targets obtained using STRING and setting the minimum required score at 0.7. The transcripts of the MPA and FPA are displayed in blue and in red respectively. Line thickness indicates the strength of data to support the interactions. Disconnected nodes were omitted from the network display.
gene arrays dedicated to mitochondrial quality control (MPA), wound healing and ECM homeostasis (FPA). As expected, IR procedure led to the up-regulation of processes involved in cardiomyocytes death and fibroblast activation including p53-dependent cascade, Mapk and Tgfb/Smad signaling. T3 supplementation was associated to the reversal of this expression signature. These findings extend and integrate previous data showing a protective role of TH against cardiomyocyte apoptosis and against the activation of p38/Mapk/Jnk signaling and pro-fibrotic cascade following heart ischemia 15,16,24,25 . In addition, we found that several transcripts of non-cardiomyocyte origin, implicated in inflammation and cell adhesion processes, were also modulated by T3 indicating that several different myocardial cell types may be involved in the protective effects of TH (see Fig. 2B). In support of this notion, previous independent works indicate beneficial actions of TH on fibroblasts, endothelial cells and vascular smooth muscle cells (reviewed in 26 ).
Further, our gene functional enrichment analysis showed that many terms referring to heart disease condition, organ morphogenesis, regulation of cell size and growth were affected by T3 (Fig. 2B). Taken together our mRNA profiling data suggest that alterations in survival, proliferation and differentiation pathways, occurring in cardiomyocytes and non-cardiomyocyte cells during the first 3d post-IR and possibly involved in disease evolution, might be reverted by an early T3 replacement.
Such a degree of complexity suggests more sophisticated regulatory levels for TH signaling in addition to the classical control of gene expression through promoter binding. Recent evidence indicates an interplay between cardiac action of TH and the modulation of myocardial miRNAs suggesting that miRNA-dependent post  Table 4. Anti-modulated T3DE-miRNAs and T3DE-targets from the mitochondria array (MPA) listed according to gene functional grouping.
translational modifications may have a critical role in TH signaling implicated in physiological cardiac hypertrophy, control of cardiomyocyte fate and regulation of tissue repair 15,16,19,20 . The present study reports for the first time a comprehensive miRNA profiling aimed to deepen our understanding of the cardiac T3 action following an acute IR event. Our results provide novel evidence that early post-IR T3 administration is paralleled by the modulation of a signature of cardiac miRNAs critically involved in the regulation of cardiac development, function and disease. Most of the miRNAs up-regulated by T3 have been previously shown to protect the infarcted myocardium by targeting cell death (miR-144/451 cluster, ref. 27 ; miR-30a, ref. 15 ; miR-133a, ref. 28 ), mitochondrial dynamic (miR-499, ref. 29 ) and profibrotic processes (miR-29, ref. 18 miR-29, 133 and 30c ref. 16 ). On the other side, the T3-down-regulated miR-31, -155 and -222, induced by IR, have been implicated in adverse cardiac remodeling and dysfunction during ischemic heart disease and heart failure [30][31][32][33] .
In addition, in the presence of T3 there was a tendency to a reduction of other stress responsive miRNAs such as miR-27 and miR-214 (Supplementary Table S2). Both in experimental models and in the clinical arena these miRNAs have been shown to evoke cardiac hypertrophy and failure by targeting the TH receptor beta1 and the TH inactivating enzyme deiodinase type III respectively 19,34,35 . These findings are in line with previous reports and confirm the critical role of miRNA modulation in TH signaling during stress conditions 19,34 .
In turn, alterations in the TH level have been shown to affect cardiac miRNA expression 36,37 . This reciprocal regulation between miRNAs and TH system strengthen the importance of maintaining the myocardial TH homeostasis to ensure a proper miRNA-dependent cardiac physiology. Our data indicate that a low dose T3  replacement strategy, sufficient to correct the post-IR lowT3S without altering plasma TH level, might be implemented in order to rescue both miRNA levels and cardiac function. In accordance with this interpretation, several lines of evidence suggest higher cardioprotective effects of physiological versus pharmacological T3 administration in the setting of cardiac disease 14,38,39 .
Our miRNA enrichment analysis on pathways emerged from the mRNA array experiments strongly suggests the causal involvement of miRNA modulation in the beneficial effect of T3 replacement. To further strengthen this association, the bioinformatic integration of the T3DE-miRNAs and anti-modulated T3DE-genes predicts a substantial contribution of the T3DE-miRNAs as direct regulators of the T3DE-gene expression profile observed in the post-IR setting. The number of overlapping T3DE-targets shared by the T3DE-miRNAs and the number of mRNAs bearing complementary sequences for more than one T3DE-miRNA suggest that an interplay between cooperativity and multiplicity of the T3DE-miRNAs may fine-tune gene expression to form complex regulatory networks within the heart tissue in the early phase of wound healing repair.
Accordingly, as a novel result, our in silico analysis evidenced that the T3DE-targets, form a network of cross talking nodes where the interactions of p53, c-JUN and Sp1 represent the connecting bridges between the mitochondria dependent cell death, adverse remodeling and pro-fibrotic cascade. These findings are consistent with a recent trascriptomic report where the gene expression signature associated to conditional heart-specific p53 knockout confers resistance to acute biomechanical stress and highlights the importance of p53 networks in the regulation of cardiac architecture, excitation-contraction coupling, mitochondrial biogenesis, and oxidative phosphorylation capacity 40 . Accordingly, we found that the miRNA-mRNA circuits associated to T3 replacement are predicted not only to inhibit the harmful pathways linked to p53 activation and adverse remodeling, but also to enhance protective processes that preserve mitochondrial function and integrity including carnitine shuttle, mitochondrial fusion, mitochondrial protein import and antioxidant activity. Studies like this might provide the basis to understand the molecular regulatory mechanisms modulated by miRNAs and thyroid hormone in cardiac IR and encourage to consider T3 replacement as a therapeutic choice in patients with acute ischemic cardiac disease and low T3S.
In conclusion, our data validate the effectiveness of a strategy of T3 replacement and support its clinical relevance. Also this work provides the first evidence that the beneficial cardiac effects of an early T3 replacement in the initial phase of the wound healing process is paralleled by the modulation of interacting molecules from different cell types putatively under the synergistic control of T3-responsive miRNAs. Although further studies are necessary to validate the predicted miRNA-gene circuitries, based on the present and previously published data, we envision a model in which stress stimuli affect the cardiac TH signaling leading to a reduction of T3 level; in this scenario the dysregulation of T3-dependent miRNAs and their target mRNAs favors cell death and adverse remodeling, which might be blunted by the maintenance of T3 cardiac homeostasis through T3 replacement.

Materials and Methods
Animal procedure. Animals used in this investigation conformed to the recommendations in the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996) and the protocol was approved by the Animal Care Committee of the Italian Ministry of Health (Endorsement n.552/20156-PR). All surgery was performed under anesthesia, and all efforts were made to minimize suffering. Myocardial infarction was produced by ligation of the left descending coronary artery of 12-15 weeks old adult male Wistar rats and weighing 310 ± 3 g. Rats were anesthetized using Zoletil ® + xylazine (50 and 3 mg/kg respectively), connected to a respirator through an oropharyngeal cannula, and ventilated (Rodent Respirator 7025, Ugo Basile, Varese, Italy) with room air. A standard limb D1-D3 electrocardiogram (ECG) was continuously monitored and recorded at 2 kHz sampling rate (ML135 PowerLab/8SP equipped with ML135 Dual Bio Amp, ADI Instruments Ltd., Oxford, UK) using subcutaneous stainless steel electrodes (MLA0112 ECG Lead Switch Box, ADI Instruments Ltd., Oxford, UK). The heart was rapidly exposed through a thoracotomy at the left fourth intercostal space and pericardial incision. The left descendent coronary artery (LAD) was surrounded at about 2-3 mm from its origin by a 6-0 silk suture and its extremities were threaded through polyethylene tubing (PE-50) to form a snare for reversible artery occlusion. Ischemia was confirmed by ST segment elevation at ECG and visually assessed regional cardiac cyanosis. After 30 min of ischemia, the knot around the vessel was loosened and unrestrained reperfusion allowed. The silk suture was left in situ, the heart was returned to its normal position and the chest closed. A control group of rats underwent all surgical procedures except for the occlusion of the LAD (sham-operated group). Post-operatively, all rats were hydrated with physiological saline and given the analgesic buprenorphin 0.05 mg/kg s.c. (Temgesic ® , Schering-Plough, Brussels, Belgium) before they gained consciousness.
Experimental protocol. To mimic the low T3 syndrome observed in the clinical setting, only the IR rats which exhibited a >50% reduction of their basal serum FT3 level 24 h after surgery were randomly treated for 48 h with a constant subcutaneous infusion of 3 μg/kg/day T3 (IRT3, n = 5) or saline (IR, n = 5) via a miniosmotic pump (Alzet, model 2ML4, Palo Alto, CA, USA). Rats that at 24 h post IR exhibited serum FT3 reduction <50% of their basal value were excluded from the present study. To fix the T3 replacement dosage we used the plasma levels of free T3 (FT3) as a surrogate for the myocardial concentration value, due to the good correlation between the two indices previously demonstrated 15,41 . Hence, we set the T3 concentration in the osmotic pump to deliver subcutaneously 3 μg/kg/day, in order to maintain the plasma level of FT3 at about 3 pg/ml.
A group of sham-operated rats was treated with constant infusion of saline and used as control (Sham group, n = 5). Three days after surgery hearts were arrested in diastole by a lethal KCl injection. Cardiac tissue samples were obtained from the ischemic reperfused region and either stored at −80 °C until use or immediately processed for mitochondria isolation.
The samples used for miRNA profiling (one Sham, two IR, and two IRT3) were collected during a previous study in which the same model and T3 treatment protocol was used but a higher T3 dose (6 μg/kg/day vs 3 μg/ kg/day of the present study) 15 . The new-dose-study samples were used for all the other experiments including the validation of miRNA profiling by qRT-PCR.
Echocardiography study. All the animals underwent ultrasound (US) examination with a high-resolution imaging system (Vevo 2100, FUJIFILM VisualSonics Inc, Toronto, Canada) 3 days after infarction. A nose cone was used for maintaining animals under gaseous anaesthesia during the examinations (1.5% isoflurane in 1.2 L/min of pure oxygen) and heart rate (HR), respiration rate (RR) and body temperature (T) were monitored and acquired using the Advancing Physiological Monitoring Unit provided with the imaging station (Vevo Imaging Station, FUJIFILM VisualSonics Inc, Toronto, Canada). The thorax, previously shaved using depilatory cream (Nair, Church & Dwight Canada Corp., Mississauga, ON, Canada), was coated with acoustic coupling gel (SonoSite Cogel, Comedical Sas, Trento, Italy). For image acquisition a 20 MHz US probe (MS250, FUJIFILM VisualSonics Inc, Toronto, Canada), held in position by a mechanical arm, was employed. Images of the heart were acquired using B-mode modality in parasternal long axis (PLAX) and short axis (SAX) views, and then analyzed offline to assess cardiac structure and function. For each animal, transmitral inflow Pulsed-wave Doppler (PW-Doppler) images were obtained in apical 4-chamber view: the ratio of the early to late ventricular filling velocities (E/A) was calculated elaborating mitral inflow data and used as diastolic function parameter.
Mitochondria isolation and functional analysis. Mitochondria were purified according to the manufacturer's protocol (MITO-ISO1; Sigma) and as previously described (Forini). Cardiac tissue was homogenized in buffer containing 10 mM Hepes, 200 mM mannitol, 70 mM sucrose and 1 mM EGTA (PH 7.5) and centrifuged at 2000 g at 4 °C for 5 min. The supernatant was centrifuged at 11000 g at 4 °C for 20 min to collect the pellet. The mitochondria were suspended in storage buffer at pH 7.5 containing 10 mM HEPES, 250 mM sucrose, 1mMATP, 0.08 mM ADP, 5 mM sodium succinate, 2 mM K2HPO4 and 1 mM DTT and stored at −80 °C until use. An aliquot of the suspended pellet was assayed for protein content with the BioRad protein assay kit.
The enzymatic activities of RC complexes were assayed spectrophotometrically on intact mitochondria and expressed as a ratio to the activity of citrate synthase (CS), a mitochondrial matrix enzyme used to account for mitochondrial mass. Kinetic assays were performed at 30 °C using a Beckman Coulter DU760 spectrophotometer following standard methods already reported in Nesti C et al. 42 . Complex I (NADH:ubiquinone reductase) activity was measured by the rotenone sensitive oxidation of NADH at 340 nm. Complex II (succinate dehydrogenase) activity was measured by the malonate sensitive reduction of succinate at 600 nm. Complex I/III (NADH: cytochrome c reductase) activity was measured by NADH dependent reduction of cytochrome c at 550 nm. Complex II/III (succinate:cytochrome c reductase) activity was measured by succinate dependent reduction of cytochrome c at 550 nm. Complex IV (cytochrome c oxidase) activity was measured by the oxidation of reduced cytochrome c at 550 nm. CS activity was determined by the formation of 5-thio-2-nitrobenzoic acid at 412 nm.
For the assessment of mitochondrial superoxide clearance, mitochondria were incubated at 37 °C in medium containing 125 mM KCl, 2 mM K 2 HPO 4 , 20 mM HEPES, 5 mM pyruvate, 4 mM MgCl 2 , 3 mM ATP and 50 μM EGTA, pH 7.0. To increase mitochondrial superoxide production, 2 μM of the complex III inhibitor Antimycin-a was added to the reaction buffer. Five μM of the mitochondrial superoxide specific dye MitoSOX Red (Invitrogen) was added to a final volume of 200 μl before starting measurement. The rate of the fluorescence decay, indicating superoxide detoxification, was recorded at 510 nm excitation wavelength and 570 nm emission wavelength. The slope of the linear tract of the curve obtained in the first 18 min of the reaction was used as an index of the superoxide clearance rate. To account for aspecific signal, such as mitoSOX-reduction or fluorescence bleaching, a blank experiment was performed in mitochondria free sample containing only the mitochondria suspension buffer, no appreciable changes over time were recorded. All mitochondrial assays were performed in triplicate using the microplate reader Infinite M200 PRO (TECAN).
SCIentIFIC REPORTS | (2018) 8:13870 | DOI:10.1038/s41598-018-32237-0 RNA extraction and cDNA synthesis. Total RNA was extracted from homogenized heart tissue using the RNeasy mini kit (QIAGEN, Italy) following the manufacturers' instructions. RNA quality and amount was determined using the Agilent Bioanalyzer 2100 and the RNA 6000 Nano Kit (Agilent Technologies) after a DNA digestion step with DNAse (invitrogen), to remove genomic DNA contamination. cDNA for quantification of gene and miRNA expression was generated in a single-step reaction from 1 μg of RNAs using the Miscript Reverse Transcription Kit (Biorad) or the miScript Reverse Transcription Kit (Qiagen) respectively, following the manufacturer's instructions. Resulting cDNAs were stored frozen at −20 °C until assayed.
PCR array profiling. The PCR-array gene expression profiling was performed using the Biorad PrimePCR plates dedicated to mitochondria gene panel (Mitochondria SAB Target List R96) and to fibrosis gene panel (Fibrosis SAB Target List R96) according to the manufacturer's instruction. Expression levels were quantified by SYBR Green chemistry in 96 well plates with a Biorad CFX96 Real-Time System (Biorad) Five biological replicates were performed for each group of rats. Each PCR-array plate contained quality control wells for assaying RNA quality, reverse transcription quality, PCR quality and DNA contamination. Cq analysis was performed by the Bio-Rad CFX Manager software with a constant baseline adjustment of the relative fluorescent units for all array runs to provide for accurate comparisons across samples. Data were normalized against the geometric mean of Tbp, Gapdh and Hprt housekeeping genes using the ΔΔCq method.
cDNA library preparation and next generation sequencing. Library preparation and sequencing was done using Illumina's NGS platform as previously described 42,43 . The purified libraries were quantified on the Agilent DNA 7500 chip and subjected to sequencing on an Illumina HiSeq. 2500 in high-output, 50 bp single-read mode in pools of 9 per lane. Sequencing chemistry v3 was used. Read data were extracted in FastQ format using the Illumina supported tool bcl2fastq v1.8.3. Sequencing produced a mean of 26.2 million reads per sample (range 5.3M-37.3M) resulting in a mean of 14.7 million reads per sample (range 3.5M-18.7M) after trimming and quality filtering. MiRNA identification and differential expression analysis. The processing and annotation of small RNA-Seq raw data was performed using miRExpress (v2.1.3) 44 to identify rattus pre-miRNAs present in mir-Base repository (rel. 20) with at least 95% of sequence identity and a length tolerance range of 4 bp for mapping. Differential expression analysis was performed using Bioconductor's package DESeq 45 . The reads count, used as measure of miRNAs quantification, was first normalized by library size factors to a common scale. p-values were estimated using a negative binomial distribution model and local regression after dispersion estimation in pooling mode. Raw p-values were finally adjusted for multiple testing using the Benjamini and Hochberg 46 procedure controlling the false discovery rate (FDR). miRNAs with an adjusted p-value < 0.05 were considered to be differentially expressed.
Quantitative real-time RT PCR. Both gene array results and NGS miRNA profiling data were validated by quantitative real-time reverse transcription PCR (qRT-PCR) as previously described 15,16 using the new-dose-study samples (n = 5 animals per group). Genes and miRNA expression levels were quantified by SYBR Green chemistry with the Rotor-Gene Q 2-Plex (QIAGEN) and normalized respectively against Hprt or small nuclear RNA-U1 (snRNA-U1) and small nucleolar RNA, H/ACA Box 55 (snora55). The primer used in the qRT-PCR experiments are listed in Supplementary Table S4. Gene Ontology and pathways analysis. To gain more mechanistic insights, analyses of the overrepresented biological processes and KEGG pathways were conducted by using the functional annotation tool available within DAVID Website (https://david.ncifcrf.gov) and the lists of T3DE-genes as input. miRNA target prediction and analysis of molecular interaction network. The list of T3DE-miRNAs resulting from NGS was analyzed by the miRWalk2 tool (http://zmf.umm.uni-heidelberg.de/apps/zmf/mirwalk2/) to associate the T3DE-miRNAs to pathways of interest and to predict miRNA targets among the lists of T3DE-genes. Finally, to create an integrated network of interaction among the T3DEmRNAs, all the mRNAs potentially targeted by the differentially expressed miRNAs were ran in the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database (https://string-db.org/) setting the minimum required interaction score at 0.700. Statistical analysis. Data were normalized to Sham pool. When conditions for parametric test were verified, differences among groups were evaluated with one way ANOVA, followed by Bonferroni post hoc correction; in these case data are expressed as mean ± SEM and differences were considered statistically significant at a value of p < 0.05. Data obtained from PCR-arrays and all the qRT-PCR quantifications were compared by means of the nonparametric Kruskall-Wallis test followed by a Mann-Whitney U-test (adjusting the α-level by Bonferroni inequality), in these cases data are express as median and interquartile range (IQR) and differences were considered statistically significant at a value of p < 0.017.

Data Availability
All relevant data are within the paper and its supplementary information files, except for NGS data that have been deposited in NCBI's Gene Expression Omnibus (Edgar et al., 2002) and are accessible through GEO Series accession number GSE116046 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE116046).