Sirt1 improves heart failure through modulating the NF-κB p65/microRNA-155/BNDF signaling cascade

Heart failure (HF) affects over 26 million people worldwide, yet the pathologies of this complex syndrome have not been completely understood. Here, we investigated the involvement of deacetylase Sirtuin 1 (Sirt1) in HF and its downstream signaling pathways. A HF model was induced by the ligation of the left coronary artery in rats, where factors associated with left ventricular echocardiography, heart hemodynamics and ventricular mass indexes were recorded. Collagen volume fraction in heart tissues was determined by Masson’s trichrome staining. Cell models of HF were also established (H2O2, 30 min) in cardiomyocytes harvested from suckling rats. HF rats presented with downregulated expressions of Sirt1, brain-derived neurotrophic factor (BDNF) and exhibited upregulated expressions of NF-κB p65 and miR-155. Repressed Sirt1 expression increased acetylation of NF-κB p65, resulting in the elevation of NF-κB p65 expression. NF-κB p65 silencing improved heart functions, decreased ventricular mass and reduced apoptosis in cardiomyocytes. MiR-155 inhibition upregulated its target gene BDNF, thereby reducing cardiomyocyte apoptosis. Sirt1 overexpression upregulated BDNF, improved heart function, and reduced apoptosis in cardiomyocytes. In conclusion, Sirt1 alleviates HF in rats through the NF-κB p65/miR-155/BDNF signaling cascade.


INTRODUCTION
Heart failure (HF) is a global health burden, affecting more than 26 million people [1]. It is a complex clinical syndrome that may be caused by functional and/or structural impairment of the heart [2]. Although HF can be caused by different etiologies, there is a growing patient population that have HF, with left ventricular impairment and ejection deficiency, which currently does not have any specific treatment guidelines [3]. Half of the HF patients were hypertensive in China; however, approximately 30% were using anti-hypertensive medications [4]. The average duration of hospitalization of HF patients in China was 9-10 days, thereby exerting a huge economic burden on society [5]. Cardiomyocyte apoptosis has been reported as one of the causes for adverse remodeling, thus contributing to HF at later phase [6]. It is, therefore, of great importance to inhibit cardiomyocyte apoptosis to improve the treatment and outcomes of HF.
Ineffective treatment for HF is partially attributed to complex pathologies that have not been fully understood. Sirtuin (Sirt) is a family of highly conserved histone/protein deacetylases that may be important for the treatment of HF [7]. Among Sirt family, Sirt1 and Sirt3 have been extensively studied in the cardiovascular system. Sirt1 is particularly noteworthy because of its effect on cardiomyocyte survival and growth under stress, which is associated with ventricular hypertrophy [8]. Moreover, Sirt1 protects cardiomyocytes against oxidative stress, ischemia/reperfusion injury, and apoptosis [9]. Sirt1 has an inhibitory effect on nuclear factor kappa-light-chain-enhancer of activated B cells

Characterization of HF rat models
We initially established an animal rat model with HF after myocardial infarction according to the method of left coronary artery ligation. Ultrasonic electrocardiogram test results suggested that HF rats had higher interventricular septal dimension (IVSD), left ventricular end diastolic diameter (LVEDD), left ventricular end systolic diameter (LVESD). In contrast, HF rats had decreased left ventricular posterior wall thickness (LVPWD), left ventricular ejection fraction (LVEF), and fractional shortening (FS) compared to that of control or sham rats ( Table 1). The LVEF in HF rats were less than or equal to 45%. There were no differences in LVEF between the control and sham rats. Hemodynamic measurements revealed that left ventricular end diastolic pressure (LVEDP) was significantly elevated, while left ventricular systolic pressure (LVSP), left ventricular pressure (+dp/dt) and the rate of decrease in left ventricular pressure (-dp/dt) were significantly decreased in HF rats compared to that of control and sham rats ( Table 2). Both left ventricular mass index (LVMI) and right ventricular mass index (RVMI) were significantly higher in HF rats than those of control and sham rats ( Figure 1A).

Sirt1 is poorly expressed while NF-κB p65 is highly expressed in heart tissues rats with HF
Messenger RNA ( Figure 1B) and protein ( Figure 1C) expressions of Sirt1 were determined using reverse transcription quantitative polymerase chain reaction (RT-qPCR) and western blot analysis. The results displayed significantly reduced mRNA and protein Sirt1 expressions in heart tissues of HF rates in contrast with that of control or sham rats. On the other hand, protein expression of NF-κB p65 and NF-κB p65 Ac was significantly increased in HF rats when compared to control and sham rats ( Figure 1C). Collagen volume fraction (CVF) ( Figure 1D) and cardiomyocytes apoptosis ( Figure 1E) were increased in HF rats compared to that of control or sham rats. The above results indicated that Sirt1 expression was elevated, NF-κB p65 in heart tissues of HF rats was upregulated, while acetylation of NF-κB p65 was inhibited in HF rats.

NF-κB p65 binds to the promoter region and elevates miR-155 expression in cardiomyocytes
The expression of miR-155 was significantly higher in HF rats and was partially reduced after NF-κB p65 silencing ( Figure 3A). MiR-155 expression remained unchanged in the control and sham rats. The enrichment of NF-κB p65 in the miR-155 promoter region was significantly higher in HF rat cardiomyocytes, which was subsequently reduced by NF-κB p65 silencing ( Figure 3B). Furthermore, NF-κB p65 overexpression increased miR-155 expression, whereas NF-κB p65 silencing reduced miR-155 expression ( Figure 3C). These results showed that NF-κB p65 upregulated miR-155 by binding to the promoter region of miR-155.

MiR-155 inhibition reduces cardiomyocyte apoptosis in cell models of HF
MiR-155 expression was increased in cardiomyocytes treated with H2O2 and reduced by a miR-155 inhibitor AGING  ( Figure 4A). Cell viability in HF cells was significantly lower than that of the control cells (left panel, Figure  4B). The miR-155 inhibitor significantly increased cell viability in HF cells (right panel, Figure 4B). H2O2 treatment also increased apoptosis and was normalized by an miR-155 inhibitor ( Figure 4C). Similarly, H2O2 treatment increased apoptosis-related proteins Bax and cleaved-caspase3 prominently while at the same time decreased anti-apoptosis-related Bcl-2 protein ( Figure   4D). The miR-155 inhibitor reversed the aforementioned effects mentioned above. These results suggested that miR-155 reduced cardiomyocyte apoptosis in HF.

MiR-155 downregulates BDNF and promotes cardiomyocyte apoptosis
Online database mirDIP, RNA22, miRWalk, and starBase searches predicted that BDNF has a binding AGING site for miR-155 ( Figure 5A). We further studied this binding relationship by using a dual luciferase reporter gene assay. Luciferase activity in the BDNF-wild-type (WT)/miR-155 mimic group was significantly lower than that of mimic NC group ( Figure 5B). However, there were no differences in luciferase activity between the BDNF-mutated (MUT)/miR-155 mimic group and mimic NC group. This indicates that miR-155 specifically binds to the BDNF ( Figure 5B). BDNF mRNA ( Figure 5C) and protein ( Figure 5D) expressions in the heart of HF rats were significantly lower than that in control and sham rats. These results suggested that BDNF may be related to HF involving miR-155.
In cardiomyocytes treated with H2O2, the miR-155 inhibitor significantly reduced the mRNA ( Figure 5E) and protein ( Figure 5F) expressions of miR-155 but on the other hand, increased BDNF levels. Conversely, miR-155 mimic increased miR-155 expression but reduced BDNF. MiR-155 mimic decreased cell viability ( Figure 5G) and increased apoptosis ( Figure 5H) in HF cardiomyocytes. BDNF overexpression normalized the effect of miR-155 mimic. Similarly, miR-155 mimic was found to upregulate Bax and cleaved-caspase3 while decreasing Bcl-2, but this trend was prevented by BDNF overexpression ( Figure 5I). These results indicated miR-155 overexpression inhibited BDNF expression and promoted cardiomyocyte apoptosis. Therefore, miR-155 inhibition up-regulated BDNF and reduced cardiomyocytes apoptosis and thus may be beneficial to HF.

Sirt1 overexpression alleviates HF in vivo
Sirt1 overexpression reduced NF-κB p65 and NF-κB p65 Ac protein expressions in heart tissues of HF rats ( Figure 7A). Sirt1 overexpression led to increased AGING Table 3. Effects of NF-κB p65 silencing on heart functions.
Sirt1 overexpression reduced IVSD, LVEDD, and LVESD while elevated LVPWD, LVEF, and FS in HF rats (Table 7). Heart hemodynamic studies showed that LVEDP was significantly decreased, while LVSP, +dp/dt, and -dp/dt were increased by Sirt1 overexpression ( Table 8). The HR was not different among the different groups. LVMI, RVMI ( Figure 7C), CVF ( Figure 7D), and cardiomyocytes apoptosis ( Figure  7E) were reduced in HF rats by the overexpression of Sirt1. These results indicated that the overexpression of Sirt1 exhibited protective effects against HF.

DISCUSSION
HF is a complex disease that has a wide variety of etiological causes and mechanisms of action, many of which are still not properly understood. There are multiple animal and cell models to mimic certain, but not all, aspects of HF [24,25]. In this study, we used an artery ligation rat model to mimic ischemic injury and an H2O2-induced cardiomyocyte injury cell model. AGING Collectively, reduced Sirt1 expression in HF lead to increased NF-κB p65 and miR-155 expressions, reduced BDNF expression, which promoted HF.
Sirt1 expression is shown to be reduced in HF [13]. The reduced expression of Sirt1 in HF is consistent to the increased acetylation in HF [26,27]. Similarly, we found that a reduction in Sirt1 expression produced an increase in levels of acetylation and expression of NF-κB p65. Therefore, we expected that increased Sirt1 expressions would prevent injuries in cardiomyocytes and hence avoid the subsequent progression to HF. We also discovered that Sirt1 overexpression increased cell viability, reduced apoptosis, and improved heart function in HF rats. These results agree with previous studies by showing the beneficial effects of Sirt1 on HF including: improving cell survival, reducing oxidative stress, and reducing ischemia/reperfusion injury [8,9,28,29]. Therefore, Sirt1 may be a novel target for the treatment of HF [7,30].
The increased levels of protein acetylation are wellknown in HF. It is postulated that acetylation is caused by ischemic stress, leading to metabolic dysfunction especially in mitochondria [26]. The acetylation of metabolic-relevant, non-histone-related proteins such as ATP synthase, long-chain acyl-CoA dehydrogenase, creatine kinase, malate dehydrogenase, and pyruvate dehydrogenase can be visibly seen in the mitochondria of failing hearts [31]. Acetylation of histone-related   proteins, on the other hand, has been shown to be responsible for an increased heart mass, such as left ventricular hypertrophy in HF [32]. In this study, we reported to the best of our knowledge, that the acetylation of NF-κB was increased in heart tissues during HF because of reduced deacetylase Sirt1. Our results supported the idea that acetylation is a potential target for treating HF [33,34]. AGING Table 5. Effects of miR-155 and BDNF inhibition on heart functions in HF rats.    Table 7. Effects of Sirt1 overexpression on heart functions in HF rats.  In the subsequent experiments, we found that a reduced Sirt1 expression led to an increased level of NF-κB. This result was in line with previous findings that Sirt1 exerts an inhibitory effect on NF-κB [10,11]. NF-κB is a transcription factor that involves many biological and pathological functions. It is speculated that NF-κB may cause overactive immune and/or inflammatory responses that promotes injury in HF [35]. On the other hand, NF-κB may promote HF by directly initiating cell death in cardiomyocytes [36]. The effects of NF-κB on HF may involve its target gene, miR-155, which is shown in previous studies [16,17]. MiR-155 may be involved in HF based on a few observations. Our results showed that miR-155 inhibition decreased ventricular hypertrophy, cardiomyocyte apoptosis and increased cardiac functions. Moreover, a previous study demonstrated that miR-155 is highly expressed in HF and has been proposed to promote ventricular hypertrophy and dysfunction [17,19]. Our results also presented that miR-155 overexpression increased cardiomyocyte apoptosis and expression of BDNF. Consistently, the inhibition of miR-155 suppresses the cardiomyocyte apoptosis induced by myocardial ischemia/reperfusion [37] or endoplasmic reticulum stress [38].

Group LVPWD (mm) IVSD (mm) LVEDD (mm) LVESD (mm) LVEF (%) FS (%)
MiR-155 has been recently found to regulate BNDF [20,21]. Similarly in our study, we found that miR-155 mimic reduced BDNF expression. Although the majority of the research on BNDF has been conducted on the brain, one report elucidates the role of BNDF in the neuronal-cardiac link [39]. For example, BNDF may have a role in the neuronal control of contraction and relaxation in cardiomyocytes, therefore having a physiological role in regulating heartbeat [40]. BNDF may also be protective against ischemia/reperfusioninduced cell deaths in the heart [41]. Our results were similar in a way that BNDF silencing prevented many beneficial effects of miR-155 inhibition including: reduced cardiomyocytes apoptosis, ventricular hypertrophy, and improved heart function in vivo. These results collectively provide strong evidence that reduced BNDF may have a role in causing ventricular dysfunction in HF. In other words, increased BNDF expression in cardiomyocytes may be beneficial to heart function in HF.

CONCLUSIONS
In conclusion, this study suggests that increased NF-κB expression in cardiomyocytes, induced by the attenuated expression of Sirt1, upregulates miR-155. An increased expression of miR-155 in cardiomyocytes inhibits the expression of BNDF, leading to ventricular dysfunction in HF. All signaling molecules in this pathway warrant further investigation as potential pharmaceutical targets against injuries in cardiomyocytes during HF. Additionally, other models should also be explored and help confirm results from this study because these two models only mimic part of the human disease.

MATERIALS AND METHODS
All animal procedures were carried out in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health and approved by the Animal Ethics Committee of The First Affiliated Hospital of Zhengzhou University.

Establishment of a rat model of HF
Healthy male Sprague-Dawley rats (207 ± 20 g, n = 120, Animal Experimental Center, Guangzhou University of Traditional Chinese Medicine, Guangzhou, Guangdong, China) were kept in a pathogen-free animal facility at (22 ± 3)°C, with a relative humidity 40-70%, in a 12 hour/12 hour light/dark cycle with free access to food and water. Rats were randomly divided into control (n = 12), sham (n = 12), and HF groups (n = 96). HF in rats was induced by the ligation of the left coronary artery between the left atrial appendage and the pulmonary artery, as described previously [42]. Sham rats received coronary artery exposure and suture threaded without ligation. Four weeks after the surgery, echocardiographic determination (SonoAce X8, Jumu Medical Devices Co., Ltd., Shanghai, China) of LVEF ≤ 45% was used to make sure that HF was induced successfully [43,44]. Animals with unsuccessful induction of HF were replaced.

Left ventricular echocardiography in HF rats
Rats were anesthetized by pentobarbital sodium (3%, i.p., P3761, Sigma, St. Louis, MO, USA) and secured on a wooden board in the supine position. Hair near the chest area was removed for Doppler ultrasonography (SSI-5000, ShuKang HengTong Science and Trade, Shandong, China) to determine the following: LVPWD, IVSD, LVEDD, LVESD, LVEF and FS.

Heart hemodynamics in HR rats
Rats were anesthetized with pentobarbital sodium and secured on an operating table in the supine position. LVSP, LVEDP, HR, +dp/dt and -dp/dt were simultaneously recorded on a multi-channel physiological recorder (p3 plus, B&E Teksystems, Beijing, China).

Ventricular mass index in HF rats
Rats' hearts were harvested and placed in a pre-cooled hydroxyethyl piperazine ethanesulfonic acid (HEPES) solution. The right and left ventricles were separated and placed on an electronic balance for determining the right ventricular mass and left ventricular mass. LVMI and RVMI were calculated by ventricular mass/body weight. Heart tissues were then fixed in 10% formaldehyde (pH 7.0) for 24 hours, embedded in paraffin, and cut into 4 μm coronal sections. Histopathological changes were observed under a microscope.

RT-qPCR
The tissue or cell homogenate (100 μL) was completely mixed with 1 mL Trizol reagent (15596-018, SolarBio Life Sciences, Beijing, China). Chloroform (200 μL) was added, mixed, and allowed to sit at room temperature for 15 min. The mixture was then centrifuged at 12,000 rpm for 15 min at 4°C. Supernatant was obtained, mixed with 0.5 mL of isopropanol, and allowed to sit for 10-30 min at room temperature. The resulting mixture was centrifuged at 12,000 rpm for 10 min at 4°C. Supernatant was discarded and RNA was precipitated. The RNA was resuspended with 1 mL of 75% ethanol and diluted with 20 μL of diethyl pyrocarbonate (DEPC)-treated water. The mixture was centrifuged at 8,000 rpm for 5 min at 4°C. Supernatant was discarded and the pellet was allowed to dry at room temperature for 5-10 min. The pellet was resuspended in 20 μL of DEPC-treated water to have its RNA concentration determined. RNA (2 μg) was used to produce cDNA by TaqMan reverse transcription reagent (Roche, Basel, Switzerland). Target genes were amplified by PCR in a 50 μL reaction system. Primer (Sigma, St. Louis, MO, USA) sequences are presented in Table 9. All samples were tested in triplicates. Glyceraldehyde phosphate dehydrogenase (GAPDH) was used as an internal reference primer for Sirt1, NF-κB p65, and BNDF. U6 was used as an internal reference primer for miR-155. The relative expression of genes to be tested was calculated by using the 2 -ΔΔCT method [47].

Western blot analysis
Heart tissue homogenate (100 μL) or cell lysate (1 mL) was digested at 4°C for 30 min. The mixture was centrifuged at 12,000 rpm for 20 min at 4°C. The protein concentration of the supernatant was determined AGING  2000, ab32503), Bcl-2 (1: 1000, ab59348) and GAPDH (1: 2500, ab9485) antibodies overnight at 4°C on a shaker. The membranes were washed with Tris-buffered saline and Tween 20 (TBST) for 5 min each time for a total of 3 times and were incubated with horseradish peroxidase -labeled secondary goat anti-rabbit IgG (1:5000, ab6721) antibodies for 1 hour at room temperature. The samples were washed by TBST for 5 min each time for a total of 3 times. The membranes were developed and images were taken. Gray intensity in protein bands were analyzed by Quantity One software. GAPDH was used as the internal reference. All samples were tested in triplicates and all antibodies were purchased from Abcam, Cambridge, UK.

Masson's trichrome staining
Paraffin-embedded myocardial tissues were dried in an oven at 65°C for 3 hours and dewaxed.

Chromatin immunoprecipitation (ChIP) assay
The enrichment of NF-κB p65 in the promoter region of miR-155 gene was determined by a ChIP kit (Millipore, Burlington, MA, USA). Cells in the logarithmic growth phase were mixed with 1% formaldehyde for 10 min at room temperature to allow DNA-protein cross-links to form. Cross-links formed were fragmented to an appropriate size by ultrasonic device for 10 seconds for 15 times with 10 second intervals after each cycle. Parts of the cross-link fragments were centrifuged at 13,000 rpm at 4°C, while the remaining fragments were used as ChIP Input. The supernatant was collected into three tubes and incubated with primary rabbit anti-rat NF-κB p65 (ab19870, Abcam, Cambridge, UK) or NC IgG antibodies overnight at 4°C. Endogenous DNA-protein complexes were precipitated by protein agarose/ sepharose. After centrifugation, supernatant was discarded. Cross-link proteins were allowed to break down at 65°C overnight. DNA fragments were then purified by phenol/chloroform extraction. The binding of NF-κB p65 to the promoter region of miR-155 (Table 9) was determined using Input as an internal reference.

Dual luciferase reporter gene assay
WT and MUT site sequences of the 3'-untranslated region (UTR) in BDNF mRNA were digested. Target gene fragments of WT and MUT were inserted into pmiR-RB-REPORT TM vectors (Guangzhou RiboBio Biotechnology Co., Ltd., Guangzhou, China) that had been digested previously with restriction endonucleases. The empty plasmid transfection served as a control. Vectors containing MUT and WT were co-transfected into HEK293T cells with NC mimics or miR-155 mimics, respectively. Cells were collected and lysed, centrifuged for 3-5 min, and allowed the supernatant to be collected after 48 hours of transfection. Relative light units (RLUs) were determined by Renilla Luciferase Assay Kit (YDJ2714, Yuduo Biotechnology Co., Ltd., Shanghai, China) using Firefly luciferase as an internal reference. Results between Renilla and Firefly luciferase were analyzed by using a dual luciferase reporter assay system (Promega Co., Madison, WI, USA).

Flow cytometry
Cells were detached with 0.25% trypsin (without ethylene diamine tetraacetic acid, EDTA), centrifuged and the supernatants discarded after transfection for 48 hours. Cells were washed 3 times with cold PBS, and centrifuged to discard the supernatant. TUNEL staining was based on Annexin-V-FITC Apoptosis Detection Kit (556547, Surej Biotechnology Co., Ltd., Shanghai, China). Staining solution was prepared by mixing Annexin-V-FITC, PI, HEPES buffer solution and Annexin-V/PI dye solution at 1:2:50 ratio. Cells at 1 × 10 6 cells/100 μL were left to stain for 15 min at room temperature and HEPES buffer solution (1 mL) was added and mixed homogeneously. Apoptotic cells were observed by flow cytometry (Bio-Rad ZE5, Hercules, CA, USA). The absorption and excitation wavelengths of FITC were 488 nm and 525 nm, respectively; the absorption and excitation wavelengths of PI-DNA were 535 nm and 615 nm, respectively.

Statistical analysis
Data were analyzed by SPSS version 21.0 (IBM, Chicago, IL, USA) and were expressed as a mean ± standard deviation. A comparison of data from 2 groups was performed by an unpaired t test, while 3 or more groups were analyzed by one-way analysis of variance (ANOVA) and Tukey's post hoc test. Data comparison between different time points was performed by repeated measures of ANOVA and Bonferroni post hoc test. Differences were considered statistically significant when p < 0.05.

AUTHOR CONTRIBUTIONS
B L, H Z, L L, ZZ Z, N J and XW Y. designed the study. B L, T Z, BW L, YK L and C Z collated the data, carried out data analyses and produced the initial draft of the manuscript. JX W, F W, DG F and J X contributed to drafting the manuscript. All authors have read and approved the final submitted manuscript.