S913 phosphorylation of Ulk1 protects the heart from aging through inhibition of cardiac senescence

Abstract

Aims

We have previously shown that prevention of GSK-3β inactivation with GSK-3β (S9A) stimulates autophagy through phosphorylation of Ulk1 at Ser913. In the current study, we investigated whether cardiac aging is accelerated in Ulk1S913A knock-in mice and whether cardiomyocyte senescence plays an important role in the development of aging cardiomyopathy in these mice.

Methods and results

In systemic heterozygous Ulk1S913A knock-in mice (shS913A), cardiac dysfunction (evidenced by lower ejection fraction and fractional shortening, lower dP/dtmaximum, higher end-diastolic pressure, and higher dP/dtminimum) and fibrosis, along with increased cardiomyocyte senescence (increased P16, P21, IL6, γH2AX, and P53), were observed at 18 months old. RNA-seq analysis showed that numerous genes are differentially expressed in shS913A and littermate wild type (WT) hearts, including those involved in glucose metabolism, cardiac fibrosis, and cellular senescence. The glycolytic activity was higher in adult mouse cardiomyocytes isolated from shS913A than in those from littermate WT mice. In cultured neonatal rat ventricular cardiomyocytes, overexpression of Ulk1S912A induced cellular senescence. Heterozygous cardiac-specific Ulk1S913A knock-in mice (chS913A) developed cardiac dysfunction, hypertrophy, fibrosis, and senescence at 7 months old. ABT-263 treatment (senolysis) attenuated the cardiac dysfunction, hypertrophy, and fibrosis while decreasing the number of γH2AX-positive cardiomyocytes.

Conclusion

The chS913A mice exhibit premature cardiac aging, which is mediated through stimulation of cardiomyocyte senescence. The results support the role of the GSK-3β-Ulk1-autophagy pathway in the heart during aging. Thus, these mice could be useful in studying cardiac aging and senescence.

Graphical Abstract

Graphical Abstract.

Time of primary review: 111 days

See the editorial comment for this article ‘ULK1 phosphorylation and cardiomyocyte senescence: unveiling a molecular nexus in cardiac ageing’, by M. Cristina Vinci, https://doi.org/10.1093/cvr/cvaf132.

1. Introduction

Aging in the heart is characterized by structural changes, including hypertrophy and fibrosis, and a decline in contractile function.¹ The prevalence of heart failure increases in elderly populations over age 65,² where aging may act as either an independent or a predisposing mechanism to promote cardiac dysfunction. Better insights into cardiac aging are needed to identify factors involved in the progression of aging-induced cardiomyopathy and improve the prevention or treatment of heart failure. To this end, the development of appropriate animal models of aging is important.

We have reported previously that aging in the heart is accompanied by increases in phosphorylation of GSK-3β at Ser9, which inactivates GSK-3β.³ The hearts of Gsk-3βS9A constitutively active knock-in (KI) mice are protected against cardiac aging.³ Increasing lines of evidence suggest that autophagy, a lysosome-dependent degradation mechanism, declines during aging, contributing to the impairment of protein and organelle quality control mechanisms.⁴ Aging-induced inhibition of autophagy is alleviated in Gsk-3βS9A KI mice.³ GSK-3β phosphorylates Ulk1 at Ser913, which in turn stimulates autophagy. We have shown that heterozygous systemic Ulk1S913A KI (shS913A) mice exhibit decreases in autophagic flux in the heart.³ This raises the possibility that prevention of GSK-3β inactivation with GSK-3β (S9A), stimulates autophagy through phosphorylation of Ulk1 at Ser913,³ thereby inhibiting aging in the heart.

Although aging of organs can be mediated through cell-autonomous declines in cellular function, aging is also mediated through non-cell autonomous effects, including cellular senescence.⁵ Cellular senescence is mediated by telomere shortening or length-independent mechanisms, namely telomere damage, both of which induce DNA damage responses and secretion of paracrine factors that facilitate remodelling of surrounding cells.⁵ The latter mechanism is collectively called the senescence associated secretory phenotype (SASP).⁶ Although cardiomyocytes in adult hearts are nearly terminally differentiated and rarely divide or divide very inefficiently, they can develop senescence through telomere length-independent mechanisms.⁷ Persistent activation of senescence in cardiomyocytes induces organ dysfunction and remodelling in the heart.^7,8 We have shown previously that the aging-induced increase in cardiomyocyte senescence is attenuated in Gsk-3β (S9A) KI mice.³ Furthermore, we have shown recently that suppression of autophagy promotes cardiac dysfunction through induction of senescence.⁹ Thus, we speculate that shS913A mice may undergo enhanced cardiac aging through cardiomyocyte senescence.

We here evaluated whether cardiac aging and senescence are increased in shS913A KI mice. We further investigated whether heterozygous cardiomyocyte-specific UlK1(S913A) (chS913A) KI is sufficient to facilitate aging and senescence in the heart. Finally, we asked whether elimination of senescent cardiomyocytes alleviates the cardiomyopathy phenotype in chS913A mice.

2. Methods

2.1. Genetically modified mice

Generation of systemic Ulk1S913A knock-in mice was reported previously.³ Conditional Ulk1S913A knock-in mice were generated by the Genome Editing Shared Resource at Rutgers, The State University of New Jersey. Using CRISPR incision in the middle of exon 25 of the mouse Ulk1 gene and homologous recombination, wild type exon 25 was replaced with the Ulk1S913A mutant, in which loxP-wild type (WT) minigene exon 25, 26, 27 and 28-loxP was inserted in front of the Ulk1S913A mutation (Figure 1). Cardiac-specific Ulk1S913A knock-in mice (cS913A) were obtained by crossing the conditional mice with αMHC-Cre mice. Cre deletes the WT mini gene, thereby allowing the mutant exon 25 after exon 24 to be spliced in. All animal experiments were approved by the Institutional Animal Care and Use Committee of the New Jersey Medical School, Rutgers, The State University of New Jersey. The animal use in this study also conformed to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (8th edition, 2011). To euthanize the mice for tissue harvest, the animals were first anaesthetized with Avertin (300 mg/kg, ip, once) and then euthanized by cervical dislocation. The avertin was used for anaesthesia before cervical dislocation for the following reasons: (i) the anaesthetic is used just before euthanasia, the adverse effects of it may not develop; (ii) our data were collected with avertin as the anaesthetic historically, switching to another anaesthetic will be jeopardize our data base; and (iii) pharmaceutical grade anaesthetics such as ketamine/xylazine or pentobarbital suppresses cardiac function and heart rate. All animal experiments used both male and female mice. The investigators were blinded to the genotype groups during the experiments.

Figure 1.

A schematic description of the generation of cardiac-specific Ulk1S913A knock-in mice. CRISPR-Cas9 was used to insert a floxed Ulk1 minigene comprising exons 25 through 28, as well as the entire Ulk1 3′UTR, upstream of exon 25 of Ulk1. The targeting vector also introduced the S913A mutation into the coding exon 25 downstream of the floxed minigene. Wild-type Ulk1 is expressed from the minigene, but tissue-specific Cre-mediated recombination removes the wild-type minigene, allowing expression of the S913A allele minigene in a conditional manner.

2.2. ABT-263 (Navitoclax) administration to mice

ABT-263 (Navitoclax, Selleckchem), dissolved in ethanol/polyethylene glycol 400/Phosal 50 PG at a ratio of 10:30:60, was administered to mice through oral gavage feeding at 50 mg per kg body weight per day (mg/kg/d) for 5 days per cycle for two cycles with a 1-week interval between the cycles.¹⁰ Mice fed the solvent were used as controls. The animals were subjected to echocardiographic measurements and then euthanized for histological analysis 1 week after the two cycles of treatment. Both male and female cardiac-specific heterozygous Ulk1S913A knock-in mice (chUlk1S913A) and littermate control mice at 5–6 months of age were used.

2.3. Primary culture of neonatal rat ventricular cardiomyocytes

The neonatal rats were deeply anaesthetized with isoflurane (5%, once for 5–10 min) inhalation and the heart quickly removed to euthanize them. Primary cultures of ventricular cardiomyocytes were prepared from 1-day-old Charles River Laboratories (Crl)/Wistar Institute (WI) BR-Wistar rats (Harlan Laboratories, Somerville, NJ, USA) and maintained in culture as described previously.¹¹ A cardiomyocyte-rich fraction was obtained by centrifugation through a discontinuous Percoll gradient.

2.4. Senescence associated-β-galactosidase (SA-β-gal) staining of cardiomyocytes

Neonatal rat ventricular myocytes (NRVMs) that were transduced with adenovirus harbouring human influenza haemagglutinin (HA) or HA-Ulk1S912A were stained with an SA-β-gal staining kit (Cell Signaling Technologies, #9860) according to the protocol provided by the manufacturer. In a separate experiment, Ad-LacZ was transduced into NRVMs at 10 MOI (multiplicity of infection) for 24 h and then the transduction efficiency was evaluated using β-gal staining. We found that the percentage of cells expressing β-gal was 98 ± 1%.

2.5. Glycolytic function analysis using Seahorse in real time

The extracellular acidification rate (ECAR) (mpH/min) in adult mouse cardiomyocytes was determined using a Seahorse XF96 Extracellular Flux Analyzer (Agilent Technologies) according to the company’s Installation and Operation Manual and as described previously.¹¹ The adult mouse cardiomyocytes were isolated according to the reported method,¹² directly plated in 96-well XF cell culture plates in the XF assay medium, and incubated for 1 h at 37°C without CO₂. For ECAR measurements, after baseline measurements, 10 mM glucose, 1 μM oligomycin, and 50 mM 2-deoxy d-glucose (2DG) were sequentially injected into each well.

2.6. Echocardiography

Echocardiography was performed on systemic heterozygous Ulk1S913A knock-in (shS913A) and littermate WT mice at two different time points, at 6 and 18 months of age. Under continuous isoflurane inhalation general anaesthesia (0.75–1%), transthoracic echocardiography was conducted using a high-resolution Micro-Ultrasound system (Vevo 3100, FUJIFILM VisualSonics Inc., Toronto, Canada). Two-dimension (short-axis)-guided M-mode measurements (at the level of the papillary muscles) of the left ventricle (LV) internal diameter were taken from three or more beats and averaged. LV end-diastolic dimension (LVIDd) was measured at the time of the apparent maximal LV diastolic dimension, whereas LV end-systolic dimension (LVIDs) was measured at the time of the most anterior systolic excursion of the posterior wall. LV ejection fraction (LVEF) and LV fractional shortening (LVFS) were analysed using Vevo LAB software (FUJIFILM VisualSonics, Inc.) to evaluate systolic function.

2.7. Immunoblotting analysis

Cardiac tissue homogenates were prepared from the LV apex. Cell lysates were prepared from primary cultures of rat ventricular myocytes 24–72 h after adenovirus infection. Both the homogenates and the cell lysates were prepared in a radioimmunoprecipitation assay buffer containing 150 mM NaCl, 1% Triton-X 100, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, and 50 mM Tris, pH 8.0, and supplemented with a protease inhibitor cocktail (Sigma), 5 mM NaF, and 1 mM sodium orthovanadate. We used antibodies against phospho-H2AX (S139) (CST, #9718), IL-6 (Thermo Fisher Scientific, #P620), P16 (SCBT, #SC-1661), Ulk1 (CST, #8054), P21 (SCBT, #SC-6246), P53 (Bioss, #8687R), GAPDH (CST, #2118), and α-tubulin (Sigma, #T6199).

2.8. RNA sequencing (RNA-seq) analysis

Total RNA was isolated from the LV apex of WT (n = 4) and shS913A knock-in (n = 3) mice and sent to LC Sciences (Houston, Texas, USA) for RNA sequencing analysis. A poly(A) RNA sequencing library was prepared following Illumina’s TruSeq-stranded-mRNA sample preparation protocol. RNA integrity was checked with an Agilent Technologies 2100 Bioanalyzer. Poly(A) tail-containing mRNAs were purified using oligo-(dT) magnetic beads with two rounds of purification. After purification, poly(A) RNA was fragmented using divalent cation buffer at an elevated temperature followed by DNA library construction. Quality control analysis and quantification of the sequencing library were performed using Agilent Technologies 2100 Bioanalyzer High Sensitivity DNA Chip. Paired-ended sequencing was performed on Illumina’s NovaSeq 6000 sequencing system. For the bioinformatics analysis, low-quality reads and adapter sequences were removed using Cutadapt¹³ and the quality of RNA-seq reads was assessed with FastQC. HISAT2 was used for genome mapping,¹⁴ followed by transcript assembly and comprehensive transcriptome reconstruction with StringTie.¹⁵ The differentially expressed mRNAs were selected with log2 (fold change) > 1 or log2 (fold change) < −1 and with statistical significance (P < 0.05) by R package edgeR.¹⁶

2.9. RT–qPCR

Total RNA was prepared from mouse LVs using TRIzol (Thermo Fisher Scientific). cDNA was generated using 1000 ng total RNA and PrimeScript RT Master Mix (Takara). qPCR was performed using TB Green Premix Ex Taq (Takara). Actin served as an internal control. The oligonucleotide primers used to carry out the qPCRs are listed in Supplementary material online, Table S1.

2.10. Immunofluorescence analysis

Mouse hearts were harvested, fixed with 10% formalin and embedded in wax and cross-sections (5 μm thick) were prepared. Neonatal cardiomyocytes were cultured on coverslips and fixed with 4% PFA. An overnight incubation with specific antibodies (IL-6 or phospho-H2AX (S139), P16 (Invitrogen, #MA5-17142), P21 (Abcam, #Ab7960) and cTNT (Invitrogen, #MA-5-12960)) was followed by 2 h of incubation with secondary antibody conjugated with Alexa Fluor 488, 555, 594 or 647 dye (Life Technologies). Samples were washed and mounted with a reagent containing DAPI (VECTASHIELD; Vector Laboratories). The analyses were performed in a blinded manner by fluorescence microscopy. For analysis of γH2AX-positive cardiomyocytes, the total number of γH2AX-positive nuclei in cardiomyocytes in the whole section was counted. The total number of cardiomyocytes in the whole section was estimated as the product of the average number of cardiomyocytes per image and the total number of images per section. The percentage of γH2AX-positive cardiomyocytes was calculated by the equation: the number of γH2AX-positive cardiomyocytes/total number of cardiomyocytes in the section.

2.11. Immuno-FISH (fluorescence in situ hybridization)

Immuno-FISH in formalin-fixed paraffin-embedded cardiac tissue sections was carried out according to the protocol provided by the manufacture of the Cy-3-labelled telomere specific (CCCTAA) peptide nucleic acid probe (HLB Panagene Inc., Republic of Korea) and a reported method.¹⁷ Briefly, the cardiac tissue sections were incubated with the anti-γH2AX antibody at 4°C overnight. Then the slides were incubated with secondary antibody for 1 h. The sections were then denatured for 5 min at 80°C in hybridization buffer containing the Cy-3-labelled probe (2.5 μg/mL), followed by hybridization for 2 h at room temperature in the dark. In depth Z stacking was used when taking images by confocal microscopy.

2.12. Cardiac histology

The LV base was harvested, fixed with 10% formalin and embedded in wax and cross-sections (5 μm thick) were prepared. Picric acid Sirius red (PASR) staining, wheat germ agglutinin-Rhodamine (WGA) staining, and terminal deoxynucleotidyl transferase dUTP nick-end labelling (TUNEL) staining were conducted, and measurements of cardiac fibrosis, cardiomyocyte cross-sectional area, and apoptosis were performed as previously described.¹⁸ Cardiomyocyte cross-sectional area was measured using images captured from WGA-stained sections. The outline of 100–200 myocytes was traced in each section. Suitable cross-sections were defined as having nearly circular capillary profiles and circular-to-oval cardiomyocyte cross-sections. No correction for oblique sectioning was made.¹⁹

2.13. Statistical analysis

The data are presented as mean ± S.E.M. and were first analysed using analysis of variance (ANOVA). If significant differences were observed, post hoc analysis was performed using the t-test with Bonferroni correction. A P value less than 0.05 was considered significant.

3. Results

3.1. The systemic Ulk1S913A knock-in mice developed cardiac dysfunction and fibrosis at 18 months of age

Phosphorylation of Ulk1 at Ser913 is decreased by aging in the heart.³ Since autophagy in the heart is suppressed in systemic heterozygous Ulk1S913A knock-in (shS913A) mice,³ as in aging hearts, we hypothesized that cardiac aging is facilitated in shS913A mouse hearts. We have shown previously that LVEF and fractional shortening (FS) determined by echocardiography, LV weight/tibial length, and lung weight/tibial length were not significantly different between young (6-month old) shS913A mice and their littermate wild type (WT) mice.³ On the other hand, at 18 months of age, shS913A mice exhibited a significantly lower LVEF and LVFS than WT mice (Figure 2A and Table 1), although the LV, RV, and lung organ weights did not differ significantly and there were no apparent differences in gross heart images between shS913A and WT mice (Table 2 and Supplementary material online, Figure S1A). Haemodynamic measurements showed that old shS913A mice exhibited a significantly lower dP/dtmax, indicating systolic dysfunction (Figure 2B). Moreover, shS913A mice exhibited a significantly higher LVEDP and dP/dtmin (Figure 2C), suggesting that old shS913A mice also develop diastolic dysfunction. shS913A mice exhibited slightly higher levels of cardiac fibrosis than WT mice when young. When the shS913A mice became older, however, they developed greater fibrosis than either WT or young shS913A mice (Figure 2D and E). Taken together, these data suggest that shS913A mice exhibit accelerated aging in the heart compared to WT mice. The histologically evaluated cardiomyocyte cross-sectional area and the percentage of TUNEL-positive cells were not significantly different between shS913A and WT at 18 months (see Supplementary material online, Figure S2A–D).

Figure 2.

Cardiac function and histology in aging shS913A mice and their littermate WT mice. (A) Left ventricular (LV) ejection fraction (EF) and LV fractional shortening (FS) at 18 months old. *P < 0.05. (B) The maximum LV dP/dt (dP/dtmax) in young (6 months old, Y) and old (18 months old, O) mice. *P < 0.05. (C) The LV end-diastolic pressure (LVEDP) and the minimum LV dP/dt (dP/dtmin) in Y and O mice. *P < 0.05. (D) Representative images from Picric acid Sirius red-stained cardiac sections. Scale bars = 100 μm. (E) The percent of fibrosis in Y and O mice. *P < 0.05. ****P < 0.0001. In (A)–(C) and (E), dark and bright data points indicate males and females, respectively. n (the number of animals used) = 5 in each group. In (A), statistical analysis was carried out using the unpaired t-test. In (B), (C), and (E), statistical analyses were conducted using Prism two-way ANOVA followed by multiple comparison t-test and/or Bonferroni’s t-test. P < 0.05 was considered statistically significant.

Table 1.

Echocardiographic measurements of old (18 months old) shS913A mice

	WT	shS913A
n	5	5
IVSd (mm)	0.92 ± 0.02	1.02 ± 0.02
LVIDd (mm)	3.53 ± 0.18	3.88 ± 0.16
LVIDs (mm)	2.55 ± 0.14	3.32 ± 0.06
LVPWd (mm)	0.90 ± 0.08	0.99 ± 0.08
EF (%)	62 ± 2	53 ± 3a
FS (%)	28 ± 1	22 ± 2a
HR (bpm)	456 ± 6	461 ± 4

The data are presented as mean ± S.E.M. Statistical analysis was carried out using the unpaired t-test.

IVSd, diastolic interventricular septum thickness; LVIDd, diastolic LV internal dimension; LVIDs, systolic LV internal dimension; LVPWd, diastolic LV posterior wall thickness; EF, ejection fraction; FS, fractional shortening; HR, heart rate; bpm, beats per minute.

^a P < 0.05 vs. WT.

Table 2.

Organ weight analysis of old (18 months old) wild type and shS913A knock-in mice

	WT	shS913A
n	8	14
Body weight (BW, g)	36.5 ± 2.7	34.5 ± 2.4
Tibia length (TL, mm)	18.8 ± 0.2	19.0 ± 0.2
Left ventricular weight (LVW, mg)	102.0 ± 3.5	107.5 ± 3.9
Lung weight (Lung, mg)	171.4 ± 2.9	182.8 ± 4.6
Right ventricular weight (RVW, mg)	21.6 ± 2.5	22.4 ± 2.2
LVW/BW (mg/g)	2.9 ± 0.1	3.2 ± 0.2
LVW/TL (mg/mm)	5.4 ± 0.2	5.6 ± 0.2
Lung/BW (mg/g)	4.9 ± 0.4	5.6 ± 0.3
Lung/TL (mg/mm)	9.1 ± 0.2	9.6 ± 0.2
RVW/BW (mg/g)	0.6 ± 0.0	0.7 ± 0.1
RVW/TL (mg/mm)	1.2 ± 0.1	1.2 ± 0.1

The data are presented as mean ± S.E.M.

3.2. shUlk1S913A mice exhibited more cardiac senescence

We investigated whether exacerbated cardiac aging is accompanied by increased cellular senescence in shS913A mice. The hearts of 18-month-old shS913A mice exhibited significantly higher levels of senescence markers, including P16, IL-6, P53, and γH2AX, than those from their littermate WT mice (Figure 3A and B). Immunofluorescent staining of cardiac sections indicated that the γH2AX positivity, a marker of DNA damage, is primarily observed in cardiomyocytes, as determined by cTNT positivity. Although the percentage of γH2AX-positive cardiomyocytes was greater in old than in young mice in both WT and shS913A mouse groups, the extent of γH2AX positivity was greater in old shS913A mice than in old WT mice (Figure 3C and D). To determine whether γH2AX colocalizes with telomeres, indicating the presence of DNA damage at the telomere, immuno-FISH analyses were carried out on cardiac tissue sections obtained from old mice. γH2AX colocalized with telomeres, and the number of telomere-associated foci in old shS913A mice was significantly greater than in old WT mice (Figure 3E and F). These results suggest that DNA damage occurred at the telomeres in shS913A mice. In order to further confirm that there were more senescent cardiomyocytes in the shS913A mice, we conducted immunofluorescence staining using antibodies against two additional senescence markers, P16 and P21. Our results showed that the percentage of P16-positive cardiomyocytes was significantly increased with age in both WT and shS913A mice, but the percentage in shS913A was significantly higher than in WT mice at both time points (see Supplementary material online, Figure S3A and B). Similarly, more P21-positive cardiomyocytes were observed in old WT and shS913A mice than in young mice. Moreover, the percentage of P21-positive cardiomyocytes was significantly higher in shS913A than in WT at both ages (see Supplementary material online, Figure S3C and D). Collectively, these data suggest that shS913A mice developed more cardiomyocyte senescence than WT mice.

Figure 3.

Cardiac senescence in shS913A and WT mice. (A) Immunoblotting of P16, IL6, P53, γH2AX, and Tubulin in the 18-month-old hearts. (B) Relative band density of P16/Tubulin, IL6/Tubulin, P53/Tubulin, and γH2AX/Tubulin. *P < 0.05. **P < 0.01. (C) Representative images of γH2AX-stained cardiac sections from young (Y, 6 months old) and old (O, 18 months old) mice. Arrow heads point to γH2AX-positive cardiomyocyte nuclei. Scale bars in images = 20 μm. Scale bars in insets = 10 μm. (D) The percentage of γH2AX-positive cardiomyocytes (CM). *P < 0.05. (E) Representative immuno-FISH images. Green, γH2AX. Red, Cy-3. Yellow, colocalization of Cy-3 and γH2AX (Seen more clearly in the insets). Arrows point to colocalization. Scale bars in images = 10 μm. Scale bars in insets = 5 μm. (F) The number of telomere-associated foci (TAF). ***P < 0.001. In (A)–(D), n (the number of animals used) = 5 in each group. In (F), n = 4 in each group. In (B) and (D), dark and bright data points indicate males and females, respectively. In (B), statistical analysis was carried out using the unpaired t-test. In (D), statistical analyses were conducted using Prism two-way ANOVA followed by multiple comparison t-test and/or Bonferroni’s t-test. P < 0.05 was considered statistically significant.

3.3. Overexpression of Ulk1S912A promotes senescence in cultured cardiomyocytes

To examine whether induction of senescence by Ulk1S912A, a human Ulk1 mutant equivalent to mouse Ulk1S913A, in cardiomyocytes is cell-autonomous, we overexpressed Ulk1S912A in cultured neonatal rat ventricular cardiomyocytes (NRVMs) and evaluated the induction of senescence markers. Overexpression of Ulk1S912A was sufficient to up-regulate senescence markers, including P16, P21, γH2AX, and SA-β-gal staining (Figure 4), suggesting that induction of senescence in cardiomyocytes by Ulk1S912A is cell-autonomous.

Figure 4.

Cellular senescence in cultured neonatal rat ventricular myocytes (NRVMs). (A) Immunoblotting of P16, P21, Ulk1, and GAPDH in NRVMs transduced with LacZ or Ulk1S912A. (B) The relative band density of P16, P21, and Ulk1 normalized by GAPDH. *P < 0.05. n (the number of experiments) = 3 in each group. (C) Senescence associated (SA) β-galactosidase (β-gal) staining of NRVMs transduced with haemagglutinin (HA) or HA-Ulk1S912A. Scale bars = 50 μm. (D) The percentage of the β-gal-positive area. ****P < 0.0001. n (the number of experiments) = 9 in each group. (E) γH2AX staining of NRVMs transduced with LacZ or Ulk1S912A. Arrows point to γH2AX-positive nuclei. Scale bars in images = 20 μm. Scale bars in insets = 10 μm. (F) The percentage of γH2AX-positive nuclei. **P < 0.01. n (the number of experiments) = 6. In this figure, statistical analysis was carried out using the unpaired t-test. P < 0.05 was considered significantly different.

3.4. RNA sequencing analyses in shS913A mouse hearts

To gain insight into the molecular mechanisms through which aging and the senescent phenotype are promoted in shS913A mouse hearts, RNA sequencing analyses were conducted using hearts obtained from WT and shS913A mice at 15 months of age. There were 638 differentially expressed genes with significant statistical differences between WT and shS913A mice. Among them, 180 genes were up-regulated in the shS913A mice, while 458 genes were down-regulated (see Supplementary material online, Table S2 and Supplementary material online, Figure S4A and B). Gene Ontology analysis revealed that shS913A mouse hearts exhibited enhanced processes related to cytokine activity, regulation of signalling receptor activity, and signal transduction compared to WT mouse hearts (see Supplementary material online, Figure S4C and D). Additionally, Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis indicated enrichment in cytokine–cytokine receptor interaction, the PI3K-Akt signalling pathway and the Jak-STAT signalling pathway in shS913A mouse hearts (see Supplementary material online, Figure S4E). The genes exhibiting higher levels in shS913A mouse hearts included genes involved in glucose metabolism (Slc2a4 (Glut4), Gck), genes involved in fibrosis (Nrep, Col4a6, Col14a1, and Col15a1), a gene involved in macrophage migration, efferocytosis, and M2 polarization (Egln3), and creatine kinase B (Ckb), a cytoplasmic enzyme involved in energy homeostasis and identified as a pro-senescence kinase²⁰ (see Supplementary material online, Figure S4F). On the other hand, the level of Pdk4²¹ was decreased in shS913A mouse hearts, which would also increase glucose metabolism. Among the atypical senescence associated secretory phenotype factors reported to be increased in aging cardiomyocytes,⁷ the expression of Edn3 was increased in shS913A mouse hearts (see Supplementary material online, Figure S4F).

3.5. Ulk1S913A increased glycolysis in cardiomyocytes

The results of the RNA-seq analysis showed that genes involved in the glycolytic pathway were affected in shS913A mice. To validate the results, RT–qPCR analyses were conducted with RNA isolated from LVs of WT and shS913A mouse hearts. mRNA expression of Glut4, Hk1, and Hk2 was up-regulated, whereas that of Pdk4 was down-regulated in shS913A mouse hearts compared to in WT mouse hearts (Figure 5A), confirming the results of the RNA-seq analysis. To examine whether glycolysis was enhanced in the cardiomyocytes of shS913A, we isolated adult mouse cardiomyocytes from these mice and littermate WT mice and evaluated the ECAR using a Seahorse analyser. Cardiomyocytes isolated from shS913A mice had increased relative basal glycolysis and glycolytic capacity, but not glycolytic reserve capacity, compared to cardiomyocytes isolated from WT mice (Figure 5B–E). These results suggest that Ulk1S913A promotes a metabolic shift in cardiomyocytes towards glycolysis.

Figure 5.

Genes related to glycolysis in shS913A mice and glycolysis in adult mouse cardiomyocytes. (A) mRNA expression of pyruvate dehydrogenase kinase 4 (PDK4), glucose transporter 4 (Glut4), hexokinase (HK) 1, and HK2 in shS913A mice. *P < 0.05. n (the number of animals) = 4. (B–E) Seahorse glycolytic flux analysis in adult mouse cardiomyocytes isolated from shS913A and WT mice. n (the number of animals) = 4. (B) Overall extracellular acidification rate (ECAR). (C) Relative basal glycolysis. ****P < 0.001. (D) Relative glycolytic capacity. **P < 0.01. (E) Relative glycolytic reserve. In (C)–(E), dark dots or circles indicate males and bright dots or circles indicate females. Statistical analysis was carried out using the unpaired t-test. P < 0.05 was considered significantly different.

3.6. Cardiac-specific Ulk1S913A (chS913A) knock-in mice developed cardiac senescence and accelerated cardiac aging

In order to further investigate whether Ulk1(S913A) signalling in cardiomyocytes is sufficient to promote aging cardiomyopathy in mice, we generated cardiac-specific Ulk1(S913A) knock-in mice (chS913A mice). There was no obvious difference in gross whole heart images between chS913A and littermate control mice (see Supplementary material online, Figure S1B). We investigated the role of chS913A in mediating cardiac dysfunction at 6–7 months old, similar to the age of the younger shS913A mice. chS913A mice have a significantly lower LVEF and LVFS than littermate control mice (Figure 6A and Table 3). chS913A mouse hearts exhibited significantly higher levels of P16, P21, IL-6, and P53 than the hearts of their littermate control mice (Figure 6B, C, and E). In addition, chS913A mouse hearts have a significantly higher percentage of γH2AX-positive cardiomyocytes than their littermate control mouse hearts (Figure 6D). Furthermore, chS913A mouse hearts exhibited significantly higher percentages of P16-positive and P21-positive cardiomyocytes than littermate control mouse hearts (see Supplementary material online, Figure S5A–D). There were also significantly more CD68-positive macrophages or monocytes in chS913A mouse hearts (see Supplementary material online, Figure S5E and F), suggesting that more inflammatory cells were recruited to the myocardium of chS913A mice. The chS913A mouse hearts exhibited significantly more cardiac fibrosis than their littermate WT mouse hearts (see Supplementary material online, Figure S6A). Although LV weight/tibial length was not significantly different between chS913A and their littermate control mice (Table 4), chS913A mice exhibited a significantly greater cardiomyocyte size than their littermate control mice (see Supplementary material online, Figure S6B). Interestingly, in contrast to shS913A mouse hearts, chS913A mouse hearts exhibited more apoptosis than littermate control mouse hearts (see Supplementary material online, Figure S6C). Taken together, these results indicate that suppression of UlK1 phosphorylation at Ser913 in cardiomyocytes is sufficient to induce accelerated cardiac aging and senescence.

Figure 6.

Cardiac function and senescence in cardiac-specific heterozygous Ulk1S913A knock-in (chS913A) mice. (A) LVEF and LVFS in 6–7-month-old chS913A and littermate control mice (Ctrl) (Myh6-Cre or Ulk1S913A mice without Myh6-Cre). **P < 0.01. n (the number of animals) = 5 in Ctrl, n = 6 in chS913A group. (B) Immunoblotting of P16, P21, IL6, P53, and tubulin. (C) Relative band density of P16, P21, IL6, and P53 normalized by tubulin. *P < 0.05. **P < 0.01. n (the number of animals) = 3 in each group. (D) γH2AX immunostaining. Red, γH2AX. Green, cTNT. Scale bars in images = 20 μm. Scale bar in inset = 10 μm. The dot plot shows the percentage of γH2AX-positive cardiomyocytes. *P < 0.05. n (the number of animals) = 5 in Ctrl, n = 6 in chS913A group. (E) IL6 immunostaining. Red, cTNT. Green, IL6. Scale bars = 20 μm. The dot plot shows the percentage of the area positive for IL6. ***P < 0.001. n (the number of animals) = 5 in Ctrl, n = 6 in chS913A group. In (A), (C), (D), and (E), dark and bright data points indicate males and females, respectively. Statistical analysis was carried out using the unpaired t-test. P < 0.05 was considered statistically significant.

Table 3.

Echocardiographic measurements of chS913A mice

	Ctrl	chS913A
n	5	6
IVSd (mm)	0.87 ± 0.03	0.69 ± 0.07
LVIDd (mm)	3.55 ± 0.15	4.59 ± 0.27a
LVIDs (mm)	2.33 ± 0.15	3.81 ± 0.34a
LVPWd (mm)	0.89 ± 0.06	0.69 ± 0.06
EF (%)	65 ± 2	37 ± 5b
FS (%)	35 ± 2	18 ± 3b
HR (bpm)	438 ± 23	444 ± 24

The data are presented as mean ± S.E.M. Statistical analysis was carried out using the unpaired t-test.

IVSd, diastolic interventricular septum thickness; LVIDd, diastolic LV internal dimension; LVIDs, systolic LV internal dimension; LVPWd, diastolic LV posterior wall thickness; EF, ejection fraction; FS, fractional shortening; HR, heart rate; bpm, beats per minute.

^a P < 0.01 vs. Ctrl.

^b P < 0.001 vs. Ctrl.

Table 4.

Organ weight analysis of chS913A knock-in mice and their littermate control mice (Ctrl)

	Ctrl	shS913A
n	5	6
Body weight (BW, g)	28.3 ± 1.4	32.3 ± 2.6
Tibia length (TL, mm)	18.5 ± 0.2	19.0 ± 0.4
Left ventricular weight (LVW, mg)	91.2 ± 4.9	105.0 ± 8.9
Lung weight (Lung, mg)	166.2 ± 4.1	185.0 ± 8.8
LVW/BW (mg/g)	3.22 ± 0.07	3.23 ± 0.13
LVW/TL (mg/mm)	4.95 ± 0.28	5.50 ± 0.40
Lung/BW (mg/g)	5.94 ± 0.30	5.76 ± 0.24
Lung/TL (mg/mm)	9.01 ± 0.27	9.71 ± 0.37

The data are presented as mean ± S.E.M.

3.7. Cellular senescence contributes to the accelerated cardiac aging in chS913A mice

To evaluate whether cardiac senescence contributes to cardiac dysfunction, fibrosis, and cardiomyocyte hypertrophy in chS913A mice, we investigated the effect of ABT-263, a senolytic. ABT-263 or PBS was administered to chS913A or their littermate control mice at 5–6 months of age. Since ABT-263 targets anti-apoptotic proteins in senescent cells,¹⁰ to test whether ABT-263 induces apoptosis in senescent cardiomyocytes, we conducted triple staining with anti-P16 antibody, anti-cTNT antibody, and TUNEL staining in myocardial sections obtained from ABT-263 treated control and chS913A mice. Our results demonstrated that ABT-263 significantly increased TUNEL- and P16-double positive cardiomyocytes in chS913A mice compared to in control mice, suggesting that ABT-263 induces apoptosis in senescent cardiomyocytes (see Supplementary material online, Figure S7A and B). The percentage of γH2AX-positive cardiomyocytes in chS913A mice treated with PBS was significantly higher than in their littermate control mice (Figure 7A). However, the percentage of γH2AX in chS913A mice treated with ABT-263 did not differ significantly from that in their littermate control mice and was significantly lower than in chS913A mice treated with PBS (Figure 7A), indicating that senolysis with ABT-263 decreased cardiac senescence in chS913A mice. In addition, the increased IL6-positivity in chS913A mice was abolished by the treatment with ABT-263 (see Supplementary material online, Figure S7C and D), further suggesting that senescence and the SASP were attenuated by the senolytic treatment. The LVEF and LVFS in chS913A mice treated with PBS were significantly lower than in their littermate control mice; however, ABT-263 treatment significantly increased the LVEF and LVFS of chS913A mice (Figure 7B and Table 5), indicating that senolysis with ABT-263 attenuated cardiac dysfunction in chS913A mice. Cardiac fibrosis in chS913A mice treated with PBS was significantly higher than in their littermate control mice (Figure 7C). However, cardiac fibrosis in chS913A mice treated with ABT-263 was significantly lower than in chS913A mice treated with PBS (Figure 7C), demonstrating that senolysis with ABT-263 decreased cardiac fibrosis in chS913A mice. Cardiomyocytes in chS913A treated with PBS were significantly larger than in their littermate control mice (Figure 7D), but cardiomyocytes in chS913A mice treated with ABT-263 were significantly smaller than in chS913A mice treated with PBS (Figure 7D), suggesting that the cardiac hypertrophy in chS913A mice was abolished by senolysis with ABT-263.

Figure 7.

Senolysis with ABT-263 in chS913A mice and littermate control mice (Ctrl). (A) γH2AX immunostaining and quantification of γH2AX-positive cardiomyocytes (CM). Red, γH2AX. Green, cTNT. Arrow points to positive γH2AX staining. Scale bars in images = 20 μm. Scale bar in inset = 10 μm. *P < 0.05. **P < 0.01. (B) LVEF. *P < 0.05. ***P < 0.001. (C) PASR staining and quantification of fibrosis. Scale bars = 100 μm. *P < 0.05. (D) WGA staining and cross-sectional area of cardiomyocytes. Scale bars = 20 μm. **P < 0.01. Dark and bright data points indicate males and females, respectively. In this figure, n (the number of animals) = 5 in each group. Statistical analyses were conducted using Prism two-way ANOVA followed by multiple comparison t-test and/or Bonferroni’s t-test. P < 0.05 was considered statistically significant.

Table 5.

Echocardiographic measurements of chS913A mice treated with ABT-263

	Ctrl		chS913A
	Veh	ABT-263	Veh	ABT-263
n	5	5	5	5
IVSd (mm)	0.59 ± 0.03	0.45 ± 0.04a	0.61 ± 0.04	0.50 ± 0.05
LVIDd (mm)	4.38 ± 0.24	3.71 ± 0.13a	4.13 ± 0.13	3.34 ± 0.19b
LVIDs (mm)	2.96 ± 0.22	2.56 ± 0.14	3.23 ± 0.15	2.19 ± 0.15b
LVPWd (mm)	0.58 ± 0.03	0.52 ± 0.05	0.62 ± 0.02	0.54 ± 0.04
EF (%)	61 ± 2	59 ± 4	45 ± 3c	64 ± 2d
FS (%)	33 ± 2	31 ± 3	22 ± 2c	34 ± 2d
HR (bpm)	466 ± 14	504 ± 16	464 ± 17	504 ± 12

The data are presented as mean ± S.E.M. Statistical analyses were conducted using Prism two-way ANOVA followed by multiple comparison t-test and/or Bonferroni’s t-test.

IVSd, diastolic interventricular septum thickness; LVIDd, diastolic LV internal dimension; LVIDs, systolic LV internal dimension; LVPWd, diastolic LV posterior wall thickness; EF, ejection fraction; FS, fractional shortening; HR, heart rate; bpm, beats per minute.

^a P < 0.05 vs. Ctrl-Veh.

^b P < 0.01 vs. chS913A-Veh.

^c P < 0.01 vs. Ctrl-Veh.

^d P < 0.001 vs. chS913A-Veh.

4. Discussion

Our results show that inhibition of Ulk1 phosphorylation at Ser913 promotes aging in the heart. Since the development of cardiomyopathy is facilitated in both systemic and cardiac-specific Ulk1(S913A) knock-in mice, down-regulation of Ulk1 phosphorylation at Ser913 in cardiomyocytes appears to be sufficient to promote cardiac aging. Cardiomyocyte senescence is promoted in the presence of Ulk1(S913A) knock-in in vivo and in the presence of Ulk1(S912A) (human sequence) in neonatal rat ventricular cardiomyocytes. Thus, down-regulation of Ulk1 phosphorylation at Ser913 promotes cellular senescence in cardiomyocytes in a cell autonomous manner. Since senolysis with ABT-263 alleviates cardiac dysfunction, senescence must play an important role in mediating cardiac dysfunction in chS913A mice.

We have shown previously that GSK-3β is inactivated through S9 phosphorylation during aging, whereas GSK-3β(S9A), a Ser9 phosphorylation resistant and constitutively active GSK-3β mutant, inhibits aging cardiomyopathy in mice.³ Ser913 of Ulk1 is phosphorylated by GSK-3β and mediates stimulation of autophagy.³ We have shown previously that GSK-3β-induced stimulation of autophagy is alleviated in the presence of Ulk1(S913A),³ suggesting that phosphorylation of Ulk1 at Ser913 plays an important role in mediating the effect of GSK-3β upon autophagy. The current study extends the role of Ulk1 phosphorylation at Ser913 to cardiac aging and suggests that inhibition of the GSK-3β-Ulk1 pathway promotes aging in the heart. Since autophagy is an important negative regulator of aging, suppression of autophagy may play an important role in mediating cardiomyopathy in chS913A mice. Further investigation is needed to test this hypothesis.

Our recent work suggests that inhibition of GSK-3β during cardiac aging could facilitate aging cardiomyopathy through stimulation of senescence.³ Together with the new observation in this study that inhibition of Ulk1 phosphorylation at Ser913 promotes senescence, which also mimics the effect of GSK-3β inhibition during aging, these findings suggest that the GSK-3β-Ulk1 pathway regulates senescence in cardiomyocytes. We believe that senescence is involved in the development of cardiomyopathy in chS913A mice, since senolytic treatment both reduced the number of senescent cardiomyocytes and improved cardiomyopathy. It should be noted that ABT-263 treatment improved cardiomyopathy in chS913A mice despite inducing death in senescent cardiomyocytes. Thus, we believe that removing senescent cardiomyocytes can improve cardiac function even though it may reduce the total number of cardiomyocytes in the heart, where regeneration is very inefficient. Although we used ABT-263, a well-established senolytic, ideally, it should be confirmed that additional senolysis-inducing interventions exhibit similar effects in future studies.

RNA sequencing analyses showed that shS913A mice exhibit a unique gene expression profile characterized by up-regulation of some of the genes involved in cellular senescence and in glucose metabolism. It should be noted that up-regulation of the glycolytic mechanism has been reported as a feature of cellular senescence induced by oncogenes.²² Since senescent cells are characterized by the presence of pro-survival mechanisms and the SASP, activation of aerobic glycolysis may play an important role in maintaining the survival and function of senescent cells in the sh913A mouse heart. Further experiments are needed to test this hypothesis.

Paracrine factors secreted from senescent cells through the SASP mediate both spreading of senescence and phenotypic remodelling of surrounding cells. The factors produced by the SASP are stimulus-dependent and differ between various senescent cell types.⁵ Our RNA sequencing analyses showed that the expression of Col15a1 and Col4a6 is stimulated in shS913A mice. It has been shown that aging-induced cardiomyopathy is characterized by up-regulation of a unique secretome.⁷ Among the SASP factors reported in an aging cardiomyopathy model,⁷ up-regulation of Edn3 was also observed in the shS913A mouse heart. Further investigation is needed to elucidate the functional significance of the SASP factors mediating aging cardiomyopathy. In addition, the signalling mechanism through which the SASP factors are modulated also remains to be elucidated.

Although both shS913A and ch913A mice exhibited cardiomyopathy and up-regulation of senescence markers, chS913A mice exhibited more prominent effects at younger ages. Our results clearly suggest that inhibiting Ulk1 phosphorylation at Ser913 in cardiomyocytes promotes cardiac dysfunction and senescence. The results also suggest, however, that inhibiting Ulk1 phosphorylation at Ser913 in non-myocytes may not promote or even alleviate cardiac dysfunction or senescence, particularly at young ages. Reports by other investigators showing that senescence in non-myocytes has beneficial effects on cardiac regeneration^23,24 and pressure overload-induced fibrosis and cardiac dysfunction²⁵ lend support for this notion. Furthermore, cardiac fibroblast senescence was associated with better cardiac function and less cardiac fibrosis in aging osteopontin-deficient mice and in aged WT mice with visceral adipose tissue removal.²⁶ Thus, further investigation is needed to clarify how the GSK-3β-Ulk1 pathway affects the function of non-myocytes.

In summary, we report that chS913A mice exhibit premature aging in the heart, which is mediated through stimulation of senescence in cardiomyocytes. Our results advanced our understanding of the role of the GSK-3β-Ulk1-autophagy pathway in the heart during aging. Thus, these mice are useful for the study of aging and cellular senescence.

Translational perspective.

Aging can lead to cardiac hypertrophy, fibrosis, and dysfunction, even heart failure. Cellular senescence, particularly cardiomyocyte senescence, may contribute to the development of cardiac dysfunction during aging. However, the key factors involved in the progression of aging-induced cardiomyopathy remain to be determined. Our animal models of accelerated aging (systemic and cardiac-specific Ulk1S913A knock-in mice) will be very helpful for finding/testing strategies to improve the outcome of heart failure in aging populations. Furthermore, our findings suggest that cardiomyocyte senescence may play a causative role in aging cardiomyopathy. Targeting cardiomyocyte senescence may offer novel strategies to mitigate heart failure progression during aging.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Authors’ contributions

P.Z. and J.S. designed the experiments and wrote the paper; P.Z. and Y.T. conducted the in vitro and in vivo experiments; E.-A.S. isolated adult mouse cardiomyocytes and performed further RNA-seq analysis; K.T., S.I., M.M., and Y.N. conducted echocardiographic measurements of mice; Y.T. and P.Z. conducted the animal experiments and analyses; Y.T. conducted immunohistochemistry and histology analyses; J.S. generated project resources. All authors reviewed and commented on the manuscript.

Funding

This study was supported in part by U.S. Public Health Service grants HL91469, HL112330, HL138720, HL144626, and HL150881 (J.S.). This work was also supported by an American Heart Association Predoctoral Fellowship Award 915784 (E.-A.S.) and Merit Award 20 Merit 35120374 (J.S.), and by the Fondation Leducq Transatlantic Network of Excellence 15CVD04 (J.S.).

Data availability

The data underlying this article are available in the article and in its online supplementary material.