Activation of FoxO1 prevents and reverses cardiac hypertrophy from diverse stimuli

Abstract

The heart is a dynamic organ capable of structural and functional remodeling in the wake of changing mechanical and/or circulating cues. While the molecular underpinnings of cardiac hypertrophy are well-defined, the mechanisms of hypertrophy regression following stimulus removal are relatively less understood. Here, we demonstrate that activation of forkhead box proteins (FoxOs), and increased expression of their autophagy gene targets, are common features of hypertrophy regression after both exercise and pregnancy in mice. Additionally, we show FoxO1 activation is sufficient to prevent and reverse adrenergic agonist-dependent pathological hypertrophy. Our findings highlight the central role of FoxO1 in regulating cardiac mass.

Graphical Abstract.

Created with BioRender.

1. Introduction

The heart exhibits remarkable plasticity and can adapt to external cues by structurally remodeling to meet new functional demands. Cardiac remodeling can be either physiological or pathological, depending on the nature and duration of the stimulus [1]. Common examples of adaptive physiological remodeling include hypertrophy that occurs with endurance exercise training or pregnancy [1]. In these cases, hypertrophy stems from increased circulatory demands and is associated with both increased left ventricular (LV) chamber diameter and wall thickness [1–3]. Physiological cardiac hypertrophy is typically associated with maintained or enhanced contractility and is completely reversible with removal of the stimulus [1–3]. Cardiac hypertrophy is also a leading risk factor for heart disease and this pathological type of remodeling, which is associated with increased LV wall thickness and reduced chamber volume, can arise from chronically elevated circulatory pressure/volume or circulating neurohumoral factors [1]. This remodeling is typically less malleable than physiological forms, although regression of pathological hypertrophy does occur in some patients and is associated with improved outcomes [1,4]. While the mechanisms of cardiac remodeling in health and disease settings are well described [2], the molecular regulation of reverse remodeling is relatively less understood.

In our previous work studying Burmese pythons, which exhibit robust and reversible cardiac hypertrophy after feeding [3], we found that activation of forkhead box protein O1 (FoxO1)-dependent autophagy contributed to hypertrophy regression [5]. In the present study, we sought to identify whether FoxO signaling was implicated in mammalian examples of reversible cardiac remodeling from exercise and pregnancy. Moreover, we explored whether activation of FoxO1 could prevent/reverse pathological cardiac remodeling.

2. Methods

2.1. Ethical Oversight

Animal studies were approved by the University of Colorado Boulder IACUC.

2.2. Exercise Model

Hamster wheels were introduced into the cages of 5-week-old C57BL/6J female mice for four weeks, as previously described [6]. After four weeks of voluntary running, hearts were collected as the hypertrophy timepoint (0 days post running, DPR). Hearts from mice in the de-training groups were collected at 3 and 7 DPR. Sedentary, age-matched mice were used as controls.

2.3. Pregnancy Model

Female C57BL/6J mice, aged 8–10 weeks, were mated with males. Mice were weighed twice weekly and those with significantly increased body weight at 13 days post-copulation were considered pregnant. A subset of mice at this stage (mid-pregnancy, MP) were euthanized and hearts collected. Seventeen days post-copulation was designated as late pregnancy (LP). After parturition, the pups were removed from the cage to stop lactation. These mice were euthanized on post-partum (PP) days 7 or 21 and their hearts collected. Age-matched females served as non-pregnant (NP) controls.

2.4. Isoproterenol Model

Seven-week-old male C57Bl6/J mice received 7.5 × 10¹¹ viral genomes of AAV9-cTNT-FOXO1_CA or AAV9-cTNT-FOXO1_ΔDBD diluted to 100 μl in saline by tail vein injection. Three weeks after AAV delivery, osmotic minipumps (Alzet, Model 2001) containing isoproterenol (ISO) or vehicle control (1 μM L-ascorbic acid in saline) were implanted subcutaneously posterior to the scapulae. Mice received 30 mg/kg/day isoproterenol for seven days, after which they were euthanized and body, heart, and LV weights were measured.

2.5. Primary Myocyte Isolation and Culture

NRVMs were isolated from 1–2-day-old Sprague-Dawley rat pups as previously described [5] and cultured in experimental media (MEM, 20 mM HEPES, 20 μM BrdU, 2 μg/mL vitamin B12, 50 U/mL PCN-G, 20 U/mL Pen-Strep) for 18 hours prior to the start of experiments. Insulin-like growth factor-1 (IGF1) and phenylephrine (PE) were used at final concentrations of 15 nM and 10 μM, respectively. Adenovirus constructs expressing constitutively active (CA) FoxO1 or a DNA-binding-domain deficient (ΔDBD) FoxO1-CA were used at 20 or 40 MOI, as described in the figure legends. For autophagic flux experiments, cells were treated with 50 nM bafilomycin A1 (BafA1) for 6 hours.

2.6. Protein Isolation and Immunoblotting

Protein was isolated in RIPA buffer supplemented with protease and phosphatase inhibitors by scraping (cells) or pulverizing while frozen (tissue). BCA assays were performed to determine protein concentration and 10 μg (cells) or 20 μg (tissue) was combined with equal volume SDS loading buffer supplemented with DTT reducing agent and boiled for 10 minutes. Proteins were separated by electrophoresis on 4–12% Bis-Tris gels (Invitrogen), transferred onto nitrocellulose membranes. Membranes were blocked with 5% milk in TBS-T for one hour prior to overnight incubation with primary antibodies at 1:1000 in 5% BSA: p-FoxO1/3 T24/32 (Cell Signaling, 9464), FoxO1 (Cell Signaling, 14952), LC3B (Cell Signaling, 3868), GAPDH (Cell Signaling, 2118). After washing with TBS-T, secondary antibodies (IR-Dye 680 or 800, LI-COR, 1:4000) were added in blocking buffer for 1 hour. Image acquisition was performed on an IQ-800 imager (Cytiva).

2.7. RNA Extraction and qPCR

RNA from NRVMs was extracted using the RNAeasy miniprep kit (Qiagen) according to the manufacturer protocol. RNA from tissue was extracted by the phenol-chloroform method with Trizol, precipitated with isopropanol, and RNA pellets washed twice with ice-cold 75% ethanol. RNA was dissolved in RNase-free water and 500 ng used for cDNA synthesis with Superscript III reverse transcriptase (Invitrogen) and random hexamer primers according to the manufacturer protocol. 5 ng cDNA was used for qPCR with SYBR green and 0.5 μM gene-specific primers (Table 1).

Table 1.

Primers for qPCR

Gene	Species	Forward (5’-3’)	Reverse (5’-3’)
Bnip3	Mus musculus	CTGAGTAGCAAGTAGAAGCTAAGG	AACTGACCACCCAAGGTAATG
Col1a1	Mus musculus	CGATGGATTCCCGTTCGAGTACG	CTTGCAGTGATAGGTGATGTT
Gabarapl1	Mus musculus	GGGTGGAGCGATTGTGATAA	CATGATGGACCTCAGGATGTAG
Il1b	Mus musculus	GGACCCATATGAGCTGAAAG	CGTTGCTTGGTTCTCCTTGTA
Map1lc3b	Mus musculus	GCGGGTGATTATAGAGCGATAC	CAAGCGCCGTCTGATTATCT
Nppb	Mus musculus	GTCCAGCAGAGACCTCAAAA	AGGCAGAGTCAGAAACTGGA
Postn	Mus musculus	CTGAAACACGGCATGGTTATTC	AATCTGGTTCCCATGGATGAC
Tnfa	Mus musculus	GCCGATGGGTTGTACCTTGTC	GAGATAGCAAATCGGCTGAC
Ulk1	Mus musculus	CAGGGTGGACACATGCTAATAC	CAGCTTGTGGACACTCAGATAC
Gapdh	Mus musculus	AGGTCGGTGTGAACGGATTTG	TGTAGACCATGTAGTTGAGGTCA
18s	Mus musculus	GCAATTATTCCCCATGAACG	GGCCTCACTAAACCATCCAA
Bnip3	Rattus norvegicus	GAGCTGAAATAGACACCCACAG	CCGACTTGACCAATCCCATATC
Gabarapl1	Rattus norvegicus	TTTGACCTCTGCCCTAATTCC	ATGTCCGTGCGAATGTCTAC
Map1lc3b	Rattus norvegicus	CCGAAACAGGTCAGGTGTATAG	CCACACTGCTGAGGTGAAA
Ulk1	Rattus norvegicus	GGCTTACAGACTGCCATTGA	GATACCACGCTGGCCTTATAC
Gapdh	Rattus norvegicus	GCAAGGATACTGAGAGCAAGAG	GGATGGAATTGTGAGGGAGATG

2.8. Cell Volume

NRVMs were trypsinzed, resuspended in 1 mM EDTA, 5% calf serum in PBS, and stored on ice. Cell volume was measured using a Multisizer 3 Coulter Counter (Beckman) with a 100 μm aperture for 60 seconds and the percent change in mean volume (representing 10,000–30,000 cells) plotted as a single data point.

2.9. Immunostaining

NRVMs were fixed by serial incubation in ice-cold 100% methanol (1 minute) and 4% paraformaldehyde (5 minutes). Permeabilization and antigen retrieval were performed in 0.1% triton (20 minutes) and 0.1M glycine pH 3.5 (30 minutes), respectively. The cells were blocked in 5% BSA for one hour and α-actinin primary antibody (Thermo, MA5–31539, 1:300) was added in blocking buffer overnight. After washing with PBS, Alexa-Fluor anti-MS 555 secondary antibody (Cell Signaling, 4409, 1:1000) was added in blocking buffer for one hour. Images were acquired on a Nikon Ti-E microscope. Image analysis was performed with Fiji.

2.10. Adenovirus and AAV Generation

Plasmid construction of GFP-FOXO1_ΔDBD and GFP-FOXO1_CA and adenovirus production were described previously [5]. The AAV production protocol was adapted from that previously described [7]. Briefly, GFP-FOXO1_ΔDBD and GFP-FOXO1_CA constructs were subcloned into an adeno-associated virus serotype-9 (AAV9) vector under the cardiac troponin T (cTNT) promoter and transformed into Sure2 competent cells. Plasmid DNA was transfected into HEK 293T cells along with pAd helper and pAAV.RepCap using calcium-phosphate precipitation. Media and cell pellets were harvested 72 hours later. The clarified viral lysate was fractionated on an Iodixanol gradient and virus collected and concentrated with a Vivaspin 6 concentrator. Titers were determined by qPCR for GFP-FOXO1 using two different primer sets (1. F: CGTGCCCTACTTCAAGGATAAG, R: GGATTGAGCATCCACCAAGA; 2. F: CCATTACCTGTCCACACAATCT, R: TCCATGCCATGTGTAATCCC) and the average used to assign genome copies/mL.

2.12. Statistical Analysis

The data throughout are presented as the mean ± standard error. Analyses of two groups were performed by two-tailed t-test. Analyses of more than two groups were performed by one-way or two-way ANOVA, as appropriate, with Tukey’s post-hoc test for multiple independent pairwise comparisons. The Shapiro-Wilk Test was used to confirm the data were normally distributed. If data did not fit Gaussian distribution, a Kruskal-Wallis test was used instead. A p-value < 0.05 was considered statistically significant. Data analysis and presentation were performed using GraphPad Prism (version 10).

2.13. Data Availability

The data in support of the study are reported herein or in the associated supplemental files. Requests for additional data, reagents, or study materials should be directed to the corresponding author L.A.L (leslie.leinwand@colorado.edu).

3. Results and Discussion

To study the effects of endurance exercise training and de-training on cardiac remodeling and FoxO transcription factor activity, we used a mouse voluntary cage-wheel model [8]. Previous work showed that exercise-dependent FoxO signaling changes were similar between male and female mice in a swimming model [9]. However, we chose to specifically study female mice here as they exhibit more robust cardiac hypertrophy in the voluntary cage-wheel model [10]. The mice ran up to 14-km per day and exhibited a ~25% increase in cardiac mass versus age-matched, sedentary controls after four weeks [6]. The wheels were then removed to induce de-training, and hearts were collected at 3- and 7-days post-running (DPR). Cardiac mass was sustained at 3DPR but significantly regressed by 7DPR (~12% greater than sedentary controls) [6]. To examine the effect of exercise and de-training on FoxO activity, we measured phosphorylation of FoxO1/3 at T24/T32, sites of inhibitory regulation by protein kinase B (Akt) that inhibit transcriptional activity of FoxOs by precluding their nuclear translocation [11]. FoxO phosphorylation increased at 0DPR (peak hypertrophy) and then significantly decreased by 7DPR (Fig. 1A–B) suggesting increased activity. FoxOs regulate expression of proteolysis genes [12], and our previous work identified that FoxO-dependent autophagy contributed to regression of adaptive cardiac hypertrophy in Burmese pythons [5]. Therefore, we next measured gene expression of FoxO targets that regulate autophagy (Bnip3, Map1lc3b, and Ulk1) and found levels increased at 7DPR (Fig. 1C). These results suggest that increased FoxO-dependent autophagy occurs with de-training.

Figure 1. Regression of physiological cardiac hypertrophy coincides with activation of FoxO signaling.

A. Western blot for p-FoxO1/3 (T24/T32), total FoxO1, and GAPDH in the exercise/de-training model; DPR = days post-running. B. p-FoxO1/3 normalized to total FoxO1. C. Expression of FoxO-regulated autophagy genes normalized to 18S and plotted as a fold-change vs. the Sedentary group. For panels B-C, n = 6 mice/group and data were analyzed by one-way ANOVA, Tukey post-hoc. D. Heart weight normalized to tibia length (TL) in the pregnancy model; n = 14 NP, 11 MP, 9 LP, 16 7PP, 7 21PP; one-way ANOVA, Kruskal-Wallis test. E. Western blot for p-FoxO1/3 (T24/T32), total FoxO1, and GAPDH in mice from the pregnancy timepoints. F-G. p-FoxO1 (T24) normalized to FoxO1 (F) and p-FoxO3 (T32) normalized to GAPDH (G); n = 5 NP, 4 MP, 4 LP, 5 7PP, 5 21PP; one-way ANOVA, Tukey post-hoc. H. Expression of FoxO-regulated autophagy genes normalized to Gapdh and plotted as a fold-change vs. NP; n = 7 NP, 4MP, 4LP, 7 7PP, 6 21PP; one-way ANOVA, Tukey post-hoc. I. Experimental paradigm for IGF1 treatment and chase experiments in NRVMs. J. Representative western blot for p-FoxO1/3 (T24/32), total FoxO1, and GAPDH. K. p-FoxO1 (T24) normalized to total FoxO1; n = 6 biological replicates per group; two-way ANOVA, Tukey post-hoc. L. Percent change in cell volume with 24 hours IGF1 treatment and 24 hours after IGF1 withdrawal (Chase); n = 6 biological replicates per group; two-tailed t-test. M. Experimental paradigm for IGF1 treatment/Ad-FoxO1 transduction experiments in NRVMs. N. Percent change cell volume with IGF1 ± FOXO1 adenovirus transduction; n = 6 biological replicates per group; one-way ANOVA, Tukey post-hoc. The data are plotted as the mean ± SEM.

Another setting of adaptive cardiac remodeling in mammals is pregnancy, where volume overload-induced stretch triggers eccentric cardiomyocyte hypertrophy [13]. To examine FoxO activity during pregnancy and with post-partum (PP) reverse remodeling, we collected hearts from mice during mid pregnancy (MP, 13 days), late pregnancy (LP, 17 days), seven days PP (7PP), and 21PP. Pups were removed from the cages after birth to stop lactation, which has previously been shown to contribute to an extended period of cardiac hypertrophy [13,14]. As expected, heart mass increased during pregnancy (+27% vs. non-pregnant) and showed signs of regression by 7PP (Fig. 1D). By western blot, we found p-FoxO1 at T24 significantly decreased at 7PP compared to LP (Fig. 1E–F). Phosphorylation of FoxO3 at T32 followed a similar trend (Fig. 1E, 1G). As with exercise, regression of pregnancy-induced hypertrophy corresponded to increased expression of FoxO-regulated autophagy genes (Fig. 1H). These data indicate that the activation of FoxO transcription factors and increased autophagy gene expression are shared features of cardiac reverse remodeling after exercise and pregnancy.

Insulin-like growth factor-1 (IGF1) is a well-established circulating cause of physiological cardiac hypertrophy and inhibits FoxO1/3 through the phosphoinositide-3-kinase (PI3K)/Akt signaling cascade [1]. Treatment of neonatal rat ventricular myocytes (NRVMs) with IGF1 (Fig. 1I) increased inhibitory phosphorylation of FoxO1 (Fig. 1J–K) and caused a 28% increase in cell volume after 24 hours (Fig. 1L). FoxO3 T32 phosphorylation also increased with IGF1-mediated hypertrophy in NRVMs (Fig. 1J). Changes in FoxO1/3 phosphorylation and cell size were rapidly reversed after IGF1 removal (Fig. 1K–L). To test whether activation of FoxO-regulated gene programs could override the hypertrophic stimulus from IGF1, we transduced IGF1-treated NRVMs with adenoviruses expressing either a constitutively active FoxO1 (CA), where the inhibitory phosphorylation sites T24, S256, and S319 were mutated to alanine (phospho-null), or a DNA-binding domain deficient FoxO1 (ΔDBD) (Fig. 1M). Transduction with FoxO1-CA significantly reduced IGF1-dependent cardiomyocyte hypertrophy compared with Empty vector control and FoxO1-ΔDBD (Fig. 1N). Given the overlapping functions of FoxO1 and FoxO3 [15], and our observations of their matching regulation in the context of physiological hypertrophy, we expect that constitutive activation of FoxO3 would also lead to a reduction in cardiomyocyte volume, as previously suggested [16]. However, as our previous investigations into adaptive hypertrophy in pythons highlighted a central role for FoxO1 specifically in hypertrophy regression [5], we elected to maintain a focus on FoxO1 for the remainder of this study.

We next examined whether activation of FoxO1 could reverse and/or prevent hypertrophy from pathological stimuli. While exercise, pregnancy, and IGF1 each modulate PI3K/Akt signaling, which regulates FoxO1, pathological hypertrophy stimuli act through distinct pathways [1]. We transduced NRVMs treated with the adrenergic receptor agonist phenylephrine (PE) with the FoxO1 adenoviruses and found that FoxO1-CA potently reversed PE-induced hypertrophy, while FoxO1-ΔDBD had no effect (Fig. 2A–B). Regression of cell size was associated with increased autophagy gene expression (Fig. 2C) and elevated autophagic flux as indicated by the relative increase in the late-stage autophagosome marker LC3-II after lysosome inhibition with bafilomycin A1 (Fig. 2D–G). Next, to determine whether these effects would translate to an in vivo setting, we used a mouse model with chronic exposure to the adrenergic receptor agonist isoproterenol (ISO). Male mice were chosen for this experiment as prior studies reported their cardiomyocytes had stronger responses to adrenergic stimulation [17,18]. We delivered serotype 9 adeno-associated viral vectors (AAV9) expressing either FoxO1-CA or FoxO1-ΔDBD under the cardiomyocyte-specific troponin T (cTNT) promoter to mice and three weeks later implanted osmotic minipumps to deliver ISO (Fig. 2H). Increased heart and LV weights were observed after one week of ISO, which were unaffected by FoxO1-ΔDBD; however, expression of FoxO1-CA completely prevented the development of cardiac hypertrophy (Fig. 2I–J). Notably, delivery of exogenous FoxO1 led to a modest, but significant reduction in endogenous FoxO1 protein levels (Fig. 2K–L), suggesting the presence of a regulatory feedback loop that tightly controls total FoxO1 abundance. As observed in vitro, expression of FoxO1-CA but not FoxO1-ΔDBD led to increased autophagy gene expression (Fig. 2M). Finally, to determine the effect of FoxO1-CA on pathology-associated signaling in the mouse heart, we measured expression of genes involved in extracellular matrix remodeling (Col1a1, Postn), hemodynamic stress (Nppb), and inflammation (Il1b, Tnfa). FoxO1-CA did not mitigate the ISO-dependent increase in expression of these genes (Fig. 2N), indicating that limiting pathological hypertrophy does not necessarily prevent disease-associated transcriptional changes.

Figure 2. Constitutive activation of FoxO1 is sufficient to reverse and prevent cardiac hypertrophy from adrenergic agonism.

A. Experimental paradigm for studying regression of PE-induced hypertrophy in NRVMs ± AdFOXO1. B. Percent change cell volume with PE ± FOXO1 adenoviruses; n = 5 biological replicates per group. C. Expression of FoxO-regulated autophagy genes normalized to Gapdh and expressed as a fold-change vs. Control; n = 6 biological replicates per group. D. Representative western blot for LC3B ± BafA1 to monitor autophagic flux. E. LC3-II expression normalized to GAPDH; n = 6 biological replicates per group. F. Representative immunofluorescence images of NRVMs transduced with the FOXO1 adenoviruses ± BafA1; Red = LC3B, representing autophagosomes, Green = GFP-FoxO1. G. Normalized autophagosome area per cell; n = 50–90 cells per group; two-way ANOVA with Tukey’s post-hoc test. H. Experimental paradigm for in vivo AAV9-FOXO1 delivery and isoproterenol treatment in mice (Created with BioRender). I-J. Heart weight (I) and LV weight (J) normalized to body weight. K. Representative western blot for FoxO1 in the AAV9-FOXO1/ISO experiments. L. Expression of endogenous FoxO1 and GFP-FoxO1 normalized to GAPDH. M. Expression of autophagy genes normalized to Gapdh and expressed as a fold-change vs. Vehicle. N. Expression of heart disease-associated genes normalized to Gapdh and expressed as a fold-change vs. Vehicle. For panels I-N, n = 6 Veh, 6 ISO, 6 AAV-FOXO1_ΔDBD/ISO, 5 AAV-FOXO1_CA/ISO. For all except G, statistical analysis was performed by one-way ANOVA with Tukey’s post-hoc test. The data are plotted as the mean ± SEM.

Our results add to findings from previous studies that highlight the important role of FoxO transcription factors in regulating cardiac mass [9,16,19,20] and support the conclusion that FoxO activation is a conserved mechanism regulating adaptive, physiological cardiac remodeling from pythons [5] to mammals. Moreover, our findings show that FoxO1 activation is sufficient to prevent and reverse pathological cardiac hypertrophy, but not heart disease-associated transcriptional changes.

Highlights.

• FoxOs are activated during hypertrophy regression after exercise and pregnancy

• Expression of FoxO-regulated autophagy genes increases during regression

• Activation of FoxO1 can prevent and reverse pathological hypertrophy

• Preventing pathological hypertrophy does not limit disease-linked gene expression