Ufmylation Suppresses Unfolded Protein Response to Prevent Peripartum Cardiomyopathy

Visual Abstract

Highlights

• •Disturbances in pregnancy-induced cardiac adaptation can result in peripartum cardiomyopathy.
• •Ufmylation is a novel ubiquitin-like protein modification that governs cellular homeostasis across various cell types. However, its role in pregnancy-induced physiological cardiac remodeling is not known.
• •Loss of UFL1 in the heart caused cardiac dysfunction after 1 cycle of pregnancy and heart failure after 3 cycles of pregnancy.
• •Loss of UFL1 evoked ER stress, suppressed mitochondrial oxidative metabolism, and induced persistent mTOR signaling.
• •Ufmylation represents a novel mechanism limiting pathological cardiac remodeling.

Summary

Ufmylation is a novel ubiquitin-like protein modification that plays a critical role in maintaining the homeostasis of different tissues, but its role in the heart remains poorly understood. Here, we showed that mice lacking UFM1 ligase 1 (UFL1), an enzyme essential for ufmylation, in the heart developed peripartum cardiomyopathy. Loss of UFL1 reversed pregnancy-induced adaptive cardiac transcriptome alterations. Moreover, loss of UFL1 triggered excessive endoplasmic reticulum stress, inhibited mitochondrial oxidative metabolism, and caused augmented mTOR signaling, leading to pronounced pathological remodeling and heart failure. These results demonstrate that ufmylation is essential for physiological cardiac remodeling and that disruption of ufmylation predisposes the heart to peripartum cardiomyopathy.

Adult hearts have the capacity to grow and adapt in response to increased demands or workload. This growth, known as cardiac hypertrophy, can be either physiological or pathological, depending on the nature of the stimuli.¹ Physiological stimuli, such as exercise and pregnancy, lead to reversible hypertrophy without adverse outcomes. In contrast, pathological stimuli, such as hemodynamic stress from hypertension or heart disease cause cardiac dysfunction and heart failure. Although rare, for reasons that remain unclear, exercise and pregnancy-induced cardiac hypertrophy can sometimes progress to heart failure or peripartum cardiomyopathy (PPCM),^2,3 which are the leading causes of accidental death among athletes and maternal mortality, respectively. The molecular mechanisms controlling physiological vs pathological cardiac hypertrophy could be distinct^1,4 but remain poorly understood. Therefore, a deeper understanding of how cardiac hypertrophy is regulated under physiological conditions could pave the way for developing approaches to address cardiomyopathies associated with athletes and pregnant women.

Ufmylation is a ubiquitin (Ub)-like protein modification by which ubiquitin-fold modifier 1 (UFM1) conjugates to its specific substrates via a ufmylation-specific E1-E2-E3 enzymatic cascade.⁵ The modifier UFM1 is a small 9.1 kilodalton protein and is synthesized as an inactive precursor form (pro-UFM1).⁵ Cleavage of its C-terminal dipeptide Ser-Cys by UFM1-specific cysteine proteases (UFSP1 and UFSP2) exposes a single glycine residue.5, 6, 7, 8 The mature form of UFM1 is further activated by E1 activating enzyme (UBA5) in an ATP-dependent reaction.⁵ The activated UFM1 is then brought to the cysteine residue of E2 conjugating enzyme (UFC1), catalyzing transthiolation of UFM1.⁹ The final step involves UFC1 guiding E3 UFM1-ligase 1 (UFL1) and targeted substrate to a close proximity and finally tethering UFM1 to the substrate.¹⁰

The ufmylation E1-E2-E3 enzyme conjugation is conserved among most eukaryotes except yeast and fungi.¹¹ Recent studies have identified an increasing list of UFM1 substrates such as UFM1 binding protein 1 (UFBP1), ASC1, CYB5R3, and RPL26.¹² These protein substrates are located at various subcellular compartments, including endoplasmic reticulum (ER), nucleus, ribosomes, and cytoplasm, and are involved in the regulation of ER homeostasis, transcriptional regulation, DNA damage repair, ER-phagy, and ER-associated quality control (CTL), underscoring the diverse biological functions of ufmylation.¹³ Systemic or cell type–specific deletion of genes encoding ufmylation components in different model systems demonstrated the physiological significance of ufmylation in development and the homeostasis of blood cells,^14,15 liver,¹⁶ brain,¹⁷ bone,¹⁸ and intestine.¹⁹ Moreover, allelic variants, loss-of-function sequence alterations, and aberrant expression of UFM1, UBA5, UFL1, and UFBP1 have been implicated in a wide range of human diseases, including encephalopathy,^17,20 Beukes familial hip dysplasia,^21,22 von Willebrand disease,²³ atrophic gastritis,²⁴ and cancer.²⁵

Despite the increasing appreciation of ufmylation in health and disease, the importance of ufmylation in the heart is not fully understood. We previously showed that ablation of Ufl1 in mouse hearts causes cardiomyopathy during ageing and exacerbates pressure overload-induced heart failure,²⁶ demonstrating a pivotal role of ufmylation in limiting pathological cardiac remodeling. However, whether ufmylation plays a role in physiological stimuli-induced cardiac remodeling, such as pregnancy-induced remodeling, remains unknown. To address this question, we generated α-myosin heavy chain (αMHC)-Cre-driven, cardiomyocyte-specific UFL1 knockout (UFL1CKO) mice to inhibit ufmylation in the heart. We then subjected these mice, along with CTL mice, to 1 or 3 cycles of pregnancy and characterized their cardiac phenotypes and transcriptomes. Our data show that UFL1CKO mice developed peripartum cardiomyopathy and heart failure, which can be attributed to pronounced ER stress, impaired fatty acid oxidative metabolism, and excessive mTOR activation. These findings demonstrate that ufmylation is essential for pregnancy-induced cardiac adaptation and identify it as a novel post-translational mechanism that protects against peripartum cardiomyopathy.

Methods

More methods can be found in the Supplemental Material.

Animal study

A transgenic mouse line bearing a UFL1 conditional knockout flox allele (UFL1^flox)²⁶ were crossed with αMHC (α-myosin heavy chain)^Cre/+ mice (the Jackson Laboratory, strain #011038) to generate cardiomyocyte-specific UFL1 knockout (UFL1^CKO) mice. Female UFL1^CKO and UFL1^flox/flox (CTL) mice were mated with male nontransgenic mice and subjected to up to 3 rounds of pregnancy. All mice were maintained on a C57BL/6J background. All animal experiments were approved by Augusta University Institutional Animal Care and Use Committee.

Echocardiography

Cardiac function was measured by echocardiography on female mice 1 week postpartum. Briefly, mice were anesthetized by inhalation of isoflurane (2.5% for induction, and 1.5% for maintenance) using a nose cone. The adequacy of anesthesia was monitored by toe pinch. Cardiac images and loops were captured using a VEVO 2100 echocardiography system with a linear 30-MHz transducer (VisualSonics). Cardiac functional parameters, such as ejection fraction (EF) rate, fractional shortening (FS) rate, left ventricular (LV) posterior wall thicknesses, LV internal dimension, LV volume, and LV mass, were analyzed off-line from the obtained M-mode images using VEVO 2100 software. Mice information, including genotyping and treatment, was blind to echocardiography performer.

RNA-seq analysis

Mouse ventricles were used for total RNA extraction using TRIzol reagent (Invitrogen, Cat#15596-026) following the manufacturer’s instructions. The isolated RNA was subjected to RNA-sequencing analysis performed by Novogene Corporation via next generation sequencing. For RNA library preparation, mRNA was enriched through poly (A) capture. Paired-end 150 strategy was adopted using the Illumina platform for sequencing. P values were adjusted with the Benjamini-Hochberg procedure. Differentially expressed genes were defined by P adj < 0.05 and a fold change larger than 1.5. Gene set enrichment software (GSEA) was used for pathway enrichment analysis.

Seahorse analysis

The oxygen consumption rate (OCR) was determined by Seahorse Bioscience XFe24 Extracellular Flux Analyzer (Agilent Technologies) using XF Cell Mito Stress Test kit (Agilent Technologies) according to the manufacturer’s instructions. Briefly, rat neonatal cardiomyocytes were seeded in Seahorse XF-24 plates at 50,000 cells/well in media containing 10% fetal bovine serum for 48 hours. Cells were then transfected with siRNAs containing 2% fetal bovine serum for another 48 hours. For OCR measurement the microplate was incubated in XF mitochondrial stress media (XF base medium add 10 mmol/L glucose, 2 mmol/l glutamine, 1 mmol/l pyruvate) at 37 °C for 1 hour without CO₂ before measurement. After OCR was recorded at baseline, oligomycin (2 μmol/L), FCCP (1 μmol/L), and rotenone/antimycin A (0.5 μmol/L) were sequentially added to reveal the key parameters showing metabolic function. Finally, basal respiration and maximal respiration were calculated. All measurements were normalized to total proteins. Mitochondrial OCR was calculated by subtracting the final OCR value (pmol O₂/min) after rotenone/antimycin A treatment from the average of the 3 OCR values before rotenone/antimycin A treatment.

Lipid droplets staining assays

Cells were stained with LipidTOX kit following the manufacturer’s protocol (Thermo Fisher #H34475). Briefly, after indicated treatments neonatal rat ventricular cardiomyocytes (NRVCs) grown on coverslips were fixed in 4% paraformaldehyde at room temperature for 10 minutes and subjected to LipidTOX labeling (1:1,000 in phosphate-buffered saline, RT for 30 minutes). The cells were counterstained with Alexa Fluor 568 phalloidin and 4′,6-diamidino-2-phenylindole to label cardiomyocytes and nuclei, respectively. Images were captured with ECHO Revolve fluorescence microscope (Echo) and were quantified using ImageJ. LipidTOX intensity was normalized with cell size. Six views/sample, 3 samples/groups were quantified.

Statistical analysis

Data are presented as the mean ± SEM. Standard or repeated measures analysis of variance with Tukey's post hoc test was used for within and between comparisons at multiple timepoints. Unpaired Student's t-test or Mann-Whitney U test was used for comparison between 2 groups for normally distributed and skewed data, respectively. Continuous data were evaluated for normality using the Shapiro-Wilk normality test. A 2-sided P value <0.05 was considered statistically significant. All statistical analyses were performed with GraphPad Prism version 10.1 (GraphPad Software).

Results

Pregnancy does not alter the expression of ufmylation enzymes in the heart

A direct link between ufmylation and PPCM has not been established. To investigate this, we first reanalyzed RNA sequencing data from LV tissues of nonfailing healthy donors and patients with PPCM (GSE141910).²⁷ This analysis did not reveal any significant alterations in the transcripts of ufmylation components, including the modifier UFM1 and key ufmylation enzymes such as UBA5, UFC1, UFL1, and UFSP2 (Supplemental Figure 1A). Similarly, we did not observe significant changes in the transcript or protein expression levels of these genes in mouse hearts subjected to 3 rounds of pregnancy (Supplemental Figures 1B to 1D). These findings suggest that alterations in ufmylation enzymes do not appear to directly contribute to PPCM development.

UFL1CKO sensitizes the heart to peripartum cardiomyopathy

Using αMHC^Cre-driven UFL1 knockout (UFL1CKO) mice, our previous study showed that UFL1-deficient hearts were functionally normal at 2 months of age but developed dilated cardiomyopathy after 4 months of age.²⁶ Interestingly, unlike other cardiomyopathies that typically exhibit cardiac hypertrophy in mouse models and humans, UFL1CKO mice showed reduced heart weight to tibial length ratio before and after the onset of cardiomyopathy,²⁶ suggesting a possible role of ufmylation in sustaining cardiac adaptation under physiological conditions.

We therefore sought to investigate the role of ufmylation in pregnancy-induced physiological cardiac remodeling. Two-month-old female CTL (Ufl1^flox/flox) and UFL1CKO mice were mated with male C57bl/6J mice and subjected up to 3 rounds of pregnancies. Age-matched female CTL and UFL1CKO mice without pregnancy were included in the study. Echocardiography measurements were recorded 1 week after each delivery (Figure 1A). After 1 pregnancy (1PP), both CTL and UFL1CKO mice showed an enlarged heart size and a similarly elevated heart weight to tibial length ratio compared with their nonpregnant counterparts (Figures 1B and 1C). Echocardiography revealed that 1PP CTL mice developed adaptive eccentric cardiac hypertrophy, as indicated by a significant increase in systolic LV wall thickness, diastolic LV internal diameters, diastolic LV volume, and LV mass, and no alterations in LV EF compared with those without pregnancy (NP) (Figures 1D to 1L). In contrast, 1PP UFL1CKO mice exhibited characteristics of dilated cardiomyopathy, as evidenced by a significant decrease in systolic and diastolic LV wall thickness and a significant increase in systolic and diastolic LV internal diameters and LV volumes, in comparison to NP UFL1CKO mice and 1PP CTL mice (Figures 1D to 1L, Supplemental Table 1). Moreover, 1PP UFL1CKO mice showed modestly but significantly compromised cardiac function compared with NP UFL1CKO mice and 1PP CTL mice (EF: 59.5 ± 6.5% CKO+1PP vs 72.3 ± 4.2% CKO and 76.7 ± 5.5% CTL+1PP) (Figure 1L). These data indicate that loss of UFL1 sensitizes the heart to peripartum cardiomyopathy.

Figure 1.

UFL1CKO Induces Cardiac Dysfunction After 1PP

(A) Schematic of experimental design. Two-month-old control (CTL) and UFL1 knockout (UFL1CKO) female mice were used and mated with C57bl/6J mice for 3 terms of pregnancy followed by echocardiography 1 week after delivery. (B) Gross morphology of the mouse heart. (C) Heart weight to tibia length ratio. (D) Representative M-mode images. (E to L) Quantification of echocardiographic parameters of CTL and UFL1CKO mice before (CTL, n = 12; conditional knockout [CKO], n = 15) and after 1 pregnancy (1PP) (CTL, n = 16; CKO, n = 13). Left ventricular diastolic posterior wall thickness (LVPWd) or left ventricular systolic posterior wall thickness (LVPWs), left ventricular diastolic internal diameter (LVIDd) or left ventricular systolic internal diameter (LVIDs), left ventricular diastolic volume (LVVd) and left ventricular systolic volume (LVVs), left ventricular (LV) mass and LV ejection fractions (EFs) are shown. Data are presented as mean ± SEM. An unpaired Student's t-test was used in C, and 1-way analysis of variance with post hoc Tukey test was used in E to L. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001. 2PP = 2 consecutive cycles of pregnancy; 3PP = 3 consecutive cycles of pregnancy; NP = without pregnancy.

Pregnant UFL1CKO mice develop heart failure

To further determine the impact of UFL1CKO on pregnancy-induced cardiac remodeling, we subjected CTL and UFL1CKO mice to 3 consecutive cycles of pregnancy (3PP). After 3PP, both CTL and UFL1CKO hearts had further elevated heart weight to tibial length ratio compared with their counterparts without pregnancy, and there is no difference in heart weight between CTL+3PP and UFL1CKO+3PP (Figures 2A and 2B). Echocardiography revealed a significant, progressive increase in LV wall thickness, stroke volume, and LV mass in CTL+3PP hearts. This occurred without alterations in LV chamber diameters, EF, and fractional shortening (Figures 2C to 2K, Supplemental Table 2), indicating that CTL hearts are tolerant of 3PP-induced cardiac stress. In contrast, pregnant UFL1CKO hearts exhibited gradual progression of LV wall thinning and chamber dilation, resulting in deteriorated cardiac function compared with UFL1CKO+NP hearts (Figures 2D to 2G, Supplemental Table 2). After 3PP, the EF of CTL hearts maintains at normal range (76.4% ± 2.6%), whereas those of UFL1CKO hearts dropped to 37.4% ± 14.3% (Figure 2J), indicative of heart failure.

Figure 2.

UFL1CKO Results in Heart Failure by 3PP

(A) Gross morphology of the mouse hearts NP and 3PP. (B) Heart weight to tibia length. (C) M-mode images. (D to K) Quantification of echocardiographic parameters before and after 3 rounds of pregnancy. LVPWd or LVPWs, LVIDs or LVIDd, stroke volume (SV), LV EFs, and fractionation shortening (FS) are shown. Data are presented as mean ± SEM. An unpaired Student's t-test was used in B and a repeated measures 2-way analysis of variance with Tukey's post hoc test was used in D to K. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001. Abbreviations as in Figure 1.

To determine the impact of UFL1 deficiency on cardiac remodeling, we assessed cardiomyocyte hypertrophy, stress marker expression, cardiac fibrosis, and cardiomyocyte cell death. Cardiomyocytes from UFL1CKO hearts exhibited a significantly larger cross-sectional area at baseline compared with those from CTL hearts, which was further increased after 3 pregnancies (Figures 3A and 3B). Although cardiac fibrotic areas were not significantly different between CTL+NP and CKO+NP hearts, more fibrosis was observed in CKO+3PP hearts compared with CTL+3PP hearts (Figure 3C). Consistently, transcriptomic analysis revealed significant up-regulation of multiple collagen genes (Col12a1, Col14a1, Col4a4, Col8a1, Col8a2, Col16a1) in CKO+3PP hearts (Figure 3D). Quantitative polymerase chain reaction (qPCR) analysis showed up-regulation of Nppa and Myh7 and down-regulation of Myh6, typical stress markers associated with pathological cardiac remodeling, in CKO hearts both at baseline and after pregnancy compared with CTL (Figure 3E). Interestingly, terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling staining did not reveal a significant difference in apoptotic cardiomyocytes between CTL+3PP and CKO+3PP hearts, suggesting that cardiomyocyte cell death is not the primary cause of cardiac dysfunction in CKO+3PP hearts (Figure 3F). Together, these findings suggest that UFL1 deficiency induces pathological cardiac remodeling after pregnancy, ultimately leading to cardiac decompensation and heart failure.

Figure 3.

UFL1 Deficiency Triggers Pathological Cardiac Remodeling Following Pregnancy

(A and B) Representative images (A) of wheat germ agglutinin (WGA) (green)-stained ventricular tissue sections showing cross-sectional cardiomyocyte area and the quantification (B). At least 500 cardiomyocytes per group were quantified, with a minimum of 150 cells per mouse heart and 3 mouse hearts per group. (C) Representative images of ventricular section stained by FastGreen and Sirius red. (D) Heatmap of indicated collagen genes that are significantly up-regulated in CKO+3PP hearts. Data were derived from RNA sequencing analysis of the indicated hearts. (E) Quantitative polymerase chain reaction analysis of indicated stress marker genes in the heart. (F) Representative images of transferase-mediated dUTP nick-end labeling (TUNEL)+ (red, arrowheads) cardiomyocytes in the hearts. The sections were counterstained with Phalloidin (green) and 4′,6-diamidino-2-phenylindole (blue) to mark cardiomyocytes and nuclei, respectively. Data are presented as mean ± SEM. Nested 1-way analysis of variance and 2-way analysis of variance followed by post hoc Tukey test were used in B and E, respectively. ∗P < 0.05; ∗∗∗∗P < 0.0001. Abbreviations as in Figure 1.

Loss of UFL1 reverses cardiac transcriptomic alterations induced by pregnancy

To gain mechanistic insights into postpartum cardiac remodeling, we conducted bulk RNA sequencing analysis on hearts with and without 3PP. Principal-component analysis identified unique gene expression profiles across 4 groups of mice (Figure 4A). In particular, gene expression patterns varied significantly between CTL+3PP and CTL+NP, as well as between UFL1CKO+3PP and CTL+3PP, consistent with the ongoing cardiac remodeling in CTL+3PP hearts and the compromised cardiac function in UFL1CKO+3PP hearts, respectively. Compared with CTL+NP hearts, there were 569 differentially expressed genes (DEGs) identified in CTL+3PP hearts (Figure 4B). GSEA revealed enrichment of up-regulated genes in oxidative phosphorylation (normalized enrichment score [NES] = 1.60), fatty acid metabolism (NES = 1.24), heme metabolism (NES = 1.52), reactive oxygen species, DNA repair, and E2F targets, whereas down-regulated genes are over-represented in apoptosis (NES = −1.21), unfolded protein response (NES = −1.37), cholesterol homeostasis (NES = −1.38), TNFα signaling via NFκB (NES = −1.44), and epithelial mesenchymal transition (NES = −1.46), among others (Figure 4C). UFL1 depletion resulted in profound alterations in cardiac transcriptome, indicated by 2003, and 1,829 DEGs identified at baseline and after 3 pregnancy cycles, respectively, compared with CTL hearts under the same conditions (Figure 4B). Notably, several pathways positively associated with CTL+3PP hearts, including oxidative phosphorylation, heme metabolism, and fatty acid metabolism, are negatively associated with UFL1CKO hearts, both with and without pregnancies. Conversely, nearly all of the top-enriched pathways negatively associated with CTL+3PP hearts show a positive association with UFL1CKO hearts under both conditions (Figure 4C). Overlapping DEGs between UFL1CKO and CTL hearts under NP and 3PP conditions identified 425 down-regulated and 648 up-regulated genes, which are mostly enriched in oxidative phosphorylation and cell death and extracellular matrix remodeling, respectively (Figure 4D). These findings suggest that pregnancy-induced cardiac adaptation involves the up-regulation of fatty acid oxidative metabolism and the repression of cell death, ER stress, inflammation, and extracellular matrix remodeling, which are reversed by UFL1 depletion.

Figure 4.

The Effects of ULF1CKO on Pregnancy-Induced Alterations in the Cardiac Transcriptome

(A) Principal-component analysis showing gene expression patterns among different groups (n = 4 mice per group). (B) Volcano plot showing significantly down-regulated (blue) and up-regulated (red) genes in CTL+3PP, CKO+NP and CKO+3PP hearts, respectively (fold change [FC] >1.5 or <−1.5 and Padj < 0.05). (C) Bubble plot showing the enrichment of differentially expressed genes in the indicated hallmark gene set. (D) Venn diagram showing the enrichment of genes commonly dysregulated in CKO+NP (vs CTL+NP) and CKO+3PP (vs CTL+3PP) hearts in the indicated pathways. Abbreviations as in Figure 1.

Loss of UFL1 induces ER stress in pregnant mice

Disruption of ER homeostasis leads to accumulation of misfolded proteins in the ER lumen, a condition referred to as ER stress.²⁸ ER stress triggers unfolded protein response, which helps cells to cope with stress by increasing protein folding capacity, attenuating protein synthesis and promoting ER-associated degradation. Previous studies suggested that ER stress contributes to the development of peripartum cardiomyopathy.²⁹ Consistently, GSEA showed a negative association between pregnancy and unfolded protein response in the hearts, which was reversed in UFL1CKO hearts, regardless of pregnancy status (Figure 5A). For example, pregnancy induced the down-regulation of chaperone genes (Hspa5, Hsp90b1, Dnajc3), transcription factors involved in ER stress response (Atf3, Xbp1), and genes regulating redox in the ER lumen and calcium homeostasis (Pdia5, Calreticulin) in the CTL+3PP hearts, compared with CTL+NP hearts. However, all of these genes were significantly up-regulated in UFL1CKO+NP hearts and, more remarkably, in UFL1CKO+3PP hearts (Figure 5B), indicating excessive ER stress in UFL1-deficient hearts. ATF4 is a key effector of unfolded protein response.^30,31 Cross-referencing of UFL1-regulated transcriptome in 3PP hearts with putative ATF4 targets³¹ showed that many genes up-regulated by UFL1CKO are ATF4 targets (Figure 5C). In line with the activation of ATF4 signaling in UFL1-deficient hearts, nearly 20% of up- and down-regulated genes induced by ATF4 overexpression in mouse hearts³² were also increased and reduced in UFL1CKO hearts, respectively (Figure 5D). Quantitative real-time PCR analysis validated the up-regulation of ER stress markers (Hsp90, Hsp70, and Pdi) in UFL1CKO+3PP hearts compared with CTL+3PP hearts (Figure 5E). Moreover, Western blotting confirmed the successful deletion of ULF1, the down-regulation of UFM1 substrate, UFBP1, and the up-regulation of ER stress response markers (BIP, GRP94, PDI) at the protein level in UFL1CKO+3PP hearts (Figures 5F and 5G). These findings suggest that uncontrolled ER stress contributes to the peripartum cardiomyopathy induced by UFL1 deficiency.

Figure 5.

Loss of UFL1 Induced ER Stress in the Hearts of Pregnant Mice

(A) Enrichment plot of unfolded protein response gene set in indicated hearts. (B) Heatmap showing the expression levels of ER stress response genes in the hearts. (C) Enrichment plot showing over-representation of putative ATF4 target genes (GSE35681) in CKO+3PP hearts. (D) Venn diagram showing commonly up-regulated (up) and down-regulated (bottom) genes in CKO+3PP hearts and ATF4 overexpressed (ATF4OE) cardiomyocytes. (E) Quantitative real-time polymerase chain reaction analysis of indicated ER stress response genes in the hearts. (F and G) Western blot (F) and quantification (G) of indicated proteins using total heart lysates. Data are presented as mean ± SEM. Two-way analysis of variance followed by Tukey’s post hoc test was used in E and G. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001. Abbreviations as in Figure 1.

We further investigated the link between ufmylation and ATF4 signaling. We found that pregnancy increased ATF4 transcript levels in both CTL and UFL1-deficient hearts; however, ATF4 proteins were barely detectable in mouse hearts, likely caused by its short half-life (Supplemental Figures 2A and 2B). In NRVCs, UFL1 silencing up-regulated baseline ATF4 transcript levels and further enhanced ER stress inducer Tunicamycin-induced ATF4 transcription and protein expression (Supplemental Figures 2C and 2D). Simultaneous silencing of ATF4 and UFL1 increased caspase-3 cleavage following Tunicamycin treatment and exacerbated cardiomyocyte cell death compared with single knockdowns (Supplemental Figures 2D to 2F), suggesting that the activation of ATF4 in UFL1-deficient cardiomyocytes is an adaptive response.

UFL1 deficiency impairs cardiac bioenergetics in postpartum mice

In the remodeling hearts, a metabolic shift from fatty acid oxidation to glucose metabolism leads to energy deficiency and underlies the pathogenesis of cardiomyopathy.^33,34 Our RNAseq data revealed up-regulation of genes involved in oxidative phosphorylation in CTL+3PP hearts and the down-regulation of genes in this category in UFL1CKO hearts (Figure 6A). These genes are either enzymes mediating fatty acid utilization (such as Echs1) or the components of electron transportation chain (such as Cox11, Uqcrb, Atp5g3) (Figure 6B). qPCR analysis validated the up-regulation of Cox11, a gene encoding a component of mitochondrial complex IV, in CTL+3PP hearts and its down-regulation in UFL1CKO hearts (Figure 6C). Meanwhile, Atp5g3, a gene encoding a subunit of mitochondrial ATP synthase, was also down-regulated in CKO+NP and CKO+3PP hearts (Figure 6C). Moreover, immunoblotting showed that NDUFB8, SDHB, UQCRC2, and ATP5A, the components of mitochondrial complexes I, II, IV, and V, were greatly reduced at protein levels in UFL1CKO hearts before and after pregnancies (Figures 6D to 6F). Although pregnancy increased mitochondrial DNA copy number relative to nuclear DNA, there was no significant difference in mitochondrial DNA copy number between CTL+3PP and CKO+3PP hearts (Supplemental Figure 3A), suggesting that loss of UFL1 does not affect mitochondrial abundance. However, UFL1 deficiency blunted pregnancy-induced up-regulation of mitochondrial genes (mt-Atp6 and mt-Nd1), which was associated with the down-regulation of Ppargc1a and Ppargc1b (Supplemental Figures 3B and 3C). These genes encode the transcriptional cofactors PGC-1α and PGC-1β, which are known regulators of mitochondrial gene expression. These findings suggest that UFL1 is required to sustain pregnancy-induced mitochondrial gene expression. In NRVCs, silencing UFL1 resulted in accumulation of lipid content (Figures 6G and 6H). Seahorse analysis revealed a significantly reduced mitochondrial respiration activity in UFL1-deficient cardiomyocytes as indicated by decreased basal and maximal respiration (Figures 6I and 6J). This was accompanied by increased mitochondrial ROS production (Figures 6K and 6L), indicating mitochondrial dysfunction. These in vivo and in vitro data suggest that loss of UFL1 impairs mitochondrial function and support that UFL1 promotes fatty acid oxidative metabolism in pregnancy-induced cardiac adaptation.

Figure 6.

UFL1 Deficiency Impaired Fatty Acid Oxidative Metabolism

(A) Gene set enrichment analysis of hallmark oxidative phosphorylation gene set. (B) Heatmap of genes involved in oxidative phosphorylation in the hearts. (C) Quantitative polymerase chain reaction analysis of indicated mouse hearts. (D to F) Western blot (D) and quantification (E and F) of the mitochondrial proteins. (G to L) NRVCs were transfected with siRNA against luciferase (siLuci) or UFL1 (siUFL1) for 48 hours before additional analyses. (G and H) LipidTOX (green) staining (G) and quantification (H) of LipidTOX fluorescent intensity. NRVCs were incubated with oleic acid (100 mmol/L) for 24 hours before the analysis. Lipid droplets were stained with LipidTOX (green) staining. Cardiomyocytes and nuclei were counterstained with Phalloidin (red) and DAPI (blue), respectively. (I and J) Seahorse analysis of oxygen consumption rate (OCR) (I) and quantification (J) of basal and maximal respiration. (K) Representative confocal images of NRVC stained with Mitotracker (deep red), MitoSOX (green), and Hoechst (blue). (L) Quantification of the fluorescent intensity in (J). Over 50 cells per group were quantified. Data are presented as mean ± SEM. Two-way analysis of variance followed by post hoc Tukey test was used in C, E, and F, and Mann-Whitney U test in H, J, and L. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Abbreviations as in Figure 1.

Persistent mTOR signaling in UFL1-deficient cardiomyocytes

Activation of mTORc1 signaling can exert both adaptive and maladaptive effects in the heart.³⁵ Interestingly, we found enrichment of the IGF1-mTOR pathway in CTL+3PP hearts, in which several upstream regulators (Igf1r, Pik3r1, Akt1, Pdk2) and downstream effectors (Eif4E) of the mTOR pathway were up-regulated in CTL+3PP hearts but down-regulated in UFL1CKO+NP and UFL1CKO+3PP hearts (Figures 7A and 7B). To validate that mTORc1 activity is altered by UFL1CKO, we measured the phosphorylation of P70S6K and 4EBP1, 2 mTOR downstream effectors. UFL1CKO hearts at 8 weeks of age displayed higher p-P70S6K levels and p-4EBP1 compared with CTL+NP and CTL+3PP hearts, which was accompanied by increased phosphorylation of AKT, a positive upstream regulator of mTOR (Figure 7C), indicating higher mTORc1 activity. To mimic pregnancy-induced cardiomyocyte hypertrophy, we treated NRVCs with IGF1. Silencing UFL1 increased phosphorylation of IGF1R, AKT1 and P70S6K at baseline and augmented IGF1-induced phosphorylation of AKT1 and P70S6K (Figure 7D). Consistent with hyperactivated mTORc1, UFL1 depletion in NRVCs led to increased protein synthesis indicated by increased puromycin incorporation into proteins and larger cardiomyocyte cell size (Figures 7E to 7H). These findings together suggest that loss of UFL1 leads to overactivation of mTOR signaling, which may contribute to pregnancy-induced cardiac dysfunction.

Figure 7.

UFL1 Deficiency Caused Hyperactivation of mTOR Signaling

(A) Gene set enrichment analysis of the IGF1-mTOR pathway gene set. (B) Heatmap of several genes involved in the IGF1 pathway. (C) Western blot of indicated proteins in the hearts. (D to H) NRVCs were transfected with indicated siRNAs, followed by IGF1 (100 ng/mL) treatment for indicated times (D) or 24 hours (E to H). (D) Western blot of indicated proteins. (E and F) Puromycin incorporation assay. Cells were labeled with puromycin (0.5 μg/mL) treatment for 6 hours. Western blot (E) and quantification (F) of puromycin-incorporated proteins. (G and H) Confocal images (G) and the quantification (H) showing cell size. Data are presented as mean ± SEM. One-way analysis of variance with post hoc Tukey test was used in F and H. ∗P < 0.05, ∗∗ P < 0.01, ∗∗∗ P < 0.001. Abbreviations as in Figure 1.

Discussion

In summary, our results demonstrate the pivotal role of UFL1-mediated ufmylation in regulation of pregnancy-induced cardiac adaptation. Mechanistically, we propose a model in which disruption of ufmylation in postpartum hearts evokes excessive ER stress, impairs cardiac biogenetics, and induces uncontrolled mTOR activation, eventually culminating in cardiac dysfunction and heart failure. Our studies uncover a novel post-translation mechanism regulating peripartum cardiomyopathy. Genetic variants of ufmylation components have been linked to various human diseases.^17,20 Our findings suggest that women carrying these variants may be at a higher risk for peripartum cardiomyopathy.

Recent investigations have begun to unveil the biological functions of the UFM1 system across a spectrum of physiological and pathological conditions. However, evidence revealing the role of ufmylation in the heart remains scant. We previously showed that UFM1 is expressed in adult mouse hearts at levels comparable to SUMO 1/2/3,²⁶ another ubiquitin-like protein extensively studied in cardiac contexts. The abundance of ufmylated proteins appeared to be increased in aging and pressure overloaded hearts.²⁶ It is also found that UFL1 expression is diminished in failing human hearts and ischemic hearts in mouse models.²⁶ Thus, alterations in the UFM1 system are associated with cardiac diseases with various etiologies. Upstream signaling controlling the stress-induced alterations of the UFM1 system is poorly understood. Pregnancy and lactation cycle do not seem to affect the homeostasis of ufmylation in the heart, because we did not observe any dysregulation of ufmylation enzymes in postpartum hearts (data not shown). Using UFL1CKO mice, we demonstrated the inhibition of ufmylation predisposes the heart to dilated cardiomyopathy and heart failure during aging and in response to hemodynamic stress. In this study, we showed that mice lacking UFL1 in the heart were unable to handle pregnancy-induced cardiac stress and quickly developed peripartum cardiomyopathy and heart failure (Figures 1 and 2). These observations support the critical role of ufmylation in both physiological and pathological cardiac remodeling, highlighting the need for future investigations into targeting ufmylation as a therapeutic strategy for cardiac diseases.

Our RNAseq data identify a series of molecular events underlying cardiac adaptation in postpartum hearts (Figure 4). These include up-regulation of genes related to oxidative phosphorylation, fatty acid metabolism and DNA damage repair, which presumably enhance cardiomyocyte performance and survival, and down-regulation of genes involved in ER stress, cell death, fibrosis and inflammation, which are generally believed to contribute to the decompensation of remodeling hearts. Notably, UFL1-deficient hearts failed to initiate such an adaptive response, which explains the cardiac dysfunction of these mice, suggesting a crucial role of ufmylation in controlling pathways contributing to cardiac compensation.

Conditional ablation of genes encoding the ufmylation major components in many cell types including blood cells, liver cells, and Paneth cells were consistently linked to elevation of ER stress markers.^14,36 In agreement with these studies, loss of UFL1 in postpartum hearts also increased ER stress markers including HSP90, HSP70, and PDI (Figure 5). Coupled with previous findings that loss of UFL1 triggered in ageing and pressure-overloaded hearts,²⁶ our findings from this study further strengthened the importance of ufmylation in regulation of ER homeostasis in cardiomyocytes. Prolonged or excessive ER stress can lead to maladaptive activation of unfolded protein response, promoting inflammation, apoptosis, and fibrosis, all of which contribute to the development of progression of various cardiac diseases, including ischemic heart disease, hypertrophy, and heart failure.^37,38 Protein kinase RNA-like endoplasmic reticulum kinase (PERK) is 1 of the 3 major transducers that mediate the unfolded protein response signaling to resolve the ER stress. A recent study reported that cardiomyocyte-specific PERK knockout mice developed peripartum cardiomyopathy after consecutive deliveries.^29,39 We also reported that UFL1 is required to activate PERK signaling in cardiomyocytes.²⁶ Therefore, deficient PERK signaling may induce ER stress and contribute to the pathogenesis of UFL1CKO-induced peripartum cardiomyopathy. In addition, given that ufmylation of RPL26 has been linked to ribosome-associated quality control at the ER,^40,41 it is possible that loss of UFL1 impairs translation-coupled protein quality control, leading to the accumulation of misfolded proteins in the ER and subsequent ER stress.

In this study, we identified a potential link between ufmylation and cardiac metabolism. This is supported by dysregulation of metabolic genes and down-regulation of several mitochondrial complexes in UFL1-deficient hearts (Figure 6). Although we cannot completely rule out whether such metabolic alterations are secondary to ongoing cardiac remodeling in UFL1CKO hearts, the deficits in fatty acid utilization and mitochondrial function in UFL1 knockdown cardiomyocytes suggest that UFL1 may have a direct effect on regulating cardiomyocyte metabolism. Deletion of Uba5 in Zebrafish caused mitochondrial damage in both the nervous system and the skeletal muscle.⁴² A proteomic study using UFM1 transgenic mice identified several mitochondrial proteins as potential UFM1 targets in the heart.⁴³ Deletion of Ufl1 in hematopoietic cells increased mitochondria mass and reactive oxygen species.¹⁴ Thus, our and other findings support the role of ufmylation in sustaining mitochondrial bioenergetics in the heart.

Another interesting finding from this study is the crosstalk between ufmylation and mTOR signaling, a master regulator of cell metabolism, growth and proliferation. The mTOR pathway has a pivotal regulatory function in unstressed and stressed hearts.³⁵ mTORC1 is required for embryonic and postnatal cardiac development and for cardiac adaptation to pressure overload. However, partial and selective pharmacologic and genetic inhibition of mTORC1 improves cardiac dysfunction and heart failure induced by increased load or genetic cardiomyopathies, chronic myocardial infarction and metabolic disorders. We showed that UFL1 deficiency led to augmented mTOR signaling in cardiomyocytes and mouse hearts (Figure 7). We speculate that while activation of mTOR signaling is necessary for initiating physiological hypertrophy in postpartum hearts, loss of UFL1 leads to persistent, uncontrolled mTOR signaling that culminates in peripartum cardiomyopathy. In support of this notion, mice with deficiency of ZFP36L2, an mRNA-destabilizing protein, in the heart exhibited mTOR activation and developed peripartum cardiomyopathy, and inhibition of mTOR with rapamycin ameliorated postpartum cardiac dysfunction.⁴⁴ Interestingly, a recent study showed that overexpression of UFL1 enhanced mTOR signaling in bovine mammary epithelial cells.⁴⁵ Another showed that deletion of UFBP1, the UFM1 substrate, impaired mTOR signaling in mouse embryonic fibroblasts.⁴⁶ Hence, the impact of ufmylation on the mTOR pathway can be cell type–specific and context-dependent.

Study limitations

Although our data demonstrate the protective role of ufmylation in preventing peripartum cardiomyopathy, several limitations of this study should be noted:

• 1.The link between ufmylation dysregulation and peripartum cardiomyopathy in humans remains unestablished, which could be addressed by examining ufmylation in ventricular samples from patients with peripartum cardiomyopathy.
• 2.The current studies on mitochondrial and cardiac metabolism are incomplete; further research is necessary to fully elucidate the effects of ufmylation on mitochondrial respiration and function in postpartum hearts.
• 3.Despite the growing list of putative UFM1 targets, how ufmylation of protein substrates regulates key cellular processes, such as ER stress and mTOR signaling, in cardiomyocytes are not well understood. Identifying UFM1 substrates in cardiomyocytes may enhance our understanding of the biological role of ufmylation in the heart and facilitate the development of therapeutic strategies targeting ufmylation for treating cardiac disease.

Conclusions

These results demonstrate that ufmylation is essential for physiological cardiac remodeling during the peripartum period through a proposed mechanism involving repression of ER stress and maladaptive mTOR signaling and promotion of mitochondrial function. Targeting ufmylation could be a potential therapeutic approach for preventing peripartum cardiomyopathy.

Perspectives.

COMPETENCY IN MEDICAL KNOWLEDGE: Peripartum cardiomyopathy is a leading cause of maternal death, highlighting the urgent need for new therapeutic targets. Ufmylation, a post-translational protein modification, plays a critical role in maintaining cellular homeostasis across various cell types. In this study, using cardiomyocyte-specific UFL1 knockout mice, we demonstrate that UFL1 deficiency in the heart promotes peripartum cardiomyopathy. Thus, targeting the ufmylation pathway could offer a potential strategy to address cardiac dysfunction associated with pregnancy and childbirth.

TRANSLATIONAL OUTLOOK: Genetic variants in ufmylation components are associated with several human diseases but have not yet been linked to cardiac conditions. Our findings from a cardiomyocyte-specific UFL1 conditional knockout mouse model suggest that women carrying these variants may be at increased risk for peripartum cardiomyopathy. Our mechanistic data support testing the effectiveness of chemical chaperones, mitochondrial antioxidants, and mTOR inhibitors in mitigating the development of peripartum cardiomyopathy.

Funding Support and Author Disclosures

This study was in part supported by the U.S. National Institutes of Health grants (1R01HL146807-01A1 to Dr Li; and 2R01HL124248-06A1, 1R01HL165205, 1R01HL173796 to Dr Su) and the American Heart Association grants (23TPA1077767 to Dr Li and 24POST1187422 to Dr Tandra). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.