Pregnancy precipitates metabolic imbalance and accelerates death in an animal model of mitochondrial cardiomyopathy

Abstract

During pregnancy, the heart undergoes major physiological and metabolic changes to increase cardiac workload, and the demand for energy production is especially elevated during the trial of labor. Normally, cardiac structure and metabolism revert to the pre-pregnancy state shortly after delivery. However, in some cases, peripartum/postpartum cardiomyopathy (PPCM) occurs, which increases a person's risk of major cardiac events following pregnancy. The molecular mechanisms underlying PPCM remain poorly understood. In this study, we investigate the transcriptional, metabolic, and bioenergetic profiles of postpartum (PP) hearts in a mouse model of cardiomyopathy caused by the pathogenic p.S55L mutation in the mitochondrial protein coiled-coil-helix-coiled-coil-helix domain containing 10 (CHCHD10). Heterozygote p.S55L mutant CHCHD10 mice develop acute heart failure during the immediate PP period. We observe cardiac remodeling, mitochondrial stress, and profound metabolic rewiring in PP mutant CHCHD10 hearts. Metabolic rewiring results in decreased levels of heme and the depletion of key cofactors of energy metabolism, including NAD(H) and ADP. These findings suggest that mutant CHCHD10 hearts fail to meet the increased energy demands associated with the trial of labor due to the insufficient turnover rate of NAD+/NADH and ADP/ATP. We propose that this metabolic insufficiency drives PP mortality in mutant CHCHD10 mice. In support of this hypothesis, dietary supplementation with nicotinamide riboside and pterostilbene, a naturally derived polyphenol, increased PP survival and cardiac energy metabolites in mutant CHCHD10 mice. Our work provides novel insights into the molecular mechanisms of PP cardiomyopathy associated with mitochondrial stress and suggests potential benefits of dietary NAD(H) supplementation.

Highlights

• •p.S55L CHCHD10 mutation causes postpartum mortality in female mice.
• •Mitochondrial integrated stress response favors antioxidant over energy metabolites.
• •Pregnancy induces a depletion of key energy metabolites in D10^S55L hearts.
• •High energy demands during labor precipitate metabolic failure in D10^S55L hearts.
• •Nicotinamide riboside and pterostilbene improve PP survival in D10^S55L mice.

1. Introduction

Postpartum (PP) maternal mortality remains alarmingly high, with a rate of 32.9 per 100,000 live births in 2021 in the United States [1]. Cardiovascular diseases, including peripartum/postpartum cardiomyopathy (PPCM) and coronary heart disease, are among the leading causes of PP morbidity and mortality [2]. Although socioeconomic status and the level of PP care can influence the mortality rate, the underlying mechanisms leading to PPCM are not well understood. PPCM is clinically defined as (1) the development of the disease in the last month of pregnancy or within 5 months of delivery, (2) absence of pre-existing heart disease prior to the last month of pregnancy, (3) unknown cause of heart failure, and (4) left ventricular systolic dysfunction [3]. Prognosis remains poor, with full recovery reported in only 23% of affected individuals and 50% experiencing heart failure-related mortality due to limited therapeutic options [[4], [5], [6]]. Limited studies in both humans and mouse models of PPCM have proposed several potential mechanisms, including inflammation [7], viral myocarditis [8], autoimmune reactions [9], oxidative stress [10], and apoptosis [11], resulting from environmental [12,13] as well as genetic factors [14]. Studying these mechanisms in animal models, particularly those involving genetic causes, has been difficult due to the lack of severity or relevance of existing mouse models of PPCM to the human disease [15].

Mitochondrial stress plays a significant role in the pathophysiology of cardiac diseases [16]. Primary mitochondrial cardiomyopathies (PMC) can arise from impairments of mitochondria within the myocardium. This condition, stemming from genetic mutations affecting mitochondrial proteins, can lead to diverse clinical manifestations, including tissue remodeling with cardiac fibrosis, arrhythmias, and heart failure [17]. Among the factors influencing the progression of PMC, mechanical and metabolic stress stand out as critical features capable of exacerbating the disease. The interplay between mitochondrial stress and metabolic alterations in PMC has been investigated in several mouse models harboring mutations in mitochondrial proteins [[18], [19], [20], [21], [22], [23], [24], [25], [26]], which show an early activation of the mitochondrial integrated stress response (ISR^mt) that drives an adaptive cellular response to restore mitochondrial function and mitigate cellular damage. The ISR^mt increases the expression of genes involved in protein quality control [19,27] and upregulates transsulfuration, serine and one-carbon (1C) metabolism, and antioxidant defenses [23,28]. Surprisingly, despite the overwhelming evidence linking mitochondrial stress to cardiomyopathy, very little is known about the role of mitochondrial stress in the development of PPCM.

Recently, we and others have shown that a heterozygote knock-in mouse model of the pathogenic p.S55L amino acid substitution in the mitochondrial protein coiled-coil-helix-coiled-coil-helix domain containing 10 (CHCHD10; D10^S55L mice) develops a fatal cardiomyopathy accompanied by early activation of the ISR^mt [24,[29], [30], [31]] and oxidative phosphorylation (OXPHOS) dysfunction [32]. Previous studies demonstrated that both male and naïve (nulliparous) female D10^S55L mice die from heart failure at approximately one year of age, but female D10^S55L mice that undergo pregnancy die much earlier, at approximately 3 months of age, of congestive heart failure following the trial of labor [29,33]. D10^S55L naïve females as young as 30 days of age show markers of cardiac ISR^mt and oxidative stress as a result of protein misfolding [24], with enhanced accumulation of amyloid fibrils [34], suggesting that mitochondrial proteotoxic stress occurs early in otherwise healthy D10^S55L mice. These observations raise the possibility of a pathological interplay between mitochondrial stress and the enhanced metabolic needs associated with pregnancy and labor.

During a healthy pregnancy, due to increased maternal hemodynamics [35,36], the heart undergoes structural and metabolic adaptations, characterized by reversible hypertrophy and enhanced energy expenditure [37]. As a result, there is an enhanced reliance on mitochondria to oxidize substrates and produce energy for heart contraction [38]. Pregnancy is accompanied by increased utilization of fatty acids as metabolic fuel in the heart, while glycolysis is downregulated [35,39]. Interestingly, pregnancy-induced cardiac structural and metabolic remodeling is dependent on FGF21, a myokine secreted by the stressed heart [40], suggesting that there is a physiological level of cardiac stress that participates in the adaptation to pregnancy.

In order to understand the implications of mitochondrial stress associated with a genetic form of mitochondrial PPCM, we performed transcriptomic and metabolomic analyses, as well as histological, molecular, and biochemical studies of PP D10^S55L and wildtype (D10^WT) mouse hearts. Our findings indicate that the metabolic rewiring associated with the p.S55L mutation results in depletion of key cofactors of energy metabolism in the heart. Dietary supplementation with a nicotinamide adenine dinucleotide (NAD) precursor improved PP survival, further highlighting the role of energy metabolite dysregulation. These findings suggest that, under enhanced metabolic requirements in the late stage of pregnancy and the trial of labor, metabolic failure causes premature death in PP D10^S55L mice.

2. Results

2.1. Cardiomyopathy in PP D10^S55L hearts is associated with hypertrophic cardiac remodeling

Pregnancy in D10^S55L mice has been previously associated with congestive heart failure and PP mortality [29]. To validate this phenotype in an independent cohort of mice, we housed 65-day-old D10^S55L and D10^WT females with D10^WT males and monitored maternal survival following pregnancy. In almost all cases (8/9 mice), D10^S55L mice died within the first day PP after their first delivery (Fig. 1A). Only one D10^S55L dam survived and was able to undergo a second pregnancy before succumbing. As expected, no D10^WT breeding females died after 3 pregnancies, which marked the end of the experiment.

Figure 1.

Postpartum is associated with cardiac remodeling in D10^S55L hearts.

(A) Probability of survival following sequential pregnancies in D10^WT and D10^S55L breeding females. (B) Schematic of study design for molecular and bioenergetic studies. (C) Ventriclar weight-to-tibia length (TL) ratio in naïve and postpartum (PP) D10^WT and D10^S55L mice (n = 5). (D) Hematoxylin and eosin (H&E) stain of naïve and PP D10^WT and D10^S55L hearts (scale bar = 50 μm). Black arrows indicate vacuolization. (E) Wheat germ agglutinin (WGA) stain of naïve and PP D10^WT and D10^S55L hearts (scale bar = 50 μm). (F) Super plots of cardiomyocyte cross-sectional area. Average area measurements per mouse are in color (n = 3–4), and individual measurements are in grey. (G) Masson's trichrome stain of naïve and PP D10^WT and D10^S55L hearts (scale bar = 50 μm). (H) α-SMA immunofluorescence in naïve and PP D10^WT and D10^S55L hearts (scale bar = 50 μm). (I) Super plots of α-SMA-positive area over the total area (n = 3). Average area measurements per mouse are in color (n = 3) and individual measurements are in grey. (J and K) Representative western blot of α-SMA, normalized to total protein stain and quantified in (K) (n = 5). For all graphs, data are represented as mean ± SEM. For statistical comparisons to naïve D10^WT hearts, ∗ = p adj <0.05, ∗∗ = p adj <0.005, ∗∗∗ = p adj <0.0005, ∗∗∗∗ = p adj <0.00005. For statistical comparisons to naïve D10^S55L hearts, ## = p adj <0.005, ### = p adj <0.0005, #### = p adj <0.00005.

To evaluate cardiac histopathology at the onset of PP mortality, we euthanized a separate cohort of D10^S55L and D10^WT dams on PP day 0 following their first pregnancy (average age 101 ± 10 days, n = 5). Aged-matched non-pregnant (naïve) D10^WT and D10^S55L female mice were used as controls (Fig. 1B). PP D10^S55L mice showed a significant increase in combined left and right ventricular weight normalized to tibia length (VW/TL) compared to naïve D10^WT controls (Fig. 1C). In contrast, D10^S55L naïve mice did not yet show increased VW/TL at this young age, indicating the absence of cardiac hypertrophy in the non-pregnant state. Furthermore, VW/TL was significantly increased in PP D10^S55L mice compared to naïve D10^S55L mice, indicating pregnancy-associated cardiac enlargement.

Histological analysis using hematoxylin and eosin (H&E) staining highlighted cytoarchitectural alterations in both naïve and PP D10^S55L left ventricles, including enlarged interstitial spaces and cardiomyocyte vacuolization (Fig. 1D), similar to histological alterations previously reported in one-year-old naïve D10^S55L hearts [29]. Wheat germ agglutinin (WGA) staining of left ventricles showed increased cardiomyocyte area in both naïve and PP D10^S55L hearts relative to naïve D10^WT controls (Fig. 1E,F). While PP D10^S55L hearts showed a further trend towards increased cardiomyocyte size, this measure did not reach statistical significance when averaged per mouse and visualized as a super plot (n = 3–4 mice per group; Fig. 1F).

Masson's trichrome stain and α-smooth muscle actin (α-SMA) immunofluorescence demonstrated extensive interstitial fibrosis (Fig. 1G) and myofibroblast differentiation (Fig. 1H,I) in both naïve and PP D10^S55L left ventricles. An increase in α-SMA expression was further confirmed by western blot of ventricular lysates from naïve and PP D10^S55L hearts compared to naïve D10^WT controls (Fig. 1J,K, S1A, B).

Since accumulation of misfolded and aggregated CHCHD10 participates in the pathogenesis of cardiomyopathy in D10^S55L mice [24,25,29,32], we next examined CHCHD10 protein levels and aggregation in isolated ventricular mitochondria. Western blot analysis confirmed the increase in CHCHD10 levels in D10^S55L mitochondria, with no significant difference between naïve and PP D10^S55L mice (Fig. S1C–E). Filter trap assay revealed a similar increase of NP-40-insoluble CHCHD10 in naïve and PP D10^S55L mitochondria (Fig. S1F and G). Together, these data indicate that PP D10^S55L hearts exhibit increased ventricular mass compared to naïve D10^S55L hearts, consistent with pregnancy-induced exacerbation of the cardiomyopathy. However, this enlargement is not accompanied by a worsening of cardiac histopathology, or CHCHD10 protein aggregation, suggesting that pregnancy unmasks a workload-associated vulnerability rather than accelerating structural disease progression.

2.2. PP D10^S55L hearts show metabolic profiles indicative of exacerbated mitochondrial stress

To investigate the molecular changes associated with the early death events in PP D10^S55L mice, we first performed bulk 3’ RNA sequencing on naïve D10^WT, PP D10^WT, naïve D10^S55L, and PP D10^S55L heart tissue (apexes of left and right ventricles). Principal component analysis (PCA) of the top 500 influential genes showed clear clustering by genotype, but not PP status (Fig. 2A). Similarly, hierarchical clustering analysis of the top 50 influential genes only showed clustering by genotype, with differences in the expression of genes mostly involved in energy metabolism, antioxidant response, cardiac stress, and ISR^mt (Fig. 2B). Because the four group comparisons showed genotype as the main driver of gene expression variance, we next performed a comparison limited to only naïve and PP hearts from each genotype but did not observe distinct clustering by PCA or hierarchical clustering (Fig. S2A–D). Volcano plots of gene expression in PP D10^WT compared to naïve D10^WT hearts (Fig. 2C) and PP D10^S55L compared to naïve D10^S55L hearts (Fig. 2D) showed a small number of differentially expressed genes (DEGs), further highlighting that the PP condition does not significantly alter the transcriptome in mice, regardless of genotype. None of the 13 DEGs in PP D10^WT hearts and 16 DEGs in PP D10^S55L hearts were shared between the two groups (Fig. 2E). Of note, in PP D10^WT hearts, transferrin receptor (Tfrc), mitochondrial ADP/ATP exchanger (ANT, Slc25a5 gene), and mitochondrial creatine kinase (Ckmt2) were upregulated, whereas metabolism genes Pdk4, Acot1, and Scd4 were downregulated (Fig. 2C), suggesting changes in iron and substrate metabolism in PP hearts. Instead, in PP D10^S55L hearts, genes associated with inflammation (C3, Crlf1, Cd74), heme metabolism (Alas2), fatty acid metabolism (Acot2), and proteostasis (Tecr1, Trabd2b, Serpina3n) were differentially expressed (Fig. 2D), suggesting that, even if the specific DEGs differed, some of the pathways were similar in the two PP groups.

Figure 2.

PP D10^S55L hearts have a unique metabolite profile indicative of stress.

(A) Principal component analysis (PCA) plot of the top 500 influencing genes. (B) Hierarchical clustering heatmap of the top 50 influencing genes. (C) Volcano plot of differential gene expression of PP D10^WT compared to naïve D10^WT hearts. (D) Volcano plot of differential gene expression of PP D10^S55L compared to naïve D10^S55L hearts. (E) Venn diagram of the differentially expressed genes (DEGs) comparing PP D10^WT to naïve D10^WT hearts and PP D10^S55L to naïve D10^S55L hearts. (F) PCA plot of the top 50 influencing metabolites. (G) Hierarchical clustering heatmap of the top 50 influencing metabolites. (H) Volcano plot of differential metabolite abundance of PP D10^WT to naïve D10^WT hearts. (I) Volcano plot of differential metabolite abundance of PP D10^S55L to naïve D10^S55L hearts. (J) Venn diagram of the differentially abundant metabolites (DAMs) comparing PP D10^WT to naïve D10^WT hearts and PP D10^S55L to naïve D10^S55L hearts.

Next, we analyzed the metabolome of the same region of the heart. Despite the small number of transcriptional changes associated with the PP state, metabolomics highlighted distinct metabolic profiles across the four groups. PCA and hierarchical clustering analyses of the top 50 influencing metabolites showed clustering by both genotype and PP state (Fig. 2F,G). Furthermore, volcano plots of metabolite intensity in PP D10^WT compared to naïve D10^WT hearts (Fig. 2H) and PP D10^S55L compared to naïve D10^S55L hearts (Fig. 2I) showed a larger proportion of differentially abundant metabolites (DAMs). Of the 65 DAMs in PP D10^WT hearts and 64 DAMs in PP D10^S55L hearts, 32 were shared (Fig. 2J). Pathway analysis of the shared DAMs in PP D10^WT and PP D10^S55L hearts showed an enrichment of amino acid metabolism, pyrimidine metabolism, and the pentose phosphate pathway (Fig. S2E). Amino acid metabolism pathways were also enriched in the DAMs unique to either PP D10^WT (Fig. S2F) or PP D10^S55L hearts (Fig. S2G), highlighting the involvement of these pathways in both the PP and disease states. In PP D10^WT hearts, the unique DAMs showed an enrichment of pathways associated with sugar metabolism, tricarboxylic acid (TCA) cycle, and nicotinate and nicotinamide metabolism (Fig. S2F). Instead, the unique DAMs in PP D10^S55L hearts showed an enrichment of purine and glutathione (GSH) metabolism (Fig. S2G), which was not observed in PP D10^WT hearts and is indicative of mitochondrial stress related to redox imbalance and ISR^mt [24]. Together, these data indicate that the PP state alters the metabolome of the heart, and that PP D10^S55L hearts have a distinct purine and antioxidant metabolic profile.

2.3. Heme and iron metabolism dysregulation is exacerbated in PP D10^S55L hearts

The changes in metabolites associated with purine and GSH metabolism in PP D10^S55L hearts were indicative of an antioxidant response resulting from the coordination of the ISR^mt and Nrf2 activation. Previously, we had shown that naïve D10^S55L hearts at a similar age (125d) had less reactive oxygen species (ROS) emission from mitochondria, but an increase in the transcripts of NADPH oxidases 2 and 4 (NOX2/4) [24]. NOXs contribute to redox signaling by generating ROS and coordinating normal heart function [41]. Furthermore, NOX4 is considered protective in models of cardiac hypertrophy by activating Nrf2 signaling [42]. We observed a significant increase in the expression of NOX2 complex (Nox2, Cyba, Ncf1/2/3, Rac1/2) and NOX4 genes in both naïve and PP D10^S55L hearts, but not PP D10^WT hearts, when compared to naïve D10^WT hearts (Fig. 2C, S3A). We also confirmed an increase in the levels of NOX2/4 in naïve and PP D10^S55L hearts by western blot (Fig. 3A–C, S3B, C). Interestingly, in mitochondrial fractions, naïve D10^S55L hearts had increased NOX4 levels, but both PP D10^WT and PP D10^S55L hearts had decreased mitochondrial NOX4 compared to naïve D10^WT hearts (Fig. 3D,E, S3D). Together, these data suggest an involvement of NOXs in ROS generation in D10^S55L disease and a specific regulation of NOX4 expression and localization in PP hearts.

Figure 3.

Antioxidant response is associated with heme depletion in PP D10^S55L hearts.

(A-C) Representative western blots of NOX2/4 expression in total ventriclar lysates, normalized to total protein stain and quantified in (B and C) (n = 5). (D and E) Representative western blots of NOX4 expression in mitochondrial lysates, normalized to total protein stain and quantified in (E) (n = 5). (F) Expression of antioxidant response genes in naïve and PP D10^S55L hearts compared to naïve D10^WT hearts represented as Log₂(FC) + SE (∗ = p adj <0.05) (n = 5). (G–J) Representative western blots of CAT, GPX4, and HO-1, normalized to total protein stain and quantified in (H–J) (n = 5). (K) Relative heme levels in total ventriclar lysates (n = 5). (L) Expression of heme-dependent and iron-responsive genes in PP D10^WT, naïve D10^S55L, and PP D10^S55L hearts compared to D10^WT naïve hearts represented as Log₂(FC) + SE (n = 5). For differential expression analyses compared to naïve D10^WT hearts, ∗ = p adj <0.05. For differential expression analysis compared to naïve D10^S55L hearts, red # = p adj <0.05. (M–O) Representative western blots of TFRC and FTH, normalized to total protein stain and quantified in (N and O) (n = 5). Unless stated otherwise, data are represented as mean ± SEM. For statistical comparisons to naïve D10^WT hearts, ∗ = p adj <0.05, ∗∗ = p adj <0.005, ∗∗∗ = p adj <0.0005, ∗∗∗∗ = p adj <0.00005. For statistical comparisons to naïve D10^S55L hearts, ## = p adj <0.005.

Alterations in the redox state of the heart trigger transcriptional responses aimed at mitigating ROS levels. Cardiomyocytes respond to oxidative stress by upregulating ATF4 and Nrf2 signaling [[43], [44], [45]]. In naïve and PP D10^S55L hearts, ATF4- and Nrf2-dependent genes were significantly upregulated, including catalase (Cat) and those related to GSH metabolism (Gss, Gsr, Gclc, and Gclm) (Fig. 3F). Nrf2 antioxidant response also involves the expression of genes related to heme catabolism, including heme oxygenase 1 (HO-1, Hmox1 gene) and Blvra/b, which were upregulated in naïve and PP D10^S55L hearts (Fig. 3F). We also confirmed increased expression of catalase (CAT) by western blot in naïve and PP D10^S55L hearts (Fig. 3G,H, S3E, F). Moreover, GPX4 protein levels were significantly increased (Fig. 3G,I, S3G, H), although not differentially expressed at the mRNA level, whereas HO-1 protein was unchanged in mutant hearts (Fig. 3G,J, S3E, F), despite being elevated at the mRNA level.

Ventricular heme levels were significantly decreased only in PP D10^S55L hearts (Fig. 3K), suggesting an imbalance between heme biosynthesis and degradation. In accord, we found a downregulation of Alas1, encoding the first and rate-limiting enzyme of heme biosynthesis, in both naïve and PP D10^S55L hearts (Fig. 3L). We also observed a decrease in the expression of erythroid tissue-specific isoform Alas2 in naïve D10^S55L hearts, while the expression of this isoform was not significantly altered in PP D10^S55L hearts (Fig. 3L). Alas1 is regulated by heme levels, but Alas2, which is not normally expressed in the heart, contains an iron-responsive element (IRE) and its expression is increased by cardiac stress [46]. When iron is low, Alas2 expression is suppressed, but when iron is abundant Alas2 levels increase to generate more heme. Alas2 was repressed in naïve D10^S55L hearts but elevated in PP D10^S55L hearts, suggesting that heme synthesis dysregulation might be related to the PP state. We looked at the expression of other IRE-containing genes and found that most were altered (Fig. 3L). In PP D10^WT, naïve D10^S55L, and PP D10^S55L hearts, transferrin receptor (Tfrc) was increased and ferroportin (Slc40a1) was decreased (Fig. 3L), in agreement with an increased need for iron during pregnancy [47,48]. Moreover, in PP D10^S55L hearts, Slc11a2, encoding DMT1, involved in transferrin-independent iron uptake, was increased. There was also an increase in ferritin light chain (Ftl1) in PP D10^WT, naïve D10^S55L, and PP D10^S55L hearts, although only significant in the mutant hearts. By western blot, TFRC was significantly increased in both PP groups, but not in naïve D10^S55L hearts (Fig. 3M, N, Fig. S3E and F). While ferritin heavy chain (FTH1) was increased in naïve D10^S55L hearts, we observed a significant decrease in FTH1 levels in PP D10^WT hearts (Fig. 3M, O, Fig. S3E and F). Together, this data indicates a strong antioxidant response in D10^S55L hearts, possibly due to elevated NOX expression. In PP hearts, the increase in TFRC was not matched by a parallel increase in FTH1, and at the same time, PP D10^S55L hearts specifically showed a decrease in heme, suggesting an iron dysregulation in PP D10^S55L hearts.

GSH is crucial for protecting the heart against oxidative damage, as it is the reducing agent for many antioxidant enzymes, including GPX4. GSH synthesis requires cysteine, which can be taken up or synthesized by cells via the transsulfuration pathway from serine and glycine. Levels of serine and glycine were markedly elevated in both naïve and PP D10^S55L hearts compared to naïve D10^WT hearts, but serine was significantly more elevated in PP D10^S55L hearts when compared to naïve D10^S55L hearts (Fig. 4A,B). Furthermore, we observed significantly increased GSH levels in both naïve and PP D10^S55L hearts (Fig. 4C), but oxidized GSH (GSSG) was only increased in PP D10^S55L hearts (Fig. 4D). The increase in GSSG levels suggests that PP D10^S55L hearts have exacerbated thiol redox imbalance. Therefore, we looked at protein glutathionylation by western blot. While naïve D10^S55L hearts showed a trend for increased protein glutathionylation compared to naïve D10^WT hearts, PP D10^S55L hearts had significantly less glutathionylated proteins compared to naïve D10^S55L hearts (Fig. 4E,F, S4A, B), suggesting that the GSH utilization is not directed towards protecting protein thiol groups from oxidation. We then measured protein S-nitrosylation, another form of reversible posttranslational modification of protein thiols, and found that both naïve and PP D10^S55L hearts had significantly less protein S-nitrosylation than naïve D10^WT hearts (Fig. 4G,H, S5A-C), suggesting that GSH is used for protein de-nitrosylation. Moreover, the levels of protein carbonylation, the oxidation of carbonyl groups, were similar in all groups (Fig. S6A–F), indicating that antioxidant mechanisms effectively prevent irreversible oxidative damage to proteins. Next, we investigated lipid peroxidation by immunostaining the left ventricles for malondialdehyde (MDA). While we observed an increase in MDA stain in naïve D10^S55L hearts (Fig. 4I,J), PP D10^S55L hearts did not show a significant increase relative to naïve D10^WT controls, suggesting that PP D10^S55L hearts utilize GSH to prevent lipid peroxidation. Overall, these findings indicate that oxidative modifications to proteins and lipids are unlikely to cause worsening of the cardiac phenotype in PP D10^S55L mice.

Figure 4.

Oxidative stress does not worsen in PP D10^S55L hearts.

(A-D) Normalized metabolite intensity values of serine (A), glycine (B), glutathione (GSH) (C), and oxidized glutathione (GSSG) (D) in PP D10^WT, naïve D10^S55L, and PP D10^S55L hearts compared to D10^WT naïve hearts represented as Log₂(FC) + SE (n = 5). (E and F) Representative western blot of protein glutathionylation (GSH expression), normalized to α-Actin expression and quantified in (F) (n = 5). (G and H) Representative western blot of protein s-nitrosylation, normalized to total protein stain and quantified in (H) (n = 5). (I) MDA immunofluorescence in naïve and PP D10^WT and D10^S55L hearts (scale bar = 50 μm). (H) Super plots of MDA-positive area over the total area. Average area measurements per mouse are in color (n = 3) and individual measurements are in grey. Unless stated otherwise, data are represented as mean ± SEM. For metabolite intensity statistical comparisons to naïve D10^WT hearts, ∗ = p < 0.05, ∗∗∗∗ = p < 0.00005. For metabolite intensity statistical comparisons to naïve D10^S55L hearts, # = p < 0.05, ## = p < 0.005. For all other statistical comparisons to naïve D10^WT hearts, ∗∗ = p adj <0.005, ∗∗∗∗ = p adj <0.00005. For all other statistical comparisons to naïve D10^S55L hearts, ## = p adj <0.005.

2.4. PP D10^S55L hearts do not show a worsening of mitochondrial defects

We investigated whether the increased workload associated with pregnancy and the trial of labor induced a worsening of mitochondrial structure and function in D10^S55L hearts. First, we imaged cardiomyocyte mitochondria by immunofluorescence using an antibody against the ATPB subunit of the mitochondrial ATP synthase (Complex V). Qualitatively, the mitochondrial network was disrupted in D10^S55L cardiomyocytes with fragmented and rounded organelles (Fig. 5A). These alterations were unchanged in PP D10^S55L cardiomyocytes. Next, we homogenized the right and left ventricles together and measured the total homogenate protein yield. We then extracted mitochondria from these homogenates and measured mitochondrial protein yield. For each sample, we calculated the ratio of mitochondrial to total homogenate protein yields and found similar ratios in all groups (Fig. 5B). This suggests that in PP D10^S55L ventricles, the relative mitochondrial content was unchanged. We then analyzed the levels of individual respiratory chain complex subunits in mitochondrial preparations by western blot and observed a small but significant increase in NDUFB8 (Complex I) and Mt–CO1 (Complex IV) in PP D10^S55L heart mitochondria (Fig. 5C, Fig. S7A–F). On the other hand, transcript levels of respiratory chain subunits were similarly decreased in naïve and PP D10^S55L hearts (Fig. 5D), as previously reported [24], suggesting that OXPHOS subunits are posttranscriptionally regulated. Together, these findings indicate that mitochondrial mass is not significantly altered in PP D10^S55L hearts, despite a small increase in protein levels of two OXPHOS subunits.

Figure 5.

Mitochondrial oxidative phosphorylation is preserved in PP D10^S55L hearts.

(A) ATPB immunofluorescence in naïve and PP D10^WT and D10^S55L hearts (scale bar = 20 μm). (B) Ratio between mitochondrial protein yield from the homogenate of combined right and left ventricles and the total protein yield from the same homogenate, expressed as percentage (%). (C) Representative western blot immunoblotted with OXPHOS cocktail antibody, normalized to VDAC expression and quantified in Fig. S6B–F (n = 5). (D) Gene expression of a subset of OXPHOS subunits in naïve and PP D10^S55L hearts compared to D10^WT naïve hearts represented as Log₂(FC) + SE (∗ = p adj <0.05) (n = 5). (E-G) Representative western blot of total ventriclar and mitochondrial lysates immunoblotted for cytochrome c, normalized to total protein stain or OXPHOS subunit expression, respectively, and quantified in (F and G) (n = 5). (H) Quantification of heme-containing cytochrome c represented as pmol Cyt c/mg mitochondrial protein (n = 5). (I and J) State 3 (I) and state 2 (J) respiration rates stimulated by succinate/glutamate or pyruvate/malate, represented as nmol O₂/min/mg heart mitochondrial protein (n = 5). (K) Respiratory control ratio (RCR; state 3/state 2) of heart mitochondria (n = 5). (L) ADP:O ratio of heart mitochondria (n = 5). Unless stated otherwise, data are represented as mean ± SEM. For statistical comparisons to naïve D10^WT hearts, ∗∗ = p < 0.005, ∗∗∗∗ = p < 0.00005. Suc., succinate; glu., glutamate; pyr., pyruvate; mal., malate; ADP, adenosine diphosphate; O, oxygen.

Interestingly, our immunoblotting analysis showed a two-fold increase in the levels of cytochrome c (cyt c) in PP D10^WT mitochondria compared to naïve D10^WT mitochondria (Fig. 5E,F, S7G). However, this marked increase in cyt c was not observed in PP D10^S55L heart mitochondria, which instead showed significantly less cyt c relative to naïve D10^WT mitochondria. In total heart tissue lysate, cyt c was unchanged in PP D10^WT hearts and decreased in both naïve and PP D10^S55L hearts (Fig. 5E,G, S7H, I). This was consistent with a decrease in Cycs mRNA levels only in naïve and PP D10^S55L hearts (Fig. 5D). To our knowledge, an increase of mouse heart mitochondrial cyt c in association with pregnancy has not been reported before, and it may play a role in the functional adaptation to cardiac energy requirements. Of note, recently a significant decrease in the levels of cyt c was reported in the hearts of young, non-pregnant D10^S55L mice, although this decrease was not associated with Cycs mRNA downregulation [32]. Next, we measured spectrophotometrically the content of the redox-active heme-containing cyt c holoenzyme in solubilized mitochondria, which confirmed an increase in PP D10^WT hearts and a trend towards a decrease in both naïve and PP D10^S55L hearts (Fig. 5H). A decrease of mitochondrial cyt c levels could suggest that it has leaked into the cytosol, where it could trigger apoptosis [49]. However, we did not find evidence of apoptotic cell death by TUNEL staining in either naïve or PP D10^S55L hearts (Fig. S7J), indicating that apoptosis is not a likely contributor to heart failure in PP D10^S55L mice.

To further characterize OXPHOS functionally, we measured mitochondrial respiration in freshly isolated intact heart mitochondria. Oxygen consumption was decreased in naïve and PP D10^S55L hearts in state 3 and state 2 succinate/glutamate-supported respiration, although this difference did not reach statistical significance (Fig. 5I,J). No differences were observed in pyruvate/malate-supported respiration in any of the groups (Fig. 5I,J). The respiratory control ratio (the ratio between state 3 phosphorylating and state 2 non-phosphorylating respiration, a marker of mitochondrial functional integrity) was modestly increased only in naïve D10^S55L heart mitochondria (Fig. 5K), but the ADP:O ratio was unchanged (Fig. 5L), indicating similar ATP production. Therefore, the OXPHOS function was unmodified in PP D10^S55L hearts relative to naïve D10^S55L hearts, suggesting that the small increase in complex I and IV subunits (Fig. 5C, Fig. S7A–F) did not have a functional impact. Furthermore, we measured the change in mitochondrial membrane potential with safranin O following ADP addition in mitochondria oxidizing either succinate/glutamate or pyruvate/malate. The physiological ADP-driven decline in membrane potential was increased in D10^WT PP hearts with succinate/glutamate, but not in naïve or PP D10^S55L hearts (Fig. S7K). This could indicate that PP D10^WT, but not PP D10^S55L, heart mitochondria have a higher rate of ADP phosphorylation relative to naïve D10^WT mice. Lastly, to test the resilience of mitochondria to uncoupling, we titrated mitochondrial membrane potential with the uncoupler SF-6847 and assessed the concentration needed to decrease mitochondrial membrane potential by 50% (UC₅₀). There were no significant differences in any group (Fig. S7L). Overall, these data showed that pregnancy does not worsen bioenergetic functions in D10^S55L hearts.

2.5. D10^S55L hearts show alterations of energy metabolites and energy storage molecules

The heart is metabolically flexible and can utilize different substrates to generate energy, but prioritizes fatty acid oxidation [38]. During physiological hypertrophy associated with pregnancy, there is an increase in fatty acid oxidation, whereas pathological hypertrophy increases glucose utilization and the generation of pyruvate [50]. Although the bioenergetic studies in isolated heart mitochondria showed that the OXPHOS machinery in PP D10^S55L mitochondria was not different than naïve D10^S55L mitochondria, these experiments were performed with excess exogenous respiratory substrates. Furthermore, pathway analyses of DAMs highlighted several bioenergetic pathways in both PP groups (Fig. S2E-G). Therefore, to understand the potential involvement of metabolic imbalance, we analyzed steady state levels of metabolites related to bioenergetics in cardiac ventricular lysates (schematic in Fig. 6A).

Figure 6.

Electron and energy carrier molecules are depleted in D10^S55L hearts.

(A) Schematic of mitochondrial respiration involving the TCA cycle and electron transport chain metabolites. Metabolite abundance in PP D10^S55L compared to D10^WT hearts is labeled in blue if decreased, red if increased, black if unchanged, and grey if not detected. (B and C) Normalized metabolite intensity values of glycolysis intermediates lactate (B) and pyruvate (C). (D–G) Normalized metabolite intensity values of β-oxidation intermediates l-palmitoylcarnitine (D), Acetyl-CoA (E), l-carnitine (F), and coenzyme A (CoA) (G). (H–M) Normalized metabolite intensity values of TCA cycle intermediates citrate (H), cis-aconitate (I), succinate (J), α-ketoglutarate (K), malate (L), and fumarate (M). (N–U) Normalized metabolite intensity values of electron transport chain substrates flavin adenine dinucleotide (FAD) (N), riboflavin (O), nicotinamide adenine dinucleotide (NAD) (P), NADH (Q), adenosine triphosphate (ATP) (R), phosphocreatine (S), creatine (T), and adenosine diphosphate (ADP) (U). For all graphs, data are represented as Log₂(FC) + SE. For statistical comparisons to naïve D10^WT hearts, ∗ = p < 0.05, ∗∗ = p < 0.005, ∗∗∗ = p < 0.0005, ∗∗∗∗ = p < 0.00005. For statistical comparisons to naïve D10^S55L hearts, ## = p < 0.005.

A significant increase in lactate (Fig. 6B), pyruvate (Fig. 6C), and l-palmitoylcarnitine (Fig. 6D) was observed only in PP D10^S55L hearts. However, there was a significant decrease in acetyl-coenzyme A (CoA) (Fig. 6E) and carnitine (Fig. 6F) in PP D10^WT, naïve D10^S55L, and PP D10^S55L hearts, whereas CoA was decreased only in the mutant groups (Fig. 6G). TCA cycle intermediates were also altered. Citrate (Fig. 6H) and cis-aconitate (Fig. 6I) were decreased in PP D10^WT hearts, increased in naïve D10^S55L hearts, and unchanged in PP D10^S55L hearts. Succinate was significantly decreased only in naïve D10^S55L hearts (Fig. 6J), but α-ketoglutarate (Fig. 6K), malate (Fig. 6L), and fumarate (Fig. 6M) were significantly decreased in both naïve and PP D10^S55L hearts. Both the TCA cycle and β-oxidation depend on key electron carriers. By metabolomics, we were able to detect flavin adenine dinucleotide (FAD), NAD, and NADH. The levels of FAD were significantly decreased in both naïve and PP D10^S55L hearts (Fig. 6N). Interestingly, the vitamin riboflavin, the precursor of FAD, was increased only in PP D10^S55L hearts (Fig. 6O). Moreover, both NAD (Fig. 6P) and NADH (Fig. 6Q) were decreased in PP D10^WT, naïve D10^S55L, and PP D10^S55L hearts.

Focusing on energy carriers, ATP was decreased in both naïve and PP D10^S55L hearts (Fig. 6R). Phosphocreatine, which stores high energy phosphates to regenerate ATP, was increased in naïve and PP D10^S55L hearts (Fig. 6S), accompanied by a decrease in creatine (Fig. 6T). Interestingly, there was also a decrease in ADP levels (Fig. 6U), suggesting that phosphocreatine may accumulate due to limited adenylate availability. Overall, these results indicate that the PP state induces widespread changes in cardiac energy and electron carriers, and some of these changes are unique to the D10^S55L mice.

2.6. Purine catabolism is enhanced in PP D10^S55L hearts

Due to the limited bioavailability of endogenous energy molecules for OXPHOS in naïve and PP D10^S55L hearts, including NAD and ADP, we investigated the metabolites involved in the biosynthesis and catabolism of these molecules. Of the detected metabolites involved in de novo NAD biosynthesis (Fig. S8A), L-tryptophan (Fig. S8B) and kynurenine (Fig. S8C) were increased in naïve D10^S55L hearts, and further elevated in PP D10^S55L hearts, compared to naïve D10^WT hearts. In contrast, all detected metabolites of the NAD salvage pathway were decreased in both naïve and PP D10^S55L hearts, including nicotinamide mononucleotide (NMN) (Fig. S8D), niacinamide (NAM) (Fig. S8E), and nicotinamide riboside (NR) (Fig. S8F). NADP was decreased in both naïve and PP D10^S55L hearts (Fig. S8G). Interestingly, NMN and NAM were also decreased in PP D10^WT hearts, in agreement with the reduced levels of NAD/NADH (Fig. 6P,Q).

Purine metabolism was also markedly altered in naïve and PP D10^S55L hearts (Fig. 7A), validating previous findings in naïve D10^S55L hearts [24]. Specifically, we found a significant decrease in the levels of inosine monophosphate (IMP) (Fig. 7B), adenosine monophosphate (AMP) (Fig. 7C), adenosine (Fig. 7D), and adenine (Fig. 7E), indicating that the de novo biosynthesis and recycling of adenylates limit ADP levels in these hearts. Similarly, guanosine diphosphate (GDP) (Fig. 7F) and guanosine monophosphate (GMP) (Fig. 7G) were decreased in naïve and PP D10^S55L hearts. Despite reduced levels of adenylates and guanylates, upstream metabolites of de novo purine biosynthesis, including d-ribose 5-phosphate (Fig. 7H), glycine (Fig. 4B), and AICAR (Fig. 7I), were increased in naïve and PP D10^S55L hearts. Furthermore, purine degradation products, including inosine (Fig. 7J), hypoxanthine (Fig. 7K), xanthosine (Fig. 7L), guanosine (Fig. 7M), guanine (Fig. 7N), xanthine (Fig. 7O), and uric acid (Fig. 7P), were increased in naive D10^S55L hearts, and further increased in PP D10^S55L hearts. Together, these findings suggest that electron carriers, mainly NAD, energy carriers, namely ATP, and their precursors are limiting in the disease state and fail to support energy production in D10^S55L hearts under conditions of high energy requirements, such as the trial of labor.

Figure 7.

Metabolism in D10^S55L hearts is rewired towards purine catabolism.

(A) Schematic of de novo purine biosynthesis and salvage purine synthesis metabolites. Metabolite abundance in PP D10^S55L compared to D10^WT hearts is labeled in blue if decreased, red if increased, black if unchanged, and grey if not detected. (B–P) Normalized metabolite intensity values of inosine monophosphate (IMP) (B), adenosine monophosphate (AMP) (C), adenosine (D), adenine (E), guanosine diphosphate (GDP) (F), guanosine monophosphate (GMP) (G), d-ribose 5-phosphate (H), AICAR (I), inosine (J), hypoxanthine (K), xanthonsine (L), guanosine (M), guanine (N), xanthine (O), and uric acid (P).

2.7. Dietary supplementation with EH301 improves survival of PP D10^S55L mice

To test the therapeutic protentional of NAD precursor supplementation with NR together with the natural polyphenol pterostilbene, we treated breeding D10^S55L female mice with EH301 [51], a combination of oral NR (370 mg/kg/day) and pterostilbene (60 mg/kg/day) and monitored their PP survival following sequential pregnancies (Fig. 8A). The treatment resulted in 50% of the females surviving the first pregnancy and being able to carry a second pregnancy, and in one case a third pregnancy (Fig. 8B). These data suggest that NR combined with pterostilbene, which has been shown to have cardioprotective effects through antioxidant and anti-inflammatory mechanisms [52], synergize in delaying PP D10^S55L female mortality.

Figure 8.

EH301 treatment improvesthesurvival of PP D10^S55L mice.

(A) Schematic of study design for survival studies with EH301 treatment. (B) Probability of survival following sequential pregnancies in breeding D10^S55L females treated with EH301 (purple line) or control diet (dotted black line). (C) PCA plot of the top 500 influencing genes. (D) Hierarchical clustering heatmap of top 50 influencing genes. (E) PCA plot of the top 50 influencing metabolites. (F) Hierarchical clustering heatmap of top 50 influencing metabolites. (G) Relative value of SIRT1 activity in heart lysates. Data are represented as mean ± SEM. (H) Schematics of our hypothesis: PP D10^S55L mice develop cardiac failure under increased energy demand associated with labor due to metabolic rewiring towards antioxidant pathways that limit the availability of key molecules such as energy and electron carriers. EH301 treatment tips the scale, increasing the abundance of energy and antioxidant molecules.

To evaluate the transcriptional and metabolic effects of EH301 treatment, we sacrificed a separate cohort of D10^S55L females following the first pregnancy. Heart transcriptomics did not show clear clustering of treated mice when compared to control diet-fed mice by PCA (Fig. 8C) or hierarchical clustering analysis (Fig. 8D). Only 10 DEGs were identified in EH301-treated compared to control diet-treated PP D10^S55L hearts, including the downregulation of Elf5 and Pigr, and the upregulation of Gm20629, Apoc3, Serpina1b, Slc28a3, Prl, Hamp2, Lama1, and F2. The small number of DEGs did not allow for pathway analysis, but a few upregulated genes may be involved in cardioprotection. Notably, Slc28a3 encodes a sodium-dependent nucleoside transporter, which could improve nucleotide balance. Moreover, Serpina1b encodes Alpha-1-antitrypsin 1-2 (A1AT2), which has been shown to have a cardioprotective effect in ischemia models [53]. On the other hand, EH301 treatment resulted in better defined clustering by PCA (Fig. 8E) and hierarchical clustering analysis (Fig. 8F) of heart metabolites. Of the 50 most influential metabolites resulting in hierarchical clustering by treatment, metabolites related to NAD, purine, energy, and antioxidant metabolism were more abundant in the EH301-treated group. Since pterostilbene was shown to have SIRT1 activating properties [54], we measured cardiac SIRT1 deacetylase activity, which was unchanged in EH301 diet-fed PP D10^S55L mice compared to control diet-fed PP D10^S55L mice (Fig. 8G). This finding suggests that the effects of EH301 on PP D10^S55L mice may be unrelated to SIRT1 deacetylase activity and possibly associated with other cardioprotective functions of pterostilbene [52]. Although preliminary, these data suggest that dietary supplementations that enhance energy carriers and antioxidant defenses can ameliorate the metabolic imbalance associated with cardiac failure in PP mitochondrial cardiomyopathy (Fig. 8H).

3. Discussion

During pregnancy, the heart undergoes significant morphological and metabolic adaptations to meet increased physiological demands. Hemodynamic changes include elevated blood volume, heart rate, and cardiac output [35,55]. To support these changes, the heart becomes hypertrophic and shifts its metabolism towards increased fatty acid utilization for energy production [35,39]. Under physiological conditions, cardiac structure, function, and metabolism return to baseline shortly after delivery. However, the cumulative stress of these adaptations can increase the risk of major cardiac events and, in some cases, lead to the development of PPCM [3].

Here, we investigated the molecular and metabolic consequences of pregnancy and the trial of labor in a mouse model of PPCM associated with the p.S55L mutation in the mitochondrial protein CHCHD10. D10^S55L dams survive pregnancy itself, but notably they develop acute heart failure and die within 24 h PP [29]. This suggests that the fatal event is triggered during the trial of labor and the immediate postpartum period. While there were extensive transcriptional changes associated with the p.S55L mutation, only a small number of DEGs were unique to the PP D10^S55L hearts compared to naïve D10^S55L hearts, suggesting that PP-associated heart failure in these mice occurs without a broad transcriptional reprogramming. This lack of PP-induced transcriptional alterations is consistant with our findings in D10^WT mice. This could appear surprising based on a previous study in PP mouse hearts where differential expression was revealed by PCA analysis, but these experiments were performed 7 days PP [56]. Furthermore, a different strain of mice, FVB/NJ, were used and these mice have a different gestational duration compared to C57BL/6NJ used here. Another study of hearts from C57BL/6 dams at PP day 0 identified a small number of DEGs [57], consistent with our findings.

The major differences that we identified across all groups were related to cardiac metabolites. Cardiac metabolic rewiring has previously been suggested in PP WT females [56]. Our findings in the immediate PP period confirm distinct metabolic profiles in both PP D10^WT and D10^S55L hearts, but they also highlight a clear separation between the two PP groups. Notable pathways affected in PP D10^S55L hearts include GSH metabolism, which is involved in antioxidant responses. The expression levels of key antioxidant systems were similarly elevated in both naïve and PP D10^S55L hearts. GPX4, the main antioxidant enzyme in the prevention of lipid peroxidation and ferroptosis, was strongly upregulated in both groups. Surprisingly, however, elevated MDA was observed in naïve D10^S55L hearts but not in PP D10^S55L hearts, suggesting that the mechanism that prevents lipid peroxidation are more effective in PP D10^S55L hearts. This is supported by increased GSSG, suggesting greater utilization of the reducing power of GSH, possibly by GPX4. Moreover, lipid peroxidation leading to ferroptosis is iron-dependent. The PP hearts show increased TRFC expression and decreased erritin levels, suggesting elevated iron uptake and utilization, as previously described in pregnant and PP humans [47]. However, the reduced heme levels accompanied by unchanged HO-1 indicate altered iron homeostasis in PP D10^S55L hearts. Therefore, it is possible that changes in iron homeostasis associated with pregnancy contribute to limiting MDA accumulation in the PP D10^S55L heart. On the other hand, GSH expenditure and GSSG elevation may underlie a deficiency in other GSH-dependent functions, such as protein glutathionylation, which was decreased in PP D10^S55L compared to naïve D10^S55L hearts. Enhanced GSH utilization in PP D10^S55L hearts could occur at the expense of linked metabolic pathways, including nucleotide homeostasis. Indeed, the most significant metabolic alterations unique to PP D10^S55L hearts included an increase in nucleotide degradation products. Importantly, key electron and energy carriers were decreased in D10^S55L hearts, likely as a consequence of an extensive metabolic rewiring that favors GSH biosynthesis over nucleotide maintenance [24].

Respiration capacity was unchanged in PP D10^WT heart mitochondria relative to naïve D10^WT hearts, indicating that the bioenergetic properties of the respiratory chain are not modified by pregnancy or the trial of labor. Similarly, OXPHOS was unchanged between PP and naïve D10^S55L heart mitochondria, suggesting that a worsening of respiratory chain function was not the cause of heart failure. It needs to be noted, however, that respiration in intact mitochondria is measured in the presence of excess exogenous substrates and ADP. While the respiratory chain machinery is unaffected in the PP state, it is possible that a lack of substrates results in defective energy production in PP D10^S55L hearts in vivo. Interestingly, we found that the PP condition was associated with a marked increase in mitochondrial cyt c protein levels in D10^WT hearts despite no change in cycs gene expression, suggesting a posttranscriptional mechanism of cyt c stabilization. Cyt c upregulation in pregnancy has not reported before, and its functional significance, as well the mechanisms, remain to be elucidated. Nevertheless, cyt c levels were found to be elevated in a patient group of hypertrophied hearts with congenital lesions [58], suggesting that there may be both physiological and pathological significance to changes in cyt c levels. Instead, cyt c was decreased in both naïve and PP D10^S55L hearts. Significant release from mitochondria was unlikely to be the reason for the decrease in cyt c, since we did not detect evidence of apoptosis in PP D10^S55L cardiomyocytes. On the other hand, a decrease in Cycs expression was evident and could explain the loss of cyt c. This finding was in agreement with a recent report in naïve D10^S55L hearts of both sexes [32] and could, at least in part, explain the trend for decreased respiration in D10^S55L heart mitochondria.

In humans, pregnancy increases cardiac output by 45% [59] and by 60–80% during the trial of labor due to decreased inferior vena cava compression and uterine contractions [60]. In mice, there is a 50–80% increase in cardiac output during the late stage of pregnancy and the PP period [56,61]. Although cardiac output during the trial of labor in mice has not been measured, it is likely that it would increase similarly to humans. This dramatic increase in cardiac output during the trial of labor is predicted to be energy intensive. In PP D10^S55L hearts, our metabolomic studies highlighted a severe depletion in the key substrates for ATP synthesis, purines and NAD, as a result of metabolic rewiring associated with ISR^mt. Based on these observations, we propose that when energy needs increase, D10^S55L hearts do not possess a sufficient pool of key molecules, including macroergic phosphate carriers (ADP/ATP coupling), electron carriers (NAD+/NADH coupling), and mature heme-containing cyt c. In addition, the decline in adenosine levels may play a role in worsening the cardiomyopathy, as adenosine has known cardioprotective functions [62]. Therefore, we investigated the effects of supplementing the diet with the NAD precursor NR together with pterostilbene (EH301) and showed that PP-mortality was ameliorated, although by only 1 or 2 pregnancies. This moderate improvement corresponded to a rebalancing of energy and antioxidant metabolites in PP D10^S55L hearts. Although we did not observe an increase in SIRT1 deacetylase activity in EH301-treated D10^S55L hearts, we speculate that pterostilbene benefits the PP D10^S55L heart due to its well known cardioprotective and antioxidant properties [63,64], which moderate the consumption of GSH and the need for metabolic rewiring. In a further confirmation of the therapeutic potential of metabolic rebalancing, we recently demonstrated that a high-fat diet that forces the production of acetyl-CoA through β-oxidation of fatty acids prevents the PP-associated mortality in D10^S55L mice [65].

Future longitudinal studies will be needed to narrow the critical window of metabolic failure in the D10^S55L pregnant heart, and whether it starts immediately before, during, or soon after the trial of labor. Moreover, additional targeted metabolic treatments could be tested in this mouse model of PPCM, such as nucleoside supplementation, which is currently in clinical trial for mitochondrial myopathies [66]. Overall, the preclinical findings presented here support metabolic supplementation approaches to complement current therapeutic standards for PPCM [15].

4. Methods

4.1. Animal models

All animal procedures were conducted in accordance with Weill Cornell Medicine Animal Care and Use Committee and performed according to the Guidelines for the Care and Use of Laboratory Animals of the National Institutes of Health. CRISPR/Cas9-generated CHCHD10^S55L knock-in mice were previously generated [29] (available as Stock #028952 from the Jackson Laboratory; www.jax.org) and maintained through breeding heterozygous males with WT C57BL/6NJ females (available as Stock #005304 from the Jackson Laboratory; www.jax.org). Mice were euthanized by cervical dislocation for transcriptomics, metabolomics, and bioenergetic assays. For immunohistochemistry, mice were euthanized with sodium pentobarbital (150 mg/kg, i.p.). The number of animals (biological replicates) was 5–9 per condition/genotype/experiment.

4.2. Immunohistochemistry

Mice were terminally anesthetized with sodium pentobarbital (150 mg/kg, i.p.) and perfused intracardially with phosphate-buffered saline (PBS). Hearts were post-fixed in 10% neutral-buffered formalin (Sigma) overnight at 4 °C followed by 70% ethanol before being paraffin embedded. The center of the heart was sectioned at 10 μm on a Vibratome (Leica). Following deparaffinization and rehydration, heart sections were stained with haematoxylin and eosin (H&E) or Masson's trichrome staining. For immunofluorescence experiments, heart sections were deparaffinized and rehydrated, incubated with boiled 10 mM sodium citrate (pH 6) for 35 min at 85 °C, blocked with 5% BSA, 0.1% Triton-X in PBS, and incubated overnight at 4 °C with the following primary antibodies: mouse anti-αSMA (ThermoFisher; 14-9760-82), mouse anti-MDA (Abcam; ab243066), and mouse anti-ATPB (Abcam; ab14730). Slides were washed in PBS, and appropriate secondary antibodies were incubated at room temperature for 1 h. Sections were imaged on a Leica TCS SP5 confocal laser-scanning microscope equipped with a HCX PL AP 20x objective (NA 0.70 IMM) and a 40x objective (NA 1.25 Oil), Argon 488 and HeNe 545 and 633 lasers, a 5 channel PMT for detection, and Leica LAS AF software. For each experiment, identical parameters were used for the WT and experimental sections, and fluorescence area was quantified using ImageJ (Fiji) software. Fluorescence area was measured in 3 fields per mouse heart (n = 3 mice/group).

4.3. TUNEL staining

Apoptosis was detected using One-step TUNEL In Situ Apoptosis Kit (Elabscience) according to the manufacturer's instructions. Briefly, heart sections were deparaffinized and rehydrated, followed by incubation with 1x proteinase K in PBS at 37 °C for 20 min. After washing in PBS, slides were incubated with 100 μL Labeling Working Solution (TdT equilibration buffer, labeling solution, TdT enzyme) at 37 °C for 20 min. Slides were washed in PBS and incubated with 1x DAPI in PBS at room temperature for 5 min. Sections were imaged on a Leica DMIRB inverted microscope system equipped with a 20x objective (Leica N Plan L 0.40 NA), LED illumination (CoolLED pE-300 ultra), Leica I3 filter, and monochrome CCD camera (Qimaging Qclick) using identical parameters for WT and experimental sections.

4.4. Wheat germ agglutinin labeling

Sections from the center of the heart were deparaffinized, rehydrated, and stained with wheat germ agglutinin (WGA) as previously described [67] with minor adjustments. Briefly, myocardial sections were stained with 40 μg/mL WGA (W7024; Invitrogen) in PBS for 30 min at room temperature. Immunofluorescence images of the left ventricle were captured with the Widefield immunofluorescence images of WGA using a Leica DMIRB inverted microscope system equipped with a 20x objective (Leica N Plan L 0.40 NA), LED illumination (CoolLED pE-300 ultra), Leica I3 filter, and monochrome CCD camera (Qimaging Qclick). Cross-sectional area was analyzed using ImageJ (Fiji) software. Cross-sectional area was measured for 36 cardiomyocytes per field and 6 fields per mouse were imaged (n = 216 cardiomyocytes total/mouse, n = 3–4 mice/group).

4.5. RNA sequencing

RNA was extracted from heart tissue using TRIzol (Life Technology) and the RNeasy Mini Kit (Qiagen) according to the manufacturer's protocols. 3′RNAseq libraries were prepared from 500 ng of RNA per sample using the Lexogen QuantSeq 3'mRNA-Seq Library Prep Kit FWD for Illumina and pooled for reduced run variability. Libraries were sequenced with paired-end 86 bps on an Illumina NextSeq500 sequencer (Cornell Genomics Facility). For EH301 diet studies, libraries were sequenced with single-end 86 bps. Raw single-end or paired-end reads were processed using Trimmomatic (version 0.39) to trim adapters and the BBDuk program in the BBMap package (version 39.27) to trim polyA and polyG tails. Trimmed paired-end reads were aligned to the mouse genome assembly GRCm39.p6 using the STAR aligner (version 2.7.0f). SAM files were converted to BAM, and overlapping reads per gene were counted using HTSeq-count (version 0.6.1) [68]. The R package DESeq2 (version 3.21) [69] was used to obtain both normalized and variance-stabilized counts, as well as for principal components and hierarchical clustering analyses. Pathway analysis was performed with Enrichr [[70], [71], [72]]. The cutoff for significance was an FDR-corrected p-value < 0.05.

4.6. Metabolomics

15 mg of cardiac tissue were homogenized in 80% methanol (Sigma) using Tissue Tearer (BioSpec) on dry ice and stored at −80 °C for 4 h. Homogenates were then centrifuged at 14,000×g for 20 min at 4 °C. The supernatant was stored at −80 °C for mass spectroscopy (Weill Cornell Medicine Meyer Cancer Center Proteomics & Metabolomics Core Facility). Peak intensities for metabolites were screened for missing values and analyzed by using MetaboAnalyst software (version 6.0), including pathway analysis and hierarchical clustering analysis.

4.7. Isolation of mitochondria

Coupled heart mitochondria were freshly isolated as previously described [73] with minor adjustments. Briefly, the apex of both the left and right ventricles were minced on ice, washed, and incubated in 0.01% Trypsin–EDTA (Invitrogen) in PBS at 37 °C for 10 min. Tissue was then rinsed and homogenized with 40 strokes using a 5 mL Dounce homogenizer with MS-EGTA buffer (225 mM D-mannitol, 75 mM sucrose, 20 mM Hepes (pH 7.4), 1 mM EGTA, 1 mg/mL fatty-acid-free BSA). Ventriclar homogenate was subjected to differential centrifugation to obtain fractions containing intact mitochondria, which were used for all bioenergetic and molecular studies. Mitochondrial proteins were quantified using the Bradford protein assay (Bio-Rad).

4.8. Western blotting

Total ventricular lysates and mitochondrial-enriched fractions were isolated from heart tissue by differential centrifugation. Protein concentration was determined by the Bradford protein assay (Bio-Rad). Total ventricular (25 μg) and mitochondrial (5 μg) lysates were denatured in 1X Laemmli Buffer (Bio-Rad) containing 2-Mercaptoethanol (Sigma) at 95 °C for 10 min and separated by electrophoresis in a 4–12% SDS–PAGE gel (Bio-Rad) and transferred to a PVDF membrane (Bio-Rad). Total protein was measured using No-Stain Protein Labeling Reagent Kit (Invitrogen) according to the manufacturer's instructions. Blots were incubated in 3% BSA in PBS with 1% Tween-20 (PBS-T) for 1 h at room temperature. Primary antibodies were incubated overnight at 4 °C. Secondary antibodies were incubated for 45 min at room temperature. For protein glutathionylation experiments, samples were denatured without the addition of 2-Mercaptoethanol (Sigma). Protein glutathionylation blots were incubated with mouse anti-GSH (Virogen) overnight at 4 °C, followed by secondary antibody incubation at room temperature for 1 h. Proteins were imaged using an Odyssey DLx Imaging System (LICOR), except for GSH, which was detected using Clarity Western ECL Blotting Substrates (Bio-Rad) and imaged on ChemiDoc Touch (Bio-Rad). The following primary antibodies were used at a 1:1000 dilution: mouse anti-αSMA (ThermoFisher; 14-9760-82), rabbit anti-CHCHD10 antibody (ProteinTech; 25671-AP), rabbit anti-NOX2 (ProteinTech; 19013-1-AP), rabbit anti-NOX4 (ProteinTech; 14347-1-AP), rabbit anti-GPX4 (Abcam; ab125066), rabbit anti-CAT (Abcam; ab1877), mouse anti–HO–1 (ProteinTech; 66743-1), rabbit anti-TFRC (Abcam; ab84036), rabbit anti-FTH1 (Abcam; ab75973), mouse anti-α-Actin (Abcam; ab88226), mouse anti-total OXPHOS cocktail (Abcam; ab110413), mouse anti-VDAC (NeuroMabs; 73–204), and rabbit anti-Cytochrome c (Cell Signaling; 4272S). Secondary antibodies IRDye 800CW donkey anti-mouse IgG (Licor) and IRDye 680RD donkey anti-rabbit IgG (Licor) were diluted 1:10,000.

4.9. Filter trap assay

Insoluble protein aggregates were detected by filter trap assay as previously described [74]. Briefly, 5 μg of mitochondrial fractions were solubilized with 0.5% NP-40 (Honeywell) in PBS for 15 min on ice. Samples were loaded onto a Bio-Dot Microfiltration apparatus (Bio-Rad) containing a cellulose acetate membrane (0.2 μm pore diameter, Whatman). Vacuum was applied to pass samples through the membrane, which was then washed with 1% Tween-20 in PBS. Trapped proteins were detected with rabbit anti-CHCHD10 antibody (ProteinTech; 25671-AP). Blots were then imaged as described above.

4.10. Protein S-nitrosylation assay

Protein S-nitrosylation was detected using the Pierce S-Nitrosylation Western Blot Kit (ThermoFisher) according to the manufacturer's instructions. Briefly, 5 mg of left and right ventricles were lysed in 400 μL of HENS buffer. Following centrifugation and protein quantification using Bradford protein assay (Bio-Rad), protein was diluted to 1.5 μg/μL in 100 μL HENS buffer. To each sample, 2 μL of 1M MMTS was added and the mixture was incubated for 40 min at room temperature. Protein was precipitated by adding 600 μL of pre-chilled acetone. Samples were centrifuged and acetone was decanted, allowing the pellet to dry for 10 min. The sample pellet was resuspended in 100 μL of HENS buffer and divided into two samples. One μL of Labeling Reagent was added to both samples. To one sample, 2 μL of 1M sodium ascorbate was added, and as a negative control 2 μL of ultrapure water was added to the other sample. Both samples were incubated at room temperature for 2 h, denatured in 1X Laemmli Buffer (Bio-Rad) containing 2-Mercaptoethanol (Sigma) at 95 °C for 10 min, separated by electrophoresis in a 4–12% SDS–PAGE gel (Bio-Rad), and transferred to a PVDF membrane (Bio-Rad). Total protein was measured using No-Stain Protein Labeling Reagent Kit (Invitrogen) according to the manufacturer's instructions. Blots were incubated in 3% BSA in PBS with 1% Tween-20 (PBS-T) for 1 h at room temperature. Anti-TMT antibody was incubated overnight at 4 °C. Anti-mouse IgG-HRP antibody was incubated for 1 h at room temperature. Proteins were imaged using an Odyssey DLx Imaging System (LICOR).

4.11. Protein carbonylation assay

The levels of protein carbonyl groups were assessed using the Protein Carbonyl Assay Kit (Abcam) according to the manufacturer's instructions. Briefly, 3 mg/mL solubilized heart protein samples were diluted 1:1 with 12% SDS (final concentration 6% SDS) and then incubated with either DNPH or Derivatization Control Solution for 15 min at room temperature, then neutralized. Equal amounts of sample and the DNP-BSA positive control were loaded into a 4–12% SDS–PAGE gel without denaturing agents or heat application. Samples were blocked in 5% BSA in TBST, followed by overnight incubation at 4 °C with an anti-DNP antibody. The following day, blots were washed and incubated with 1X HRP-conjugated secondary antibody at room temperature for 1 h. Blots were imaged as described above.

4.12. Measurement of heme

Total heme was measured by incubating 500 μg of total ventricular lysate in 500 μl of 2M oxalic acid at 95 °C for 30 min. Samples were centrifuged at 1,000 × g at 4 °C for 10 min to remove debris. 200 μl of the supernatant was transferred to a black 96-well plate clear-bottomed microplate (Greiner Bio-One), and fluorescence was assessed at ex 404 nm/em 630 nm using a SpectraMAX microplate reader (Molecular Devices).

4.13. Determination of cytochrome c tissue content

Cytochrome c absolute content was measured as previously described [75,76]. Briefly, ventricular mitochondria were resuspended at 2 mg protein per mL in 80 μL of PBS + Triton X-100. Fully oxidized or reduced (with excess sodium dithionite) absorbance spectra were recorded between 500 and 650 nm on a Cary Varian 4000 UV-Vis spectrophotometer (Agilent Technologies). The absolute content of cytochrome c was determined in reduced minus oxidized spectrum as described [75,76] using an extinction coefficient of ε 550-535 nm = 18,500 M⁻¹ cm⁻¹ in microcuvettes.

4.14. Measurement of oxygen consumption

Oxygen consumption was measured using 0.75 mg protein/mL of freshly isolated ventricular mitochondria at 37 °C using an Oxygraph-2k (Oroboros) as previously described [77]. Oxygen consumption was measured with specific substrates for either complex I (2 mM malate, 5 mM pyruvate) or complex II (5 mM succinate, 2 mM glutamate) in MS-EGTA buffer with 4 mM K₂HPO4 and 2 mM MgCl₂. State 3 respiration was measured following ADP addition, and state 2 respiration was measured following carboxyatractyloside addition. RCR was calculated by dividing state 3 by state 2. ADP:O ratio was calculated as previously described [75].

4.15. Determination of mitochondrial membrane potential changes

Mitochondrial membrane potential was measured using safranin O (Abcam) as previously described [78]. Briefly, 1 μM safranin O was added to the reaction for oxygen consumption measurements prior to the addition of 0.75 mg protein/mL heart mitochondria in the 2-mL chamber of the Oxygraph-2k (Oroboros) with attached Fluorescence LED2-Module and Fluorescence-Sensor Blue with the filter set for safranin O (excitation at 495 mm and emission at 587 nm). For uncoupler titrations, 20 nM SF-6847 was added gradually to reach the maximum fluorescence signal.

4.16. EH301 supplemented diet treatment

For dietary supplementation of D10^S55L female mice with EH301 containing nicotinamide riboside (NR; 370 mg/kg/day) and pterostilbene (PT; 60 mg/kg/day), provided by Elysium Health, was added to drinking water and chow, respectively. Supplementation started at the time of breeding and continued throughout pregnancy. Control diet was normal chow without pterostilbene.

4.17. SIRT1 activity measurements

SIRT1 activity was measured using the SIRT1 Fluorometric Drug Discovery Kit (Enzo Life Sciences) according to the manufacturer's instructions. Briefly, 40 μg of ventricular lysate was incubated at 37 °C for 10 min to degrade endogenous NAD. Next, 1x FLOUR DE LYS SIRT1 (deacetylase substrate) and 200 μM NAD⁺ (sirtuin substrate) in Assay Buffer were added to the samples, which were incubated at 37 °C for 1 h. 1x Developer II and 2 mM nicotinamide were added to the samples and incubated at room temperature for 45 min. Fluorescence was then measured in the wells of a clear microplate with ex 360 nm/em 460 nm using a SpectraMAX microplate reader (Molecular Devices).

4.18. Quantification and statistical analysis

All numerical data are expressed as mean ± standard error of the mean (SEM). Statistical comparisons were made in GraphPad Prism (GraphPad Software Inc.). One-way ANOVA with Bonferroni's multiple correction was used for sample comparisons. Differences were considered statistically significant for p adjusted < 0.05. Biological replicates information for each experiment is indicated in the figure legends.

4.19. Data availability

The source data for transcriptomic studies are available through Gene Expression Omnnibus gene repository (GSE325788) and metabolomics data through Metabolomics Workbench (ST004713). Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

CRediT authorship contribution statement

Nicole M. Sayles: Writing – review & editing, Writing – original draft, Resources, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Gabriella Casalena: Methodology, Investigation, Conceptualization. Dazhi Zhao: Methodology. Ryan W. Dellinger: Writing – review & editing, Formal analysis, Conceptualization. Holly E. Holmes: Formal analysis, Conceptualization. Onorina Manzo: Methodology. Alexander Galkin: Writing – review & editing, Supervision, Methodology, Investigation, Formal analysis, Conceptualization. Annarita Di Lorenzo: Writing – review & editing, Investigation, Conceptualization. Giovanni Manfredi: Writing – review & editing, Writing – original draft, Supervision, Funding acquisition, Formal analysis, Conceptualization.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Giovanni Manfredi reports equipment, drugs, or supplies was provided by Elysium Health. Ryan W. Dellinger reports a relationship with Elysium Health that includes: employment. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article.

Fig. S1.

CHCHD10 expression is elevated in D10^S55L heart mitochondria, independent of PP condition

(A) Total protein stain of membrane used to normalize α-SMA levels in Fig. 1J. (B) Western blot of α-SMA and total protein stain of the corresponding membrane. (C-E) Western blots of CHCHD10 and total protein stain, quantified in (E) (n = 5). (F-G) Filter trap assay immunoblotted (IB) for CHCHD10, quantified in (G) (n = 5). Data are represented as mean ± SEM. n = 5/condition/genotype.∗∗ = p adj <0.005, ∗∗∗∗ = p adj <0.00005.

Fig. S2.

Naïve D10^S55L and PP D10^S55L hearts are transcriptionally similar

(A) PCA plot of the top 500 influencing genes comparing only naïve and PP D10^WT hearts. (B) Hierarchical clustering heatmap of the top 50 influencing genes comparing only naïve and PP D10^WT hearts. (C) PCA plot of top 500 influencing genes comparing only naïve and PP D10^S55L hearts. (D) Hierarchical clustering heatmap of top50 influencing genes comparing only naïve and PP D10^S55L hearts. (E-G) Pathway analysis of the 32 DAMs shared between PP D10WT and D10S55L hearts (E), 33 DAMs unique to PP D10WT hearts (F), and 32 DAMs unique to PP D10S55L hearts (G).

Fig. S3.

D10^S55L hearts upregulate NOX expression and antioxidant proteins

(A) Expression of NOX genes in naïve and PP D10^S55L hearts compared to naïve D10^WT hearts represented as Log₂(FC) + SE (∗ = p adj <0.05) (n = 5). (B) Total protein stain of membrane used to normalize total ventricular lysate NOX2/4 levels in Fig. 3A. (B) Western blot of total ventricular lysate NOX2/4 expression and total protein stain of corresponding membrane. (D) Western blot of mitochondrial NOX4 and VDAC used for normalization. (E) Total protein stain of membrane used to normalize CAT and HO-1 levels in Fig. 3G and TFRC and FTH1 levels in Fig. 3M. (F) Western blot of CAT, HO-1, TFRC, and FTH1 expression and total protein stain of corresponding membrane. (G) Total protein stain of membrane used to normalize GPX4 levels in Fig. 3G. (H) Western blot of GPX4 expression and total protein stain of corresponding membrane.

Fig. S4.

D10^S55L hearts have decreased levels of protein glutathionylation

(A and B) Western blot of protein glutathionylation (GSH expression) and α-Actin expression (used for normalization).

Fig. S5.

D10^S55L hearts have decreased levels of protein nitrosylation

(A) Total protein stain of membrane used to normalize protein S-nitrosylation levels in Fig. 4G. (C and D) Western blot of protein S-nitrosylation levels and total protein stain of the corresponding membrane.

Fig. S6.

D10^S55L hearts have unchanged levels of protein carbonylation

(A-F) Western blots of protein carbonylation with or without 2,4-dinitrophenylhydrazine (DNPH; negative control), quantified in (F).

Fig. S7.

Reduced levels of mitochondrial Cyc t in D10^S55L hearts are not associated with OXPHOS impairment or apoptosis

(A) Western blot immunoblotted with OXPHOS cocktail, normalized to VDAC expression. (B–F) Quantifications of OXPHOS complex (CI-CV) subunits: NDUFB8 (B), SDHB (C), UQCRC2 (D), MtCO1 (E), ATP5A (F) (n = 5). (G) Western blot of mitochondrial cyt c levels. (H) Total protein stain used to normalize total ventricular lysate cyt c levels in Fig. 5E. (I) Western blot of total ventricular lysate cyt c levels and total protein stain of the corresponding membrane. (J) Representative TUNEL stain of naïve and PP D10^WT and D10^S55L hearts. DAPI is in blue and TUNEL is in magenta (scale bar = 50 μm). (K) Percentage (%) change in membrane potential following ADP stimulation of heart mitochondria (n = 5). (L) Half-maximal uncoupling concentration (UC₅₀) of SF-6847 in nM (n = 5). Data are represented as mean ± SEM. ∗ = p adj <0.05. Succ., succinate; glut., glutamate; pyruv., pyruvate; malat., malate; ADP, adenosine diphosphate.

Fig. S8.

NAD salvage pathway intermediates are depleted in P D10^S55L hearts

(A) Schematic of de novo nicotinamide adenine dinucleotide (NAD⁺) biosynthesis and salvage pathway metabolites. Metabolite abundance in PP D10^S55L compared to D10^WT hearts labeled in blue if decreased, red if increased, black if unchanged, and grey if not detected. (B and C) Metabolite abundance of detected de novo NAD biosynthesis intermediates l-tryptophan (B) and kynurenine (C). (D-G) Metabolite abundance of detected NAD salvage pathway intermediates nicotinamide mononucleotide (NMN) (D), niacinamide (NAM) (E), nicotinamide riboside (NR) (F), and NADP (G).

Data availability

Data will be made available on request.