Enhancing Cardiac Serine Biosynthesis Mitigates the Progression of Dilated Cardiomyopathy

Abstract

Genetic dilated cardiomyopathy (DCM) is a leading cause of heart failure. However, disease-modifying therapies remain limited. Metabolic dysfunction has emerged as a key driver of DCM pathogenesis, and impaired serine biosynthesis, catalyzed by the rate-limiting enzyme phosphoglycerate dehydrogenase (PHGDH), has recently been identified as a potential therapeutic target. Here, we evaluated the therapeutic potential of increasing serine biosynthesis through AAV9-mediated PHGDH gene augmentation in a transgenic TM54 mouse model of DCM with established pathology. Longitudinal echocardiography showed preserved systolic function and prevented ventricular dilatation in TM54 mice treated with AAV9-PHGDH compared to AAV9-GFP controls. Histological analysis revealed reduced myocardial fibrosis and cardiomyocyte hypertrophy in AAV9-PHGDH-treated TM54 hearts, indicating a reversal of pathological remodeling. Metabolic profiling, including targeted metabolomics and in vivo ¹³C-glucose tracing analysis, revealed that serine levels increased in hearts treated with AAV9-PHGDH, accompanied by decreases in glucose-derived pyruvate and lactate. At the same time, mitochondrial oxidative metabolism remained intact, indicating a shift of glycolytic carbon toward serine biosynthesis. Collectively, these findings show that enhancing cardiac serine synthesis through PHGDH gene augmentation therapy preserves contractile function and mitigates disease progression in vivo, suggesting a novel metabolic therapeutic strategy for DCM.

Introduction

Dilated cardiomyopathy (DCM) is a progressive cardiac disorder characterized by left ventricular dilation, systolic dysfunction, and metabolic remodeling, ultimately leading to heart failure. It is associated with pathogenic variants in over 50 genes of diverse ontologies. [1] Despite advances in medical management, current therapies for genetic DCM remain palliative, and patients with disease-causing mutations often experience poor outcomes, necessitating heart transplantation. [2] This underscores the urgent need for innovative, disease-modifying therapeutic strategies.

Recent studies suggest that metabolic dysfunction is a key driver of heart failure pathogenesis and a promising therapeutic target. [3] Metabolomic analysis of failing human DCM hearts reveals significant impairments in glucose utilization and a reduction in glycolytic intermediates. [4] Serine biosynthesis is one of the most depleted pathways in DCM, serving as an essential metabolic hub that connects glycolysis to nucleotide metabolism, redox homeostasis, and sphingolipid synthesis. [5] Serine depletion may exacerbate the metabolic insufficiency observed in DCM, impairing mitochondrial function, reducing membrane integrity, and disrupting antioxidant defenses. Evidence suggests that activation of the cardiac serine biosynthesis pathway, mediated by phosphoglycerate dehydrogenase (PHGDH), the rate-limiting enzyme in de novo serine synthesis (Fig. 1A), may serve as an adaptive response in stressed myocardium. [6–8] However, it remains unclear whether enhancing endogenous cardiac serine production can mitigate the progression of the disease.

Figure 1. AAV9-PHGDH gene therapy preserves cardiac function and prevents ventricular dilation in TM54 mice with dilated cardiomyopathy.

A) Schematic of the de novo serine biosynthesis pathway, which branches from glycolysis. PHGDH catalyzes the first rate-limiting step in serine biosynthesis, converting the glycolytic intermediate 3-phosphoglycerate (3-PG) to 3-phosphohydroxy-pyruvate (3-PHP).

B) Schematic of the AAV9-PHGDH, a self-complementary gene therapy vector. cTnT - chicken Troponin T promoter; hPHGDH – human PHGDH cDNA; ITR, Inverted Terminal Repeat; pA, SV40 polyA sequence.

C) Experimental timeline. At six weeks of age, TM54 mice and non-transgenic (NTG) littermates were injected with AAV9-PHGDH or the control AAV9-GFP virus. M-Mode and 4D Echocardiography were performed at baseline and serially after treatment, with endpoint analyses at ten weeks.

D) Representative M-mode echocardiography images from TM54 and NTG mice at baseline and 10 weeks after treatment. Quantitative M-mode data are provided in Supplementary Table 1.

E) Longitudinal changes in left ventricular ejection fraction (EF) at baseline and 10 weeks post-AAV delivery, assessed by 4D echocardiography. Time 0 indicates the baseline measurements at the start of treatment. Data are presented as mean ± SEM. P values, calculated by two-way ANOVA with repeated measures followed by Tukey’s multiple comparisons test, are shown for the AAV9-PHGDH vs AAV9-GFP in the TM54 mice cohorts. Box-and-whisker plots illustrate the minimum, 25th percentile, median, 75th percentile, and maximum.

F) Left ventricular EF at 10 weeks after AAV treatment was assessed by 4D echocardiography (volumetric analysis; see Supplementary Table 2). Time 0 indicates the baseline measurements at the start of treatment. Data are shown as mean ± SEM. P values were calculated using two-way ANOVA with repeated measures, followed by Tukey’s multiple comparisons test. P values are presented for the AAV9-PHGDH vs AAV9-GFP comparison in the TM54 mouse cohorts. Box-and-whisker plots depict the minimum, 25th percentile, median, 75th percentile, and maximum.

G) Longitudinal changes in left ventricular volume at systole (ESV) from baseline to 10 weeks post-AAV delivery were assessed using 4D echocardiography. Time 0 indicates the baseline measurements at the time of treatment. Data are presented as mean ± SEM. P values, calculated with two-way ANOVA with repeated measures followed by Tukey’s multiple comparisons test, are shown for the AAV9-PHGDH versus AAV9-GFP in the TM54 mouse cohorts.

H) Left ventricular end-systolic volume (ESV) was measured at 10 weeks after AAV treatment. Time 0 indicates the baseline measurements at the time of treatment. P values were calculated using baseline and 10-week post-AAV measurements with two-way ANOVA with repeated measures, followed by Tukey’s multiple comparisons tests. Box-and-whisker plots show the minimum, 25th percentile, median, 75th percentile, and maximum values.

In this study, we investigated whether cardiomyocyte-specific PHGDH augmentation confers cardioprotection in vivo. We found that enhancing cardiac serine biosynthesis preserved cardiac function and prevented adverse remodeling, establishing a mechanistic link between serine biosynthesis and heart failure progression. These findings identify PHGDH gene augmentation is a novel cardioprotective metabolic strategy for DCM.

Results

AAV-Mediated PHGDH Gene Therapy Preserves Systolic Function

The TM54 mouse is a DCM model in which a mutant α-tropomyosin E54K protein is expressed by a cardiomyocyte-selective transgene. [9] We observed that PHGDH, both mRNA and protein, was decreased in expression in the hearts of 16-week-old TM54 mice with severe systolic dysfunction, but not 6-week-old mice with a milder DCM phenotype (Supplementary Fig. 1). To test whether enhancing serine biosynthesis would attenuate the progression of cardiac dysfunction in DCM, we developed an AAV9 vector expressing a human PHGDH cDNA (AAV9-PHGDH) [10] under the control of the cardiomyocyte-specific chicken troponin T promoter (cTnT) (Fig. 1B).

The TM54 and control non-transgenic (NTG) littermate mice were treated at 6 weeks of age with either AAV9-PHGDH or a control AAV9-GFP vector (Fig. 1C), when systolic dysfunction, but not PHGDH deficiency, was already apparent. Cardiac function was evaluated before and 2-, 6-, and 10-weeks post-treatment by serial M-mode and 4D echocardiography (Fig. 1D-H; Supplementary Tables 1 & 2). The AAV9-PHGDH and AAV9-GFP cohorts had similar levels of LV dilation and systolic dysfunction before treatment. This included ~11% reduced fractional shortening by M-mode and ~14% reduced ejection fraction by 4D imaging between TM54 and control non-transgenic (NTG) littermate control cohorts.

In AAV9-GFP-treated mice, EF progressively declined, mirroring the expected trajectory of disease progression. In contrast, AAV9-PHGDH treatment stabilized systolic function, resulting in a 15% improvement in EF compared to AAV9-GFP at the endpoint (4D EF %, 35.05 ± 1.65% vs. 19.91 ± 1.47%, P<0.0001) (Fig. 1E-F; Supplementary Table 2). Additionally, AAV9-PHGDH treatment significantly reduced LV end-systolic volume (ESV) and end-diastolic volume (EDV) compared to AAV9-GFP controls, indicating attenuation of ventricular dilation (ESV: 40.60 ± 2.93 μL vs. 56.25 ± 3.40 μL, P < 0.0001; EDV: 62.07 ± 3.23 μL vs. 72.59 ± 3.15 μL, P = 0.03; Supplementary Tables 1 and 2). This suggests that PHGDH therapy prevents ventricular dilation, a hallmark of DCM progression. Notably, AAV9-PHGDH had no discernible impact on cardiac function in NTG animals, confirming that its beneficial effects were specific to the DCM hearts (Fig. 1G-H; Supplementary Tables 1 and 2).

RNAscope and RT-qPCR confirmed robust, cardiomyocyte-restricted transgene expression at 10 weeks post-delivery, while Western blot analysis showed about a 20% increase in PHGDH protein levels in AAV9-PHGDH-treated hearts compared to controls (Supplementary Fig. 2).

AAV9-PHGDH Reduces Fibrosis and Reverses Cardiomyocyte Hypertrophy

DCM is associated with progressive interstitial myocardial fibrosis and myocyte hypertrophy. Picrosirius Red staining under polarized light revealed a significant reduction in interstitial fibrosis in the hearts of 16-week-old mice in AAV9–PHGDH–treated TM54 mice compared to AAV9–GFP–treated controls (Fig. 2A and Supplementary Fig. 3). Moreover, cardiomyocyte hypertrophy was reversed in AAV9-PHGDH-treated TM54 mice, along with a trend towards normalizing heart weight, indicating that PHGDH overexpression reduces pathological remodeling in DCM (Fig. 2B; Supplementary Figure 4). Finally, histopathological assessment by H&E staining revealed no evidence of transgene-associated inflammation, indicating that PHGDH overexpression does not elicit an inflammatory response in the heart (Supplementary Fig. 5)

Figure 2. PHGDH gene therapy reduces fibrosis, preserves cardiomyocyte size, and redirects glucose carbons into serine biosynthesis in TM54 hearts.

A. Representative images of Picrosirius red-stained left ventricular tissue sections were analyzed for collagen content using a circularly polarized light microscope at the endpoint. Box-and-whisker plots depict the minimum, 25th percentile, median, 75th percentile, and maximum values. Scale bar: 150 μm. P values were computed using two-way ANOVA followed by Tukey’s multiple comparison tests.

B. Representative images of Wheat germ agglutinin-stained left ventricular tissue sections and quantification of cardiomyocyte cross-sectional area at the endpoint are provided. The cell size at 6 weeks of age is shown for comparison. Box-and-whisker plots display the minimum, 25th percentile, median, 75th percentile, and maximum values. Scale bar: 50 μm. P values were calculated using two-way ANOVA followed by Tukey multiple comparison tests.

C. Serine levels in hearts of TM54 mice treated with AAV9-PHGDH or AAV9-GFP for 10 weeks, measured by targeted metabolomics. Data are presented as mean ± SEM (n = 4). P values are from a Student’s t-test.

D. Schematic of [U-¹³C₆]glucose carbon fate in the first turn of the TCA cycle. ¹³C-Isotopologues are designated as M+n, where n represents the number of ¹³C atoms incorporated from [¹³C₆]-glucose into the metabolite (e.g., M0 = unlabeled, M+1 = one ¹³C atom incorporated, M+2 = two ¹³C atoms incorporated, etc.).

E. Relative abundance of pyruvate ¹³C-isotopologues from the in vivo [¹³C₆]-glucose tracing study. TM54 mice were treated with AAV9-PHGDH or AAV9-GFP for 10 weeks and gavaged with 150 mg of [¹³C₆]-glucose for 1 hour before harvesting the hearts. Data are presented as mean ± SEM (n = 4). P values are from two-way ANOVA with Sidak’s multiple comparisons test.

F-G) Fractional distribution of ¹³C-isotopologues of the TCA intermediates citrate (F) and α-kG (G) ¹³C isotopologues from [¹³C₆]-glucose in TM54 hearts following AAV9-GFP or AAV9-PHGDH treatment. Data are presented as mean ± SEM (n = 4). P values are from two-way ANOVA with Sidak’s multiple comparisons test. α-kG, α-ketoglutarate.

H) Schematic summary of findings: PHGDH overexpression redirects glucose carbons from glycolysis to the serine biosynthesis pathway, providing cardioprotection in DCM.

AAV9-PHGDH Alters Metabolic Flux and Enhances Cardiac Serine Levels

To investigate the metabolic changes underlying the cardioprotective effects of AAV9-PHGDH, ventricular tissue from TM54 mice was analyzed using targeted metabolomics hydrophilic interaction liquid chromatography-mass spectrometry (HILIC-MS) at endpoint. Mice treated with AAV9-PHGDH exhibited a significant increase in cardiac serine levels compared to AAV9-GFP controls, confirming that the PHGDH gene therapy increased serine abundance in the DCM heart (Fig. 2C). Targeted metabolomics (GS-MS) of a broad panel of amino acids showed no significant differences between the groups except for an increase in glutamine levels (Supplementary Fig. 6)

To further assess changes in cardiac glucose metabolism, we performed in vivo ¹³C-glucose isotope tracing at physiological levels, coupled with GC–MS analysis to comprehensively map glucose carbon flux (Fig. 2D). The PHGDH-treated hearts showed a significant reduction in glucose conversion to pyruvate compared to controls, as indicated by a decrease in M+3 pyruvate (Fig. 2E). In line with this reduced glycolytic activity, ¹³C-lactate (M+3) levels were also lower in PHGDH-treated hearts compared to controls (Supplementary Fig. 7A). Although glucose-derived serine was undetectable, consistent with the heart’s low baseline serine synthesis [11], fractional labeling of TCA cycle intermediates revealed strong entry (citrate M+2) and propagation (αKG M+2, Malate M+2 and fumarate M+2) of glucose-derived carbon with no significant differences between groups (Fig. 2F-G; 2Supplementary Fig. 7B-C).

In parallel, expression analysis of a comprehensive metabolic gene panel showed that the expression of genes involved in serine biosynthesis, one-carbon metabolism, the TCA cycle, fatty acid oxidation, and glutathione metabolism remained unchanged (Supplementary Fig. 8).

Together, these metabolomic and transcriptomic data suggest that restoring PHGDH gene augmentation reprograms cardiac metabolism away from glycolysis while preserving oxidative capacity (Fig. 2F).

Discussion

Despite advances in managing symptoms, DCM remains an unmet clinical need due to the lack of disease-modifying therapies. This study demonstrates that enhancing de novo serine biosynthesis through AAV9-mediated PHGDH gene augmentation preserves systolic function and attenuates pathological remodeling in a preclinical model of genetic DCM. Our findings provide proof of concept that rewiring metabolic pathways in the failing heart can confer significant cardioprotection.

Our metabolic analyses provide mechanistic insight into this protective effect. PHGDH gene augmentation reduced glycolytic flux to pyruvate and lactate, increased myocardial serine abundance, while preserving mitochondrial oxidative metabolism. Notably, the expression of genes involved in central metabolic pathways remained unchanged, suggesting that this metabolic shift was predominantly regulated post-transcriptionally. These findings support a capacity-building model in which PHGDH redirects a fraction of glycolytic carbon towards de novo serine biosynthesis, thereby replenishing intracellular serine pools to bolster the anabolic pathways of the failing heart.

Prior studies suggest that increased serine availability fuels one-carbon metabolism, which in turn provides essential nucleotide precursors for biosynthesis and NADPH, thereby maintaining the redox homeostasis of cardiomyocytes [5], potentially explaining the observed cardioprotection.

Although we did not directly measure one-carbon flux, our findings are consistent with the biosynthesis capacity-building model of cardioprotection, which is supported by previous experimental and clinical studies linking enhanced de novo serine biosynthesis and one-carbon metabolism to better cardiac outcomes. For example, in patients with advanced heart failure who responded to LVAD therapy, increased PHGDH expression was observed, along with a higher one-carbon flux [8]. A similar metabolic rewiring has been observed in experimental models of heart failure, where activation of the serine and one-carbon pathway has been shown to protect the heart. Blocking this pathway, either genetically or pharmacologically, negates these benefits [6,7]. Supporting this, reduced serine synthesis was previously linked to contractile and mitochondrial dysfunction in human iPSC-derived cardiomyocytes, both of which were restored by increasing serine biosynthesis [12]. Like older TM54 mice with severe dysfunction, metabolomic analysis of end-stage human hearts identified serine as one of the most significantly depleted metabolites [4], indicating that serine deficiency is a metabolic vulnerability and that correcting it could be beneficial. Overall, these studies suggest that impaired serine supply and reduced one-carbon metabolism are hallmarks of the failing heart, and that maintaining this pathway is sufficient to slow disease progression.

Collectively, our findings suggest that serine biosynthesis is a modifiable metabolic vulnerability in DCM. Specifically, PHGDH-driven serine metabolism redirects glucose metabolism to cardioprotective anabolic processes, providing a novel metabolic therapeutic approach for heart failure.

Limitations:

First, although we observed consistent improvements in function, remodeling, and metabolism, the precise mechanistic connection between serine biosynthesis and cardioprotection needs to be further defined. Second, our analysis was limited to one genetic model of DCM; expanding this research to other animal models, including large mammals, is necessary to confirm translational relevance. Lastly, although PHGDH overexpression was achieved through AAV-mediated gene augmentation, long-term safety, dose-response, and tissue-specific effects still need to be evaluated.

Methods

Animal Studies

Male and female FVB/N-Tg(Myh6-Tpm1∗E54K) and NTG mice were obtained from Jax (Strain #:035610). The mice were randomized and received either AAV9-PHGDH or AAV9-GFP via a single tail vein injection of 1.5 × 10¹³ viral genomes/kg at 6 weeks of age. Research protocols were approved by the Administrative Panel on Laboratory Animal Care and the Institutional Animal Care and Use Committee at Stanford University.

Adeno-Associated Virus

The human PHGDH CDS (NM_006623) was cloned into a self‐complementary AAV genome vector downstream of the chicken cardiac troponin-T promoter and pseudotyped into rAAV9 capsids (Penn Core). The green fluorescent protein (GFP)-expressing vector (AAV9-GFP) was used as a control.

Echocardiography

Mice were anesthetized with 1%–3% isoflurane for echocardiographic assessments using a Vevo 3100 system (FujiFilm), while maintaining heart rates between 450 and 530 bpm. Serial imaging was performed at baseline (6 weeks of age) and at 2-, 6-, 8-, and 10-weeks post-treatment. Image acquisition included M-mode, B-mode, and 4D echocardiography, with datasets captured at 0.2 mm increments across the left ventricle. A complete cardiac cycle was recorded with automated ECG and respiratory gating. Left ventricular dimensions were obtained from M-mode, whereas absolute volumes and ejection fraction (EF) were derived from 4D echocardiography. Data were analyzed using the VevoLAB software (Fujifilm). The operator was blinded to the treatment groups.

Histochemistry

The mouse heart was arrested with 1M KCl, fixed in 4% formalin, and embedded in paraffin. Deparaffinized sections (5 μm) were stained with Picrosirius Red, Hematoxylin and Eosin, and Alexa Fluor-594 WGA. The cross-sectional area of over 100–200 myocytes was measured using WGA-stained sections. Myocardial interstitial fibrosis, excluding perivascular regions, was assessed in Picrosirius Red-stained sections under polarized light microscopy. The cross-sectional area and collagen content were analyzed with ImageJ.

Heart Metabolomic analysis

Metabolites were extracted from 20 mg of heart tissue homogenized in 1 mL methanol:water (1:1) with internal standards (100 nmol Norvaline, 20 μmol PIPES) using a ball mill. Then, 1 mL chloroform and 2 μL formic acid were added and vortexed for 5 minutes, followed by centrifugation at 21,000 g (4°C) for phase separation. The polar phase was collected, dried overnight in a vacuum centrifuge, and resuspended in 100 μL of 75% methanol for HILIC-HRMS or kept dry for online derivatization before GC-MS analysis.

Polar metabolites like serine were quantified using HILIC-HRMS with a Vanquish Flex UHPLC on an InfinityLab Poroshell 120 HILIC-z column (2.1 × 100mm, 2.7 μm, Agilent). Mobile phases included ammonium carbonate buffers in water (A) and 95:5 acetonitrile: water (B) with 5μM medronic acid. The flow rate was 0.3 mL/min with a gradient: starting at 100% B, decreasing to 90% B over 4 min, to 50% B over 6 min, then to 30% B in 0.5 min, held for 1 min before returning to 100% B and equilibrating at 45°C for 8 min. Data acquisition was via a Q-Exactive orbitrap mass spectrometer, with HESI conditions and MS parameters as described. [13] Data analysis used TraceFinder and a 10-point standard curve for absolute concentrations.

After online derivatization, labeled metabolites (e.g., ¹³C-pyruvate, ¹³C-lactate) were analyzed via GC-MS (DB-35MS column, 30 m × 0.25 mm × 0.25 μm, Agilent J&W). Data were processed, corrected for natural isotope abundance, and normalized to Norvaline and protein content using AMDIS_32 and MATLAB scripts.[13]

Highlights.

1. Evidence from in vivo studies shows for the first time that enhancing PHGDH-mediated serine biosynthesis preserves cardiac function in DCM.

   • Prior studies have identified serine depletion in human DCM hearts and iPSC-CMs. However, this study demonstrates that restoring serine biosynthesis in vivo via PHGDH gene therapy attenuates disease progression and prevents adverse remodeling.
2. PHGDH gene therapy redirects metabolic carbon toward serine biosynthesis, highlighting a novel metabolic intervention for heart failure.

   • Metabolomic analysis confirms that the PHGDH gene therapy redirects glycolytic intermediates toward serine production, supporting the notion that metabolic reprogramming can be utilized as a therapeutic strategy for DCM.
3. Bridges the gap between metabolic dysfunction and targeted intervention, positioning serine biosynthesis as a promising therapeutic target.

   • This study goes beyond correlative evidence from human and in vitro studies, providing in vivo data that support PHGDH and serine metabolism as a novel metabolic target for future translational applications in heart failure therapy.