Development of a Quantitative Systems Pharmacology Model to Interrogate Mitochondrial Metabolism in Heart Failure

Abstract

The metabolic hallmarks of heart failure (HF) include diminished ATP hydrolysis potential and alterations in myocardial energy substrate metabolism, such as a switch in substrate utilization away from fatty acid (FA) to carbohydrate oxidation and reduced metabolic flexibility. However, the mechanisms underlying these phenomena and their potential contributions to impaired exercise tolerance are poorly understood. We developed a comprehensive quantitative systems pharmacology (QSP) model of mitochondrial metabolism to interrogate specific pathways hypothesized to contribute to reductions in reserve cardiac power output in heart failure. The aim of this work was to understand how changes in mitochondrial function and cardiac energetics associated with heart failure may affect exercise capacity. To accomplish this task, we coupled published in silico models of oxidative phosphorylation and the tricarboxylic acid cycle with a model of β-oxidation and extended the model to incorporate an updated representation of the enzyme pyruvate dehydrogenase (PDH) to account for the role of PDH in substrate selection. We tested several hypotheses to determine how metabolic dysfunction, such as a decrease in PDH activity or altered mitochondrial volume, could lead to marked changes in energetic biomarkers, such as myocardial phosphocreatine-ATP ratio (PCr/ATP). The model predicts expected changes in fuel selection and also demonstrates PDH activity is responsible for substrate-dependent switch driven by feedback from NAD, NADH, ATP, ADP, CoASH, Acetyl-CoA and pyruvate in healthy and simulated HF conditions. Through simulations, we also found elevated malonyl-coA may contribute to lower PCr/ATP ratio during exercise conditions as observed in some HF patients.

Figure 1:

Overview of In silico models describing pathways implicated in cardiac energetics. Pathway Key: Glycolysis (green), Electron Transport Chain (Red), TCA Cycle (Yellow), β-Oxidation (light purple), Transporters (Dark Purple, Reactive Oxygen Species (ROS) (Blue), SCOT enzyme (Mint green). In silico model scope and summary (top to bottom and left to right): ▪ ODE model with 13 unconstrained parameters describing fatty acid catabolic pathway [26]. ▪ ODE model with 41 parameters and 16 adjustable describing mitochondrial respiration [21]. ▪ ODE model with 31 adjustable parameters describing oxidative phosphorylation and TCA cycle to predict the role of ADP and NADH on TCA flux [25]. ▪ ODE model describing highlighted processes phenomenologically [32]. ▪ ODE model describing the kinetics of ANT transport [31]. ▪ ODE model describing ROS [22]. ▪ ODE model describing competition of β-Oxidation pathway [20] ▪ ODE model phenomenologically describing the fate of lipids in mitochondrial respiration [26]. ▪ ODE unconstrained model phenomenologically describing mitochondrial energetics [27].

Figure 2.

In silico Mitochondrial Metabolism Model Schematic adapted from Wu et al. [25, 33] and van Eunen et al.[20] (A) The mitochondria model characterizes the interplay of the following pathways: (I) Pyruvate enters the mitochondria to generate acetyl-coA (II) β-oxidation converts acyl-carnitines to acetyl-coA and produces NADH and FADH₂ (III) TCA cycle produces NADH from Acetyl-CoA (IV) electron transport chain (ETC) provides the proton motive force for complex V (C_V). (B) Overview of PDH regulation by pyruvate dehydrogenase kinase (PDK) and pyruvate dehydrogenase phosphatase which are further influenced by the concentrations of acetyl-CoA, NADH, pyruvate, and ATP.

Figure 3.

Data Fitting and Model Validation. (A) Observed mitochondrial oxygen consumption (mVO₂) (black bars) supplied with C_n Acyl-carnitine in media without L-carnitine present compared with in silico model fitted mVO₂ (gray bars). (B) Observed mitochondrial oxygen consumption (mVO₂) (black bars) supplied with C_n Acyl-carnitine in media including the addition of L-carnitine compared with in silico model fitted mVO₂ (gray bars) (C) Fixed observed ATP concentrations over time (open circles ) and model simulation (solid line) (D) Added phosphocreatine (PCr) concentrations over time (open circles ) and model simulation of PCr addition (solid line) (E) Observed change in ADP concentration (open circles) and model predicted change in ADP (solid line) (F) Observed change in mitochondrial oxygen consumption (mVO₂) (open circles) and model predicted mitochondrial oxygen consumption (solid line) following stepwise additions of PCr.

Figure 4.

Substrate selection and energetics in healthy cardiomyocytes. Solid line represents the mean values and the shaded region represents the 90% prediction interval. (A) Carbohydrate oxidation predominates with increasing pyruvate concentrations illustrating the switch in substrate selection under the fed condition (light gray). In the fasted condition (dark gray) fatty acid metabolism predominates to supply ATP production. (B) Fraction of carbohydrate oxidation trends upward with increased free energy of ATP hydrolysis (ΔG_ATP) from rest to exercise. (C) Inorganic phosphate (pi) also trends upward with increased free energy of ATP hydrolysis (ΔG_ATP) from rest to exercise. (D) PCr/ATP ratio declines from 2.1 to 1.5 with increased free energy of ATP hydrolysis (ΔG_ATP) from rest to exercise.

Figure 5.

Contribution of pathways to PCr/ATP ratio within each case study. Error bars represent the 95% prediction intervals. Fractional PCr/ATP ratio is shown compared to the healthy control (black bar), reduced adenine nucleotide pools (AMP, ADP, ATP) (dark gray bar) downregulated β-oxidation and PDH pathways (light gray bar), reduced functional mitochondrial content (white bar). Fractional changes in PCr/ATP for each pathway are more pronounced under the exercise conditions compared to rest conditions. * p < 0.05, t-test with respect to control for each group.