Mitochondrial β-oxidation of fatty acids plays a major role in energy production, especially during periods of fasting or low intensity exercise. The primary sources of fatty acids for oxidation are dietary and mobilization of triacylglycerols mainly stored in adipocytes of adipose tissue. The release of metabolic energy, in the form of fatty acids, is controlled by a complex series of interrelated cascades that result in the activation of hormone-sensitive lipase, which hydrolyzes fatty acids from triacylglycerols and diacylglycerols. The final fatty acid is released from monoacylglycerols through the action of monoacylglycerol lipase, an enzyme active in the absence of hormonal stimulation. Once released, these fatty acids travel through the blood to other tissues such as muscle where they are oxidized to provide energy through the mitochondrial β-oxidation pathway. The β-oxidation spiral of fatty acid metabolism involves these four steps: oxidation, hydration, a second oxidation, and finally thiolysis. These occur in repeating cycles through the sequential removal of 2 carbons and production of acetyl-CoA, which then enters the Krebs cycle for oxidation and ATP production (Figure 1). Another destination of acetyl-CoA is the production of ketone bodies in the liver that are transported to tissues like the heart and brain for energy production during starvation. Fatty acids with an odd number of carbons in the acyl chain are left at the end with propionyl-CoA, which is converted to succinyl-CoA that then also enters the Krebs cycle. Furthermore, unsaturated fatty acids with bonds in the cis configuration require three separate enzymatic steps to prepare themselves for the β-oxidation pathway.

Mitochondrial fatty acid oxidation deficiencies are due to genetic defects in enzymes of fatty acid β-oxidation and transport proteins (clinically often summarized as FATMO – fatty acid transport and mitochondrial oxidation). Genetic defects have been identified in most of the genes where nearly all types of sequence variations (mutation types) have been associated with disease [1]. In particular, defects in fatty acid metabolism associated with clinical disorders include defects in acyl-CoA dehydrogenase and β-hydroxyacyl-CoA dehydrogenase, which catalyzes the first and third steps in β-oxidation, respectively. Several acyl-CoA dehydrogenases were previously isolated and described [2–6]. In general, these enzymes can be classified due to their fatty acid chain length specificity in short-chain (SCAD), medium-chain (MCAD), long-chain (LCAD), and very long-chain acyl-CoA dehydrogenases (VLCAD). The enzymatic activity distributions of the fatty acid chain length specificity of these four enzymes overlap to some extent [2–4, 7], as summarized in Figure 2A. Consequently, the deficiency of one particular enzyme cannot be compensated by the accumulated activity of the other non-impaired enzymes (Figure 2B).

Ten of the disorders directly affecting mitochondrial fatty acid oxidation have been well-defined, implicating episodic hypoketotic hypoglycemia provoked by catabolic stress, multiple organ failure, muscle weakness, or hypertrophic cardiomyopathy. Additionally, syndromes of severe maternal illness (HELLP syndrome and AFLP) have been associated with pregnancies carrying a fetus affected by fatty acid oxidation deficiencies. The incidence of one of these disorders, MCAD deficiency (MIM 201450), is 1:14 600 in almost 8.2 million newborns worldwide [8]. In the first years of life this inherited deficiency may become apparent following a prolonged fasting period, sometimes in combination with infection or fever. An acute attack usually features symptoms of lethargy, nausea and vomiting, which rapidly progresses to coma within 1–2 h. Up to 25% of MCAD patients die during their first attack; or suffer permanent brain damage from cerebral edema. The clinical phenotypes of most of the disorders of fatty acid metabolism are very similar [9].

The introduction of tandem mass spectrometry (MS/MS) for the analysis of plasma acylcarnitines has greatly facilitated the identification of patients with a defect in fatty acid β-oxidation and has unquestionably been the most striking recent advance in newborn screening. Pre-symptomatic diagnosis is important to prevent morbidity as most of the diagnosed defects are treatable and the prognosis is generally favorable [10–12].

Besides statistical model building and data mining based approaches [13–15], computational Systems Biology is essential to combine knowledge of human physiology and pathology starting from genomics, molecular biology, and the environment through the levels of cells, tissues, and organs all the way up to integrated systems behavior. Applying Systems Biology approaches within the context of human health and disease will definitely gain new insights. Eventually, a new discipline – Systems Medicine – will emerge at the interface between Medicine and Systems Biology [16–18]. Higher levels of organization are extremely complex and even models at the cell and subcellular levels are forced to resort to simplifications to minimize modeling and computational complexity [19–21]. Additionally, some parameters and constants for kinetics, binding and concentrations of biomolecules are typically not known, thus reducing the model's ability to respond correctly to dynamic changes in external conditions. A high-quality network of human-specific metabolic pathways including detailed knowledge about all metabolic reactions concerned is essential to design tailored kinetic models for better understanding of human metabolism and its relationship with diseases. While such large networks are used to analyze the global structure or functional connectivity of the network [22], deterministic and stochastic models are mainly used for simulating specific metabolic pathways as well as regulatory and signaling networks [23].

To date, little is known about fatty acids kinetics, especially during catabolic stress or exercise when the demand for fatty acid oxidation is increased. Here we introduce a detailed kinetic model of mitochondrial metabolism with specific focus on fatty acid β-oxidation to simulate and predict the dynamic response of that metabolic network toward distinct enzyme deficiencies. The simulation results are compared and validated using experimental data. Finally, the dynamic response to changes in the input to the system representing catabolic stress is simulated and results are interpreted in a biological and clinical context, followed by a discussion on limitations of the model.

The present computer simulation attempts to contribute to a better understanding (explanation) of pathophysiological aspects of a group of hereditary disorders impairing mitochondrial β-oxidation. Several enzymes are involved in mitochondrial fatty acid oxidation. For all of them, genetic defects impairing their function have been described. The availability of modern analytical methods has facilitated newborn screening for these disorders. For the identification of patients with a defect in β-oxidation, acylcarnitines in blood were evaluated showing a characteristic profile depending on the affected enzyme even in asymptomatic stage. We simulated the steady-state concentrations of acyl-CoAs in acyl-CoA dehydrogenase deficiencies and compared the results to acylcarnitine data from screening programs in Middle Europe and South Australia. Results indicate that the overall characteristics of the simulated accumulation of acyl-CoA show good agreement with experimental data and findings in the literature (Figure 3 and Figure 4).

Differences in simulated C16 acyl-CoA and measured C16 acylcarnitine might be caused by the special biological role of C16 acylcarnitine. From a biochemical perspective it is not clearly evident why a decrease of acyl-CoA dehydrogenase activity (Figure 2B) would not impact the concentration of C16 acyl-CoA in the mitochondria, leading to a much higher C16 acyl-CoA than C14 acyl-CoA concentration, which is reflected by the measured acylcarnitines (Table 1). It seems that there is an abundant pool of C16 acylcarnitine in the blood, whose level is almost not affected by mitochondrial fatty acid oxidation deficiencies (as can be seen in Table 1). This might be reasonable, since C16 acylcarnitine is an ester of an important saturated fatty acid, which is involved in several biological processes within the body – it is needed for energy, hormone production, cellular membranes and for organ padding as well as for important signaling and stabilization processes in the body.

Differences in simulated acyl-CoA ratios and measured acylcarnitine ratios shown in Table 2 are mainly given by the relatively small value of the simulated C4 acyl-CoA concentration and the relatively large value of the C12 acyl-CoA concentration. Higher values for C4 and lower values for C12 can be mainly generated by changing the activity of the enzymes for chain length 4 and 12. The problem is that we do not exactly know the total enzyme activity because the enzyme activities are based on data from rat liver mitochondria [3–5] (and not on data from human mitochondria). Additionally, mitochondria from different tissues can show different total acyl-CoA dehydrogenase activity as well as possible residual activities of the deficient enzyme are also changing the total acyl-CoA dehydrogenase activity of the individual patient. This might be supported by the high inter-patient variance of the measured acylcarnitines, which is caused not only by the experimental error but also by the biological variance between patients. For example, for some patient data the C12 acylcarnitine concentration is higher than the C14 acylcarnitine concentration (data are not shown). The discrepancy with C14 and C12 (for the simulation data the C12 ratio is higher than the C14 ratio which was not found in the experimental data) can be considered negligible, because of the small difference between C14 and C12 compared to the strong increase of C8 for MCADD. The ratio C8/C14 is more than 10 fold higher than the ratio C12/C14.

The clinical phenotypes of most of the disorders are very similar. As MCAD deficiency is the most prevalent defect among them we focused our discussion and biochemical interpretation on this particular defect. Nevertheless, the aspects are relevant to the pathogenesis of all fatty acid oxidation defects. Patients with MCAD deficiency are without clinical manifestations until a prolonged fasting period sometimes in combination with infection or fever. As a physiological response to this catabolic stress triacylglycerols from adipose tissue are released and energy production switches from carbohydrate to lipid utilization. In healthy individuals, subsequently ketone body formation by the liver is increased to provide this metabolic fuel for brain and muscles.

The response to catabolic stress in patients with MCAD deficiency shows a marked increase in plasma fatty acids, mitochondrial acylcarnitines and acyl-CoAs. Severe symptoms of lethargy and nausea develop as a consequence of encephalopathy, and patients can become dangerously ill, sometimes before plasma glucose falls to hypoglycemic levels. The progression to severe sickness proceeds within a few hours. Patients often die in the course of the first episode or at least suffer from persistent brain damage. The underlying pathogenetic mechanisms have been poorly understood until now. To simulate the response to fasting we assumed a 20% increase of palmitoyl CoA (C16) and observed significant differences of the calculated metabolic changes in acyl-CoA deficiencies with respect to the healthy controls (Figure 5). One major consequence of the disorder is inadequate ketone body formation to meet tissue energy demands under conditions of fasting and catabolic stress. Our calculations showed that formation of acetyl-CoA – substrate for energy production via the tricarboxylic acid (TCA) cycle and ketogenesis – is impaired (Figure 7). The simulation impressively showed the low production rate of acetyl-CoA within the first hours, which corresponds to the rapid disease progression after onset. In addition, inadequate acetyl-CoA production has secondary effects on flux through the TCA cycle, on regulation of fatty acid oxidation, and on efficiency of gluconeogenesis, which contribute to pathogenesis [24].

The simulated accumulation of specific acyl-CoAs according to the investigated enzyme deficiencies are in agreement with the accumulation of plasma free fatty acid intermediates, which enter the central nervous system and exert toxic effects, which may explain the observed encephalopathy and cerebral edema. In vitro experiments on cerebral cortex of rats indicate that inhibition of energy metabolism and oxidative stress induction by the accumulating fatty acids may contribute to the pathophysiology of encephalopathy [30, 31].

Although several patients have been found to have VLCAD deficiency, none have been documented with LCAD deficiency [32]. This could arise from either gestational loss due to LCAD deficiency as seen in the mouse model, a failure to recognize LCAD deficiency because the phenotype differs so greatly from other inborn errors of fatty acid metabolism, or absence of disease resulting from LCAD deficiency in humans [33, 34]. The dynamic behavior of the simulation model of LCAD deficiency exhibits the highest accumulation of fatty acids (179-fold of C14 as can be seen in Figure 4) along with marked increase of these substrates during fasting (Figure 5 and Table 4) and the lowest production rate of acetyl-CoA (Figure 7). These findings might confirm gestational loss to be the explanation that no human cases of LCAD deficiency have been described.

Our model can be extended to comprehensively test and study deficiencies of mitochondrial trifunctional protein and β-hydroxy-acyl-CoA dehydrogenase, or other diseases of fatty acid oxidation such as carnitine cycle, electron transfer and ketone synthesis defects. Furthermore, differences in the expression level of the enzymes in different cells and tissues and their consequences on the dynamical behavior of the β-oxidation can be investigated. Future work will incorporate the enzymatic steps for unsaturated fatty acids.

In summary, this work provides a stimulating example for Systems Biology in the context of human disease revealing insights into dynamic properties of complex biochemical networks under the constraints of various disease conditions. As analytical technologies for global and targeted measurements mature, especially with regards to metabolites, new findings and hypothesis can be verified utilizing quantitative data. Furthermore, while mitochondrial deficiencies are often treated with metabolites to stimulate the enzyme activities, models will allow evaluation of the influences of metabolite treatments at the mitochondrial level, visualization of the dynamic behavior of the pathway and exploration of a hypothetical rationale of the treatment. In this respect, computational biology proves to be able to uncover insights, which can hardly be obtained from experimental data alone.