Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 7 (6), 1891-901

Modeling Cholesterol Metabolism by Gene Expression Profiling in the Hippocampus

Affiliations

Modeling Cholesterol Metabolism by Gene Expression Profiling in the Hippocampus

Christopher M Valdez et al. Mol Biosyst.

Abstract

An important part of the challenge of building models of biochemical reactions is determining reaction rate constants that transform substrates into products. We present a method to derive enzymatic kinetic values from mRNA expression levels for modeling biological networks without requiring further tuning. The core metabolic reactions of cholesterol in the brain, particularly in the hippocampus, were simulated. To build the model the baseline mRNA expression levels of genes involved in cholesterol metabolism were obtained from the Allen Mouse Brain Atlas. The model is capable of replicating the trends of relative cholesterol levels in Alzheimer's and Huntington's diseases; and reliably simulated SLOS, desmosterolosis, and Dhcr14/Lbr knockout studies. A sensitivity analysis correctly uncovers the Hmgcr, Idi2 and Fdft1 sites that regulate cholesterol homeostasis. Overall, our model and methodology can be used to pinpoint key reactions, which, upon manipulation, may predict altered cholesterol levels and reveal insights into potential drug therapy targets under diseased conditions.

Figures

Fig. 1
Fig. 1
Schematic diagram of the cholesterol synthesis and degradation pathways. Metabolites M1 to M52 are described in Table S1 (ESI†). The enzyme encoding genes are indicated in each reaction. The grey arrows indicate enzymatic reactions carried out by Dhcr24.
Fig. 2
Fig. 2
Adult hippocampus cholesterol expression profile. The expression profile of the adult mouse hippocampus plots the 24 genes involved in cholesterol metabolism. The gene expression intensity values were obtained from in situ hybridization data collected from the Allen Mouse Brain Atlas
Fig. 3
Fig. 3
Final concentration of metabolites calculated from the cholesterol pathway model described in Fig. 1. The value of the metabolite indicates the name of the metabolite found in Table S1 (ESI†). Simulation time was 1 × 106 arbitrary time units. The first metabolite that feeds the reaction was not plotted.
Fig. 4
Fig. 4
The cholesterol model replicates knockout and desmosterolosis conditions. (A) Production of cholesterol is stopped only when both Tm7sf2 and Lbr are knocked out of the model. (B) The decrease of cholesterol level in desmosterolosis, due to Dhcr24 knockout, is reproduced by the model as well as the indefinite accumulation of desmosterol. In the model desmosterol accumulates because we are not explicitly modeling its metabolism similar to the human condition.
Fig. 5
Fig. 5
The cholesterol model replicates SLOS disease which is due to mutations in Dhcr7 that decrease enzyme activity. As the Dhcr7 reaction decreases by three orders of magnitude, 7-dehydrocholesterol and 27-hydroxy-7-dehydrocholesterol increase (A and B) and cholesterol (C) decreases.
Fig. 6
Fig. 6
Sensitivity analysis of cholesterol model shows strong influences of Idi2 and Fdft1 in cholesterol regulation. The sensitivity analysis was performed by using the algorithm proposed by Ingalls and Sauro, 2003. In most cases, changes in rate constants did not result in modifications in cholesterol production (not shown). In the mevalonate pathway (A), Hmgcr showed the strongest response, which was transitory. The isoprenoid and squalene synthesis (B and C) showed strong and constant influence. The rest of the pathway (Cholesterol I–III), showed transitory sensitivity (D–F). The negative sensitivity shown in (F) comes from the trivial case of changing the rate constant of cholesterol degradation reactions.
Fig. 7
Fig. 7
Predicting the metabolic profiles of Alzheimer’s and Huntington’s diseases. (A) Percent difference between baseline cholesterol metabolic profile and three different progressive stages of Alzheimer’s disease (incipient, moderate, and severe). (B) Percent difference between baseline cholesterol metabolic profile and Huntington’s disease. The metabolite numbers correspond to the names in Table S1 (ESI†).
Fig. 8
Fig. 8
Recovering baseline cholesterol metabolic profile. (A) Using the severe Alzheimer’s disease model we modified the value of enzyme produced by Idi2 (O) to recover the value of cholesterol in the baseline model (•). (B) Using the severe Alzheimer’s disease model as a starting point, we modified the value of the enzyme produced by Fdft1 (O) to recover the value of cholesterol in the baseline model (•). (C) We tested the model to a systematic change of Idi2 and Fdft1 to recover the value of cholesterol in the baseline model (•). The values found for Idi2 and Fdft1 were different for those determined in A or B. (D) The metabolic profile calculated using the modifications determined in C is the closest (in mean squared difference) to the baseline metabolic profile from all the metabolic profiles calculated in the mesh.

Similar articles

See all similar articles

Cited by 2 PubMed Central articles

Publication types

Feedback