Learn more Check out. Citing Literature. Related Information.
Dynamic Adaptation of Nutrient Utilization in Humans
Close Figure Viewer. Browse All Figures Return to Figure. Previous Figure Next Figure. Email or Customer ID. Forgot your password? Forgot password? Old Password. New Password. Password Changed Successfully Your password has been changed. Symptoms are controlled by removing lactose, sucrose, and glucose from the diet. In this condition , excess iron is deposited in several organs, and can cause:. MSUD disrupts the metabolism of certain amino acids, causing rapid degeneration of the neurons. If not treated, it causes death within the first few months after birth.
Treatment involves limiting the dietary intake of branched-chain amino acids. PKU causes an inability to produce the enzyme, phenylalanine hydroxylase, resulting in organ damage, mental retardation, and unusual posture.
Metabolic disorders are highly complex and rare. Electrolytes are naturally occurring compounds that control important bodily functions. Important electrolytes include calcium, magnesium, potassium…. Metabolic syndrome X is a group of five risk factors that can increase your chances of developing heart disease, diabetes, and stroke. Antidiuretic hormone ADH is a hormone that helps your kidneys manage the amount of water in your body. Red blood cells are normally shaped like discs, which allows them to travel through blood vessels.
Sickle cell disease causes red blood cells to be…. Maple syrup urine disease MSUD is a metabolic disorder in which your body can't break down certain amino acids. Read more on how to treat this rare…. Metabolism tests can tell you how effectively your body burns calories, and uses oxygen during workouts. They're a valuable tool which can help you…. A healthy lifestyle is about more than just proper nutrition and exercise. The second innovation of our method is the new concept of reaction profiling for each metabolic reaction and the new measurement for reaction similarity in the context of a specific disease, such as cancer.
By this method, each reaction can be represented quantitatively as its flux states in NCI cell lines, and thus computational models can be used to predict novel drug-reaction interactions for cancer drugs, which leads to the discovery of their novel drug targets. The genome-scale human metabolic network reconstructed by Duarte et al. In fact, only 59 cell lines have available gene expression data. We also downloaded protein sequence data from KEGG database [ 17 ], and computed the protein sequence similarities using Smith-Waterman method by Matlab.
Drug-target association data was downloaded from DrugBank. A network-based method is proposed by Shlomi et al. This inspires us to analyze cell line-specific metabolism based on gene expression data in NCI cancer cell lines. By integrating the metabolic network and the cell line-specific gene expression, we can predict reaction flux and activity for all the reactions in each cell line by our proposed linear programming model.
Thus, each metabolic reaction in the metabolic network has 59 flux levels in 59 cell lines. Reaction profiling is the procedure to represent each reaction as its flux values in NCI cell lines. Then for each metabolic reaction, if one of its enzyme genes is highly expressed, the reaction is considered as highly expressed.
If all of its enzyme genes are lowly expressed or undetermined, the reaction is considered as lowly expressed or undetermined, respectively. The boolean gene-to-reaction mapping can be obtained from the metabolic network model. Thus, based on the expression states of the corresponding genes, all the reactions are classified into a highly expressed subset R H , a lowly expressed subset R L and an undetermined subset. The reasons for the relaxation are twofold.
First, the relaxation allows the activity score of reactions to be continuous rather than binary value, which is more reasonable and flexible. The LP model is presented as follows:. Lowly expressed reactions are considered to be more likely to be inactive if they carry almost zero metabolic flux. Their closeness to one means i is likely to be active, while the closeness to zero means it is likely to be inactive.
We first apply LP model in each of NCI cancer cell lines to predict flux distribution for all the metabolic reactions, and then, we can obtain for each reaction its flux profile in these 59 cancer cell lines. Based on any reaction similarity metric s , we can define the conditional probability:. P E t is defined in the same way. We now define a similarity metric based on reaction flux profiles and then check whether it's a possible ideal metric.
The flux of reaction i in j th cell line, f ij , can be obtained from the optimal solution v of the above linear programming model for j th cell line. Each reaction is thus profiled using its 59 flux values in NCI cell lines, and can be considered as a data point in dimensional flux feature space. We use cosine similarity discussed more in Additional file 1 of reaction flux profiles to measure the similarity among metabolic reactions:. This is called reaction flux RF similarity between reaction i and j. Denote the set of target enzymes for reaction i by M i. Target sequence TS similarity for protein s and t is denoted by s TS s, t normalized to be between 0 and 1.
Then the reaction similarity for reaction i and j based on target sequence is defined as the maximum, average or minimum of all the possible enzyme target sequence similarities , , and. For simplicity, this is called reaction structure RS similarity s RS since it is defined based on enzyme targets' structure. The integration of s RF and s RS may provide an optimal measurement. For a specific drug, we can obtain its true interactions with m reactions from known drug-target interactions.
We aim to predict whether a new metabolic reaction interacts with the drug or not. Here we propose a Kernel KNN K-Nearest Neighborhood method for this binary classification problem using the similarities between reactions. Here x new is called the association score of the new reaction and the specific drug. Higher association score indicates a higher probability of the reaction interacting with the drug. Note we still call this model as Kernel KNN method even if s may not be positive semi-definite.
However, Kernel KNN considers the weight of voting from different individuals in the neighborhood. This task involves four stages including flux analysis, reaction profiling, drug-reaction interaction prediction and drug-target prediction. In the first stage, metabolic reactions in the metabolic network are first classified as highly or lowly expressed based on the expression levels of the genes which encode the enzymes catalyzing the reactions.
Integrating the classified reactions with metabolic stoichiometry, the LP model is used to predict the flux distribution for metabolic reactions in the different cellular environments of 59 cancer cell lines. After generating reaction profiles in stage two by putting reaction fluxes of 59 cell lines together, reaction flux similarity is constructed to measure the similarities among reactions functionally. With reaction flux similarity and their known interactions with drugs, stage three applies the proposed Kernel KNN model to predict novel drug-reaction interactions.
The last stage is to obtain novel drug targets using the hints from drug-reaction interactions. In this section, we discussed how to construct the Drug-Reaction Network based on the known drug-target interactions, and then reported some results from the analysis of this network. Flow chart for drug-target prediction. The 4-stage task includes flux analysis, reaction profiling, drug-reaction prediction and finally drug target prediction. Although relationships between drugs and targets have been depicted in a global view [ 18 ], the relationship between drug and enzymatic reactions remains uncharacterized.
The effect of a drug on human metabolism takes place through its target protein, which, as an enzyme, can catalyze its corresponding metabolic reactions directly. Thus, if any target of a drug catalyzes a reaction as an enzyme, the drug can interact with the reaction through the enzyme target and the reaction is considered to be a target reaction of the drug.
An illustrative example is shown in Figure 2. The green circle, yellow rectangle and red triangle represent drug, protein and reaction, respectively. Black edges and green edges indicate drug-target relationships and reaction-enzyme relationships. The blue solid edge between two reactions in Figure 2 indicates that two reactions interact with the same drug through different enzyme targets, while the blue dashed edge means that two reactions interact with the same drug through the same enzyme targets. An illustrative example of the relationship between drugs, enzyme targets and metabolic reactions.
Green circle, yellow rectangle and red triangle represent drug, protein and reaction, respectively. Black edge and green edge indicate drug-target interaction and reaction-enzyme relationship. The blue solid edge between two reactions indicates that two reactions interact with the same drug through different enzyme target, while blue dashed edge means that two reactions interact with the same drug through the same enzyme targets.
We constructed a Drug-Reaction Network DRN to depict the interactions between the approved drugs and metabolic reactions. We used Cytoscape [ 19 ] to show the network in Figure 3. The nodes in DRN include drugs and enzymatic reactions, and the edges between drugs and reactions indicate their interactions.
Two kinds of data are required to construct this bipartite graph: drug-target interaction data, which was obtained from the DrugBank [ 20 ] database, and gene-to-reaction mapping data, which was obtained from metabolic network Human Recon 1 reconstructed by Duarte et al. All drug-reaction interactions and the corresponding enzyme targets are listed in Additional file 2 , Table S1. Figure 4 shows the degree distribution in DRN.
We report the numbers of nodes both drugs and reactions, reactions only and drugs only with different degrees from 1 to 20, which are the network degree distribution, reaction degree distribution and drug degree distribution, respectively. Maximal degree of the drugs and reactions are and 35, respectively. DRN is generated by using the known interactions between approved drugs and their target reactions. Circles and triangles correspond to drugs and reactions, respectively. An edge between a drug node and a reaction node is placed if the drug targets at least one enzyme of the reaction.
The area of the drug reaction node is proportional to the number of reactions drugs the drug reaction interacts with.
Drug nodes are colored according to their Anatomical Therapeutic Chemical Classification, and reactions are colored according to their subsystems obtained from human metabolic network data. Green and blue represent the histograms for drug nodes and reaction nodes, respectively.
Overview of Metabolic Reactions – Anatomy and Physiology
We found that around half of the approved drugs in DRN are for metabolic diseases, cardiovascular diseases and antineoplastic drugs. This implies that among all 14 classes of drugs, these three classes are related to metabolic network most closely. Large percentage of these drugs in DRN shows that it may be a good way through DRN to get deep understanding of the mechanism of these drugs.
Particularly, for metabolism drugs, we found that targets of the remaining are related to metabolism by their own specific protein function, but not by catalyzing metabolic reactions as an enzyme, thus this part of metabolism drugs are not involved in DRN. In order to illustrate which pathways the drugs interact with, we also colored metabolic reactions according to their pathways. The most frequent pathway in DRN is transport.
For simplicity, we combine all these six transport pathways to one pathway as transport in Figure 3. The less frequent pathways are Carnitine shuttle 99 , Keratan sulfate 44 , Folate Metabolism 39 , Tyrosine metabolism 33 , Nucleotides 33 and Steroid Metabolism Thus, Figure 3 illustrates not only the interactions between the cancer drugs and enzymatic reactions, but also the pathways each drug interacts with. From Figure 3 , we can see that the resulting network is naturally clustered by major therapeutic classes and major pathways, although DRN layout is generated without the knowledge of drug classes and reaction classes.
The clustering of drugs may be mainly because they have the same targets. The most obvious cluster of drugs is the tightly connected drugs for the nervous system. Antineoplastic drugs, drugs for blood diseases, metabolic diseases, cardiovascular diseases, respiratory diseases and Musculoskeletal diseases are also clustered together, although less distinct than the cluster of drugs for nervous system. Reactions are clustered less obviously than drugs. The most obvious cluster of metabolic reactions is reactions related to folate metabolism and vitamin metabolism.
Further observation from Figure 3 is the relationship between drug classes and reaction pathways.
- Lunar surface models.
- Concurrent Prolog - Vol. 2: Collected Papers;
- Graffiti Japan;
Most of the neurological drugs, blood drugs and cardiovascular drugs are connected with transport reactions. Many cancer drugs and some of cardiovascular drugs are related to the nucleotides reactions. Reaction profiling we propose here is a method to quantify metabolic reactions. Using this method, the profile for each metabolic reaction is identified by its flux states in different cellular environments. Compared with the traditional equation representation of metabolic reaction, this new representation makes quantitative analysis of reactions obviously easier.
In each of NCI cell lines, LP model was used to predict metabolic flux distribution for all the metabolic reactions based on gene expression data in the cell line. Thus, each metabolic reaction was only designated one profile as its signature, which was used to quantitatively measure the similarity among reactions and thereby for prediction of novel drug-reaction interactions.
We only focused on enzymatic reactions since only enzymatic reactions have direct interaction with drugs through enzymes. Around of total metabolic reactions in the human metabolic network can be catalyzed by enzymes. Some of the enzymes in metabolic network have been identified as drug targets. We only consider reactions that are active in at least one cancer cell line. We report, in Figure 5 , the heat map of these reaction profiles using cosine similarity and complete linkage algorithm.
Red and green indicates different directions of metabolic reactions, while black indicates a flux value very close to zero. From the figure, we can see that both the enzymatic reactions and cancer cell lines cluster as two major groups. Especially, for enzymatic reactions, one group involves reactions with many positive fluxes red , while the other group involves reactions with many negative fluxes green.
Note that very few differences between cell lines can be seen in some metabolic reactions, which may be due to the common biological properties of cancer cell lines, for example, uncontrolled growth, invasion, and sometimes metastasis. Although the two groups of cell lines are not as distinct as reactions due to these common properties, they may imply further study for these cancer cell lines. Heat map of metabolic reaction profiles using complete linkage algorithm. Rows represent different reactions, and columns represent different cell lines.
Chemical Reactions in Metabolic Processes
Red and green indicate different directions of metabolic reactions, while black indicates a flux value very close to zero. To understand more about the relationship between reactions, enzyme targets, and drugs, we listed a table Additional file 3 with reaction pairs in DRN that share the same associated drugs, their reaction flux similarity scores, their common enzymes, and their common associated drugs.
Different reactions can interact with the same drugs in two ways. One is through the same enzymes, while the other is through different enzymes that are targets of the same drugs.