5
Screening Technologies III: Metabolomics

Metabolomics, like genomics and proteomics, can be used to assess drug safety. Rather than patterns of gene expression or protein–protein interactions, metabolomics (the characterization of small molecule metabolites produced in response to particular stimuli) is used to study the effects of drugs on various biochemical pathways. According to Klaus Weinberger, Chief Scientific Officer at Biocrates, one advantage of metabolomics over other approaches is that scientists currently have a stronger qualitative understanding of underlying biochemical pathways than of protein–protein interactions or interactions at the transcription level. The presentations addressing metabolomics illustrated how the technology is being used to gather information on toxicities and their underlying mechanisms. They highlighted four categories of metabolites that can provide insights at varying levels of complexity:

  • Markers for the activities of single enzymes

  • Direct multiparametric markers, which can indicate lipid elevation or lowering, metabolic control, insulin sensitivity, or inflammation

  • Multiparametric surrogate markers, which offer details about questions that are difficult to analyze directly, such as gluconeogenesis/ glycolysis, oxidative stress, and tissue damage and apoptosis

  • Mode-of-action markers, which indicate the presence of such responses as lipid signaling and regulatory metabolites



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 50
5 Screening Technologies III: Metabolomics M etabolomics, like genomics and proteomics, can be used to assess drug safety. Rather than patterns of gene expression or protein– protein interactions, metabolomics (the characterization of small molecule metabolites produced in response to particular stimuli) is used to study the effects of drugs on various biochemical pathways. According to Klaus Weinberger, Chief Scientific Officer at Biocrates, one advantage of metabolomics over other approaches is that scientists currently have a stronger qualitative understanding of underlying biochemical pathways than of protein–protein interactions or interactions at the transcription level. The presentations addressing metabolomics illustrated how the technology is being used to gather information on toxicities and their underlying mechanisms. They highlighted four categories of metabolites that can provide insights at varying levels of complexity: • Markers for the activities of single enzymes • Direct multiparametric markers, which can indicate lipid elevation or lowering, metabolic control, insulin sensitivity, or inflammation • Multiparametric surrogate markers, which offer details about questions that are difficult to analyze directly, such as gluconeogenesis/ glycolysis, oxidative stress, and tissue damage and apoptosis • Mode-of-action markers, which indicate the presence of such responses as lipid signaling and regulatory metabolites 0

OCR for page 50
 SCREENING TECHNOLOGIES III: METABOLOMICS METABOLOMICS AT METABOLON1 Dr. Milburn discussed some of the advantages of studying metabolites as opposed to proteins or gene expression. Biochemical molecules are the end result of many biological processes, and they can reflect the impact of a number of factors, such as the environment, a patient’s overall health, and any drugs a patient might be taking. While the technology used to analyze the human metabolome is complex, the metabolome is smaller than the genome or the proteome. According to the most recent estimates, there are only about 2,400 metabolites in the human body—significantly fewer than the approximately 25,000 genes, 100,000 transcripts, and mil- lions of proteins with which other fields must work. In a sense, metabolomic analysis can be thought of as an expansion of the traditional diagnostic tests performed on blood or urine and used to measure the levels of, for example, blood urea nitrogen, creatinine, and glucose. While these molecules represent a small portion of the total biochemistry of the body, the aim of metabolomics is to look at all, or at least a large proportion, of the body’s small molecules. The Metabolon Process At Metabolon, the goal is to be able to identify and quantitatively measure all of the small molecules in any sample type—urine, blood, tis- sue, or cell extract. The process used is illustrated in Figure 5-1. Sample preparation begins with four different fractionation steps to extract all polar and nonpolar molecules with a mass of 50–1,500 daltons. Once these small molecules have been separated out through these four extraction steps, they are pooled back together, and that sample is then split for analysis by two different platforms—a liquid chromatography–mass spec- trometry system (LC-MS) and a gas chromatography–mass spectrometry system (GC-MS). Company scientists use both of these platforms because small molecules can be very polar as well as very nonpolar; the two chro- matography methods work well together for profiling of most of the small molecules in the samples. Metabolon has developed proprietary software that makes it possible to identify automatically all the ions that are scanned by the spectrom- eters. Using automated processing techniques based on the biological variation of the compounds within samples, the researchers are able to reconstruct the original molecules to which the ions belonged before going through the system. With the help of a standard chemical library, 1This section is based on the presentation of Michael Milburn, Chief Scientific Officer, Matabolon.

OCR for page 50
 EMERGING SAFETY SCIENCE Sample Preparation FIguRE 5-1 Overview of the Metabolon process. Once the sample has been pre- pared, it is analyzed by a liquid chromatography–mass spectrometry (MS) method (with and without electrospray ionization [ESI]) and a gas chromatography–mass spectrometry method (with electron ionization). Proprietary software is used to analyze the data, and identify the molecules present in the sample and quantify their amounts. Because this is a fixed image,i.e., a bitmap, NOTE: QA = quality assurance; QC individual elements without redrawing, we can't darken = quality control. SOURCE: Milburn, 2007. so we have increased contrast overall. A side effect is that the gradated shading is now solid black the molecules are identified, and their amounts are quantified. The end fig 5-1 result is a data set that identifies all the small molecules seen in the sample and their relative amounts. Using these techniques, Metabolon is able to detect and analyze Revised metabolites that capture the vast majority of the biochemistry that occurs in the human body. Company scientists predict that after refining their methods, they will be able to identify even more molecules. Examples Several examples illustrate how the technology is used. The first is a study carried out with Bristol-Myers Squibb researchers who were inter- ested in different HIV protease inhibitors, many of which have a side effect of lipidystrophy, a degenerative condition in the body’s adipose tis-

OCR for page 50
 SCREENING TECHNOLOGIES III: METABOLOMICS sue. Metabolon researchers exposed cultured liver and fat cells to the dif- ferent HIV protease inhibitors and conducted a metabolomic analysis. The results were displayed in scatter plots (see Figure 5-2). The dots on each horizontal line represent the readout of a particular molecule. Since the experiment was run a number of times, there are many read- ings for each metabolite, and these are plotted along the horizontal axis in terms of their Z scores, where the zero point is the mean value of the control group for that particular molecule. Thus the amount of scattering in the scatter plot—how tight or how loose it is—offers a rough visual measure of the extent to which the total biochemistry of the cells was perturbed by the compound under examination. The scatter plots of the different HIV protease inhibitors display very different levels of perturbation. The earlier protease inhibitors, such as lopinavir and nelfinavir, tend to show much more overall biochemical perturbation, while the newer ones show less. Atazanavir caused the least amount of perturbation and was most similar to the vehicle control group. This result agreed with the clinical data, which indicated that some of the earlier protease inhibitors had a greater lipidystrophic side effect; the newer compounds, which had been specifically developed to have less of that effect, did indeed have fewer lipidystrophic side effects. Following this global analysis, the researchers tried to identify which compounds were being affected by the drugs and which pathways were being altered. In the liver cells, they found an increase in metabolized biochemicals produced in fatty acids—fatty-acid triglyceride metabolites. In the fat cells, on the other hand, they found an impairment of the Krebs cycle intermediates. Thus the analysis implied that the drugs were caus- ing a large change in the energy metabolism of the fat cells. A second example illustrates how this metabolomics technology can be used to examine a drug’s mechanism of action. The study was performed using an oncology drug in a myeloma cell line. This drug was known to induce apoptosis in about 1 day, but the mechanism of apoptosis was unclear. A simple study was conducted to look for significantly altered compounds in cell cultures treated with the drug or with a control at four time points spread over 27 hours and with six cultures per group. The number of biochemicals that were significantly changed (with p ≤0.1, q ≤0.2) rose over time and then leveled off after about 20 hours, at 65 compounds. The amount of change was more than twice as great as was commonly seen in such experiments—an indication that there had been very large perturbations of biochemistry because the cells had gone into an apoptotic state. By examining the individual biochemicals whose levels were altered, the Metabolon researchers were able to identify those metabolites that were most strongly affected. The level of sorbitol, for example, rose

OCR for page 50
 Control Atazanavir Indinavir Ritonavir Lopinavir Nelfinavir FIguRE 5-2 Global metabolomic analysis of commercially available HIV protease inhibitors. The scatter plots displayed above, depending upon how tight or how loose the scatter is, indicate different levels of cellular perturbation. Earlier protease inhibi- tors, such as lopinavir and nelfinavir, show more overall biochemical perturbation, while the newer protease inhibitors show less. 5-2 Atazanavir caused the least amount of perturbation and was most similar to the vehicle control group. SOURCE: Milburn, 2007. new type

OCR for page 50
 SCREENING TECHNOLOGIES III: METABOLOMICS steadily throughout the experiment, increasing by 3-fold at 13 hours and by 23-fold at 27 hours. Another compound, fructose-1-phosphate, per- formed similarly, increasing steadily throughout the course of the experi- ment until it was up 30-fold at the end of the 27 hours. From changes seen in the individual biochemicals, the researchers could also identify biochemical pathways that were being affected—both the subpathways and the superpathways. Among the superpathways, for instance, carbohydrate metabolism and lipid metabolism were strongly altered. Because the study drug clearly up-regulated the level of sorbitol, a molecule known from a number of previously published studies to induce apoptosis, the researchers attributed the mechanism of apoptosis to an increase in sorbitol. Summary As indicated by the above examples, metabolomic analysis is an effi- cient and valuable technology. With a turnaround time of 3–4 weeks, these studies can be performed relatively quickly. Further, this technology makes it possible to obtain highly specific data by analyzing biochemicals individually. These data can help in evaluating the side effects of test compounds, as well as in understanding mechanisms of action and of toxicity. METABOLOMICS AT BIOCRATES2 Weinberger’s presentation paralleled that of Milburn in a number of ways, addressing various means by metabolomics can be used to help ensure drug safety. The Biocrates Process The technology platforms used by Biocrates and Metabolon are simi- lar. The Biocrates platform provides fully automated sample preparation, mass spectrometric identification, and quantitation; bioinformatics is used for technical validations, visualization of statistics, and biochemical inter- pretation; and the entire process is based on an in-house bio bank or on samples collected from the partners with which Biocrates works. The analytical portfolio includes more than 1,000 annotated metabo- lites encompassing the main areas of intermediary metabolism. It contains primary and secondary amines, such as proteinogenic and nonproteino- genic amino acids, acylcarnitines and free carnitine, reducing monosac- 2This section is based on the presentation of Dr. Weinberger.

OCR for page 50
 EMERGING SAFETY SCIENCE charides and oligosaccharides, phospholipids, glycolipids, prostaglan- dins, bile acids, and many other metabolites. The technology can be applied to basic research, the agriculture and nutrition industries, clinical diagnostics and theranostics, and pharma- ceutical research and development. Pharmaceutical applications include studies of drug metabolism and pharmacokinetics, safety and toxicology, and pharmacodynamics and efficacy. Four categories of metabolites have been established, all of which offer insights at differing levels of complexity: • Biomarkers for the activities of single enzymes, which are relatively straightforward, as a simple ratio of product concentration to substrate concentration will generally provide an idea of the quantitative activity of a single enzyme • Direct multiparametric markers, or groups of markers that can indicate lipid elevation or lowering, metabolic control, insulin sensitivity, or inflammation • Multiparametric surrogate markers, or groups of markers that offer details about questions that are difficult to analyze directly, such as gluconeogenesis/glycolysis, oxidative stress, and tissue damage and apoptosis • Mode-of-action markers, or markers that indicate the presence of such responses as lipid signaling and regulatory metabolites An Example To illustrate the utility of metabolomics, Weinerger described a study of puromycin-induced toxicity. The study was conducted using four groups of six Sprague-Dawley rats: a vehicle control, and a low-dose (10 mg/kg), a medium-dose (20 mg/kg), and a high-dose (40 mg/kg) group. The researchers were blinded to the test compound given to the rats, so they did not know it was puromycin. Histopathology at 3 weeks revealed no damage in the control or the low-dose group, and only moderate nephrosis in the medium-dose group. The high-dose group, by contrast, developed end-stage renal disease after only 2 weeks and had to be sacri- ficed at that point. Plasma and urine samples were taken on days 3, 7, 14, and 22 in the first three groups and on days 3, 7, and 14 in the high-dose group. These samples underwent a metabolomics analysis that was cor- related with histopathology, pathophysiology, expression profiling, and proteomics. The analysis revealed a marker for general tissue damage. Acylcar- nitines are compounds that are produced in the mitochondria of energy- metabolizing cells, and in healthy tissue they remain inside these cells, so

OCR for page 50
 SCREENING TECHNOLOGIES III: METABOLOMICS that there are very low levels of circulating acylcarnitine in plasma. The metabolomic analysis showed that in the low-dose group, acylcarnitine plasma levels were very similar to those in the control group, but in the medium-dose and high-dose groups, there was a clear time-dependent increase in the circulating acylcarnitine levels, which implied tissue leak- age and, in particular, mitochondrial damage. The observance of general tissue damage prompted the researchers to seek to identify where the damage had occurred, and they checked for known markers of organ-specific toxicity. In testing for hepatotoxic- ity, they could not show a significant change in the bioassays for the bile acids, so they concluded that the liver was not the main site of the tissue damage. The analysis did, however, identify a number of markers for a wide variety of kidney-specific outcomes and mechanisms. It identified mark- ers for • moderate polyuria and, in at least half a dozen compound classes, tubular dysfunction; • general inflammation and oxidative stress, such as dose-dependent activation of COX, 12-LOX, and 15-LOX, although there was no sign of systemic oxidative stress; and • time-dependent moderate ketosis and the de-repression of NO synthase. In contrast with previously published, studies, in which tryptophan depletion was found to be due to increased synthesis of kynurenine, the markers showed that in this case, the tryptophan depletion was due mainly to conversion to serotonin, which implied that there was an addi- tional vasoconstrictor in this model. Using the information gained from all of these markers, the researchers were able to form a biochemically functional and plausible model of what was taking place in the study. Summary As evidenced by the above example, metabolomics can be useful in attempting to determine causes or sites of drug toxicity. Knowing as much as possible about how a drug might affect a specific pathway helps researchers see a more complete picture as they try to formulate answers. An important factor to consider in using metabolomics is heterogeneity of responses. In the above example, the animals used were genetically identical; in clinical settings, there will be widespread genetic variabil- ity. Researchers will need to determine whether the observed effects of

OCR for page 50
 EMERGING SAFETY SCIENCE metabolomic analysis are great enough to be significantly higher than the biological variability among a population. Although metabolomics can make a valuable contribution to under- standing disease, researchers continue to characterize disease from the perspective of different disciplines (e.g., pathology, physiology, and clini- cal chemistry). Weinberger asserted that the community must aim to unite these different disciplines in the assessment of molecular pathology. Further, he suggested that throughout the pharmaceutical industry, phar- macology, preclinical, clinical, and toxicology departments should focus on the same question of drug reaction utilizing all available perspectives. Such unification of disciplines could help reinforce evidence-based drug development.