112N16 Tox Focus

View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by PubMed Central Environews Focus A 938 VOLUME 112 | NUMBER 16 | November 2004 • Environmental Health Perspectives Focus | Systems Biology Systems Biology The Big Picture enomics, proteomics, and metab- “Traditionally, scientists—toxicologists G olomics have all vastly advanced included—have relied on a reductionist our understanding of human biology and approach to biology,” says William Suk, disease. But the functioning of even a director of the NIEHS Center for Risk and simple system such as a single yeast cell or Integrated Sciences. Even now, many stud- bacterium is much more complicated ies examine complex systems by looking at than the sum of its genes or proteins or cellular components in isolation. For metabolites; it’s the activity of all those instance, a common experiment involves components and their relationships to using DNA microarrays to observe the one another that add up to a living organ- effect of a chemical exposure on thousands ism. Recognizing that complexity, the of genes at once. This technique can emerging field of systems biology quickly tell a scientist which genes may be attempts to harness the power of mathe- vulnerable to that exposure. But a systems matics, engineering, and computer sci- biology approach would attempt to model ence to analyze and integrate data from all not only the chemical’s effect on gene the “omics” and ultimately create working expression but also how that expression Photodisc, Matt Ray/EHP, Joseph Tart/EHP Photodisc, Matt Ray/EHP, models of entire biological systems. will affect protein function, and in turn Environmental Health Perspectives • VOLUME 112 | NUMBER 16 | November 2004 A 939 Focus | Systems Biology how the exposure will affect cell signaling. delineation of a complete parts list of all suggests the answer to that problem lies in “There’s nothing wrong with what we’ve the genes,” he says. “If you know all the highly miniaturized sensors and detectors been doing,” Suk says. “But systems biol- genes, you have the ability to do DNA developed through nanotechnology and ogy is going to take it to another level.” arrays, follow the behavior of all the mes- microfluidics. senger RNAs, and even the proteins, in Another “omics” that promises to fill Building a New Science principle.” in gaps in systems biology models is From one perspective, systems biology is Measuring gene expression is one metabolomics—the evaluation of tissues nothing new. At the turn of the twentieth important component of systems biology, and biofluids such as urine, blood plasma, century, physiologists such as Walter B. and methods for doing so are fairly well and saliva for metabolite changes that may Cannon were developing the concept of developed for the needs of this field. result from environmental exposures or homeostasis—the self-regulatory mecha- However, proteomics—the science of ana- from disease. Because metabolites (which nisms, hunger and thirst for example, that lyzing all the proteins present in a system include carbohydrates, amino acids, and a living organism uses to keep its internal at any one time—still has some maturing lipids) are the actual by-products of pro- cessing food into energy, this “omics” has the potential to paint a picture of what has actually happened in the cell. Work in metabolomics seeks to go We’re exposed to lots of beyond sampling single metabolites to developing profiles of four or five related chemicals but at very low con- metabolites. But to create meaningful centrations over time. We need metabolite profiles, scientists need better tools for measuring tiny amounts of tools to help us understand metabolites and for determining which how complex exposures are most important in the activity of the cell. Existing analytical tools such as perturb complex systems. nuclear magnetic resonance spectrometry and mass spectrometry go only so far. — William Suk Indeed, the Metabolomics Technology Development initiative of the NIH Roadmap for Medical Research encour- ages projects that develop new methods systems in balance despite an ever-changing to do before scientists can integrate its for measuring metabolites that are present external environment. The term “systems data payload into a true systems approach. only in low concentrations or at specific biology” was first used in the 1960s, when Proteomics has been hailed as having even subcellular locations. theoretical biologists began creating com- more potential than genomics, because puter-run mathematical models of biolog- whereas DNA is a set of static instructions The Measure Is in the Models ical systems. for an organism, proteins—the machines For those working on a systems biology But the field took a leap forward begin- that actually carry out the work—are a approach, the goal of developing these ning in the 1990s, when the high-through- more fluid medium and may reflect the various “omics” technologies is to com- put tools developed for the sequencing of effects of chemical exposure more accu- bine their data into interactive models. the human genome brought experimental rately. But to integrate protein expression Hood says, “The ultimate object of sys- scientists up to the speed of theoretical biol- into systems biology, scientists need to tems biology is to understand how the ogists. The widespread use of the Internet better understand its relationship to gene elements and their interactions together has also made possible for the first time the expression. DeLisi says scientists still don’t give rise to the emergent properties of the international collaborations and sharing of understand why in many cases there is a system.” huge amounts of data that systems biology tight correlation between gene and pro- To do that, scientists begin by model- requires. “The way that computer science tein expression, while in others (as with ing individual components such as pro- has responded to genomics is one of the transcription factors) the correlation is tein networks and signal transduction great stories of the sociology of twentieth- very loose. “That [understanding] will pathways. Initially, says Hood, the models century science,” says Charles DeLisi, senior develop over the next five to ten years,” he are descriptive. They may involve perhaps associate provost for bioscience and chair of predicts. a relatively simple equation showing rela- the Bioinformatics Program at Boston Other researchers have pointed to the tionships between a few proteins in a cell. University. Computer scientists have taken a need for more quantitative techniques to As more information comes to light, the great interest in biology and have stepped not only detect the presence of proteins models become more graphical. A graphi- up to collaborate with biologists to develop but also determine their size, purity, and cal model may visualize a cell as a very the tools needed to sequence genomes and concentration in a system. Hood agrees complicated flow chart, as a series of pin- analyze the resulting data. that to achieve truly global analyses of wheels, or as a spiderweb. Relationships Leroy Hood, a biochemist who is pres- complex biological systems, proteomics between elements are depicted through ident and cofounder of the nonprofit technology needs more development. color or distance. Institute for Systems Biology, agrees. “The big problem that proteomics has is Next, researchers can experiment with “What uniquely defines the systems biol- that proteins that are expressed at very low an actual system such as yeast to see what ogy that I’m thinking about has really levels are generally invisible to the analyt- will happen at the organism level when come from the genome project and its ic techniques we have,” he says. Hood one component of the system is perturbed. A 940 VOLUME 112 | NUMBER 16 | November 2004 • Environmental Health Perspectives Focus | Systems Biology “You can do genetic perturbations where frames and having engineers design big under ideal conditions, but which enable you knock out genes, for example, or envi- databases,” says Eric Neumann, global the plane to stabilize if conditions sudden- ronmental perturbations where you give head of knowledge management at ly change. Likewise, biological organisms or take away certain kind of sugars,” Hood Aventis Pharmaceuticals. All of this com- include complex control systems that kick says. “And then you observe how all the putational power is aimed at trying to in only during potential threats—such as other elements behave in response to those combine information from experiments variations in temperature or nutrients—to perturbations.” Those experiments will that don’t have an integrated design—for keep the organism stable. usually yield data that aren’t explained by instance, trying to relate gene expression In general, this complexity makes the the model. “So you formulate hypotheses data from one study to protein data col- organism robust. But some scientists to explain the discrepancies, and you go lected in a separate study. hypothesize that such complexity can back and do more of these global and inte- Neumann says this integration can be leave a system vulnerable to unplanned grated experiments,” Hood says. done better for one-tenth of the cost by disruptions such as genetic mutation. The It’s a long way from modeling yeast to adopting good experimental design and mutation may be tiny, but because the modeling a human being. But Hood focusing more on the downstream. gene is involved in such a complex, mul- believes that the knowledge gained from a Neumann’s ideal experiment involves col- tilayered control network, the tiny muta- model of a simple system can be scaled up. lecting gene microarray data, protein lev- tion can trigger a “cascading failure”—a Such comparative genomics—the ability els, metabolite levels, clinical phenotypes, kind of domino effect that leads to a to learn about complex systems by model- and serum biomarkers in one experiment.

Load more