Predictive Metabolic Engineering: A Goal for Systems Biology

The landmark sequencing of Escherichia coli, Brewer's yeast (Saccharomyces cerevisiae), Arabidopsis, and human genomes has facilitated the emergence of systems biology—a science that is currently gaining the attention of molecular biologists. In addition to satisfying our intellectual desire to have a more detailed understanding of plant biology, we need to understand how plants work well enough to undertake truly predictive engineering of morphology and chemical composition. This would allow us to efficiently produce plants that are more productive sources of food and fiber as well as novel nutritional and industrial chemicals. What is currently described as genetic engineering of plant metabolism is really tinkering. Very rarely do our genetic manipulations cause the predicted effects—we usually discover new rate-limiting steps that prevent accumulation of the desired compound or induction of diversionary or catabolic pathways, or we observe undesirable pleiotropic effects that limit the usefulness of the modification. To enable true engineering of plants will require major changes in the way we approach science as well as major technological breakthroughs in analytical chemistry, bioinformatics, and other areas of genomics. We must generate robust mathematical models of metabolic pathways that accurately describe current experimental observations. These models can then be used to generate experimentally testable hypotheses, the results of which can be used to refine the model (Fig. 1). The goal is to go through this process enough times to create a model that begins to predict the behavior of the plant following perturbations by adding transgenes, making mutations altering cell fate, or environmental perturbations.