The Genom of Homo sapiens.pdf

346 WESTON ET AL.these differences and then trigger a second round ofperturbations generating new global data sets to testthe hypothesis. At each round the model will be refined.This process will continue iteratively until themodel and experimental data are consistent with oneanother. The ultimate objective is to move toward anaccurate mathematical model. Hence, systems biologyis hypothesis-driven, global, quantitative, integrative,and iterative.MODEL ORGANISMS FOR SYSTEMSANALYSISWe have exploited two organisms for applying systemsapproaches: the yeast, Saccharomyces cerevisiae, and thearchaeon, Halobacterium sp. S. cerevisiae is an idealmodel system in that the organism is single-celled, itsgenome is fully sequenced, extensive genetic manipulationsare possible, it is easy to grow, and an enormous experimentalliterature is available. Moreover, deletionstrains are available for more than 95% of the ~6200 openreading frames (ORFs) in the yeast genome (Giaever etal. 2002). Because of these advantages, genome-widestudies such as microarray analysis, proteomics, genomewidebinding analysis, and large-scale genetic interactionstudies became available for yeast much sooner than formost other organisms. Similarly, technologies for characterizingprotein–protein interactions are most advancedand easiest to study using yeast. As a result, a vast amountof global information is available, including protein–proteinand protein–DNA interactions, changes in mRNAexpression, and phenotypic assays.Halobacterium sp. offers a striking contrast to yeast inthat far less biological information is available, hence itaffords a test of what systems biology can do in a lesswell developed model organism. It is easily cultured, itsgenome is sequenced, an array of genetic, biochemical,and genomic tools have been developed, and its robustand interesting physiology make Halobacterium sp. idealfor evaluating the response of relevant biological systemsto fluctuating environmental factors (DasSarma andFleischmann 1995; Ng et al. 1998, 2000; Peck et al. 2000;Baliga et al. 2001, 2002). As in other model organisms,certain biological responses such as phototrophy (conversionof light to energy), aerotaxis (movement toward oxygen),and phototaxis have been characterized, one or afew genes or proteins at a time, and serve as benchmarksto help validate the global systems-level interpretationsof genome-wide data (Oesterhelt and Stoeckenius 1973;Spudich and Bogomolni 1984; Spudich et al. 1989;Oesterhelt 1995; Oren 1999; Peck et al. 2001). This organismalso has unusual protein chemical features thatwill enormously facilitate global structural studies.GALACTOSE UTILIZATION IN YEAST:A BENCHMARK SYSTEMThe galactose system in yeast is one of the best-studiedeukaryotic systems of gene regulation, because of its relativesimplicity and because the central transcription factor,Gal4, is one of the strongest transcriptional activatorsdescribed to date. The galactose system refers to a biochemicalpathway that enables cells to utilize galactose asa carbon source, and the regulatory network that controlsthe switching on (in the presence of galactose) and off (inthe absence of galactose) of the pathway (for review, seeLohr et al. 1995; Reece 2000). A permease (Gal2) transportsgalactose into the cell, whereas the enzymatic proteinssubsequently convert intracellular galactose to glucose-6-phosphate.These include galactokinase (Gal1),uridylyltransferase (Gal7), epimerase (Gal10), and phosphoglucomutase(Gal5/Pgm2). Three regulatory proteins,Gal3, Gal4, and Gal80, exert transcriptional control overthe transporter, the enzymes, and to a certain extent, eachother. Gal4 binds to specific sequences upstream of theGAL genes (with the exception of the GAL4 gene itself)to potently activate transcription. In the presence of glucose,this is prevented by the action of Gal80, a co-repressorprotein, which binds to and inhibits Gal4. In themost recent model, upon galactose induction, Gal80 istranslocated to the cytoplasm where it interacts with Gal3(Peng and Hopper 2000, 2002). This Gal80–Gal3 interactionis important for relieving the repressive function ofGal80. Gal4, no longer repressed by Gal80, activates theGAL genes through a mechanism involving a phosphorylatedversion of the transcription factor (Mylin et al. 1989,1990).Despite years of research on the galactose system, andits well-characterized regulatory network, recent findingsindicate that even this relatively simple system is muchmore complex than previously thought. First, a systemsbiology study of this system (outlined below) revealedthat the cellular response to galactose extends far beyondthe activation of the Gal genes (Ideker et al. 2001). Second,the activation and repression of the Gal genes requiresa more extensive repertoire of regulatory proteinsthan simply Gal4, Gal3, and Gal80. A major undertakingin our laboratory has been to characterize the regulatorynetworks that control the galactose response in yeast,with respect both to the regulation of the Gal genes themselvesand to the control of genes indirectly affected byactivation of the galactose system.STEADY-STATE PERTURBATION OF THEGALACTOSE SYSTEMAs described above, an important task in solving generegulatory networks for any given pathway is to catalogthe changes in gene and protein expression controlled bythat pathway. To do this comprehensively requires integratinginformation both from steady-state perturbationexperiments and from kinetic analyses. With respect tothe galactose system, the former was carried out throughsystematic environmental (presence or absence of galactose)and genetic (knockouts) perturbations to the system.Employing nine strains of yeast, each with a differentgalactose gene knocked out, and the wild type, the mRNAlevels of ~6200 yeast genes were monitored, with the systemon and off for each genetic perturbation. Nine hundredninety-seven mRNAs were changed in one or moreof these perturbations. In addition, the quantitativechanges in protein expression for 300 proteins of the