Chemical Reaction Kinetics Is Back: Attempts to Deal with Complexity in

Developing a Quantitative Molecular View to Understanding Life

ew strategies to handle complexity in biology have been and are developed under the catch phrase “systems biology.” What stands in the core of this PETER SCHUSTER N recent field of research is the concept to understand and model cells and organisms as high-dimensional dynamical systems and to determine the necessary input parameters by experiment. Regulation of gene activities and metabolic func- tions are encapsulated in differential equations that have their origin in chemical Peter Schuster is Professor of reaction kinetics. Needless to say, this ap- Theoretical Chemistry at Vienna proach has to envisage enormous com- University. His publications, about 300 Over many decades, molecular plexity. On the other hand, solution of papers and nine books, deal with the biology has been extraordinarily analysis and prediction of molecular large numbers of kinetic equations, up to successful in applying a one thousand and more commonly structures from small molecules to qualitative molecular view to biopolymers and the dynamics of rather stiff equations, is routine in com- evolutionary processes. The books understanding life. bustion chemistry and flame modeling. include “The Hydrogen Bond”, three What’s new, however, is the fact that cel- collective volumes together with Georg lular reaction networks have a number of Zundel and Camille Sandorfy; “The unique properties unknown in physics and chemistry. They are not only self- Hypercycle—A Principle of Natural Self-Organization” together with regulated but they are also capable of reproduction, they are robust and don’t Manfred Eigen; and “Evolutionary change their state under often not so small changes in the environment, and they can Dynamics—Exploring the Interplay of tolerate loss of one or the other constituent without loosing function. Both the Selection, Accident, Neutrality, and experimental [1] and the computational approach to systems biology [2, 3] have Function”, a collective volume together made substantial progress within the last few years. Somehow, the mathematical with James P. Crutchfield. His recent analysis of the basic properties of genetic and metabolic networks is lagging behind research focuses on optimization of RNA structures as a model for [4, 5]. Despite undoubted success [6], many fundamental questions are still unan- molecular evolution and the role of swered. selectively neutral variants. Peter Over many decades, has been extraordinarily successful in Schuster is at the Institu¨t fu¨r applying a qualitative molecular view to understanding life. This qualitative image of Theoretische Chemie und Moleculare is based on yes-or-no answers rather than the conventional quantitative Strukturbiologie der Universita¨t Wien, Waeringerstraße 17, A-1090 Wien, Austria. Correspondence to: Peter Schuster, E-mail: [email protected]

14 COMPLEXITY © 2004 Periodicals, Inc., Vol. 10, No. 1 DOI 10.1002/cplx.20056 description applied in physics and struction of raw images come to an end chemistry. It uses rough pictures re- [8]? Has the period of gathering facts Enormously complex regulation placing the commonly very fine details reached the limits because the volume networks, rather than simple of molecular structures. Function is il- and the diversity of data escape the cascades, are formed by gene lustrated by means of cartoons rather imaginative power of the human ? interaction and a new kind of than equations. This was not always so. I don’t think so, but the current mass network theory that allows for Until the 1970s, biochemical kinetics production of experimental data indeed asking the appropriate questions, was central to the research in molecular provides new and hitherto unknown is required. biosciences and explored cooperative challenges: How are databases created processes, such as allosterically induced that allow for fast and unambiguous re- conformational changes of biopoly- trieval and provide convenient tools for active structural or housekeeping genes mers, and the mechanisms of enzyme comparison of information from very and about the same number of specific regulatory genes that define the state of reactions. Within the last three decades different sources? The conventional the and its role within the organ to of the twentieth century, however, techniques of data processing work fine which it belongs. All this is executed by mathematics and quantitative thinking with precisely defined objects, for ex- means of a complex network, inter- were largely banned from molecular bi- ample, sequences and structures at weaving gene activities in subtle man- ology. I’ve heard hard-nosed professors atomic resolution. Gene expression ner. The problem is to cut, or better yet, of molecular biology at European uni- data from microarrays are different in to release this Gordian knot. versities saying, “Molecular biology as I this respect because they have an indis- Chemical reaction kinetics such as understand it, is qualitative in nature!” pensable quantitative component that combustion or polymerization have Let me take, for a moment, the position is poorly reflected by a yes-no-maybe plenty of experience with high-dimen- of devil’s advocate: Molecular , sional ordinary or partial differential one might well say, is even in a pre- equations. On the other hand, non- Linnaean state because there is no sign Gene expression data from linear chemical systems have been for the beginning of the development of microarrays are different in this respect because they have an investigatedingreatdetail[9].Beingauto- a systematic and generally accepted no- indispensable quantitative catalytic processes, these multistep re- menclature of genes and gene products. component that is poorly actions represent excellent examples of Instead, molecular continue reflected by a yes-no-maybe multiple steady states, oscillations, de- to name the genes they’ve discovered, classification. terministic chaos, and spatial pattern after the entries in their laboratory formation. The way from the relatively notebooks, or, they use more or less simple nonlinear chemical model sys- arbitrary names taken from various classification. Theory is required to tems to the characterization of the sources. I am not denying that there are bring order into the more complex states of cells by means of attractors is serious attempts toward a more system- “heaps of facts” before they can be pro- elaborate and hard to go into detail, but atical nomenclature, but they have not cessed efficiently by computer tech- it is straightforward. Complex chemical (yet) made it into daily laboratory work. niques. The really fundamental prob- reactions can also be a suitable study Starting in the mid-1990s, biologists be- lem, however, arises from the numbers model for the development of novel re- gan to feel the lack of comprehensive of genes, which lie between a few thou- verse engineering tools [10] for the theory and quantitative thinking. As Sir sand in bacteria and some 30,000 in study of biological complexity. Perhaps had already said a de- humans, and the nature of their inter- engineering theory is a good method for cade earlier, “No new principle has de- actions: Enormously complex regula- providing insight into the interplay be- clared itself from below a heap of facts.” tion networks, rather than simple cas- tween resilience, robustness, modular- Even the pioneer of molecular biolo- cades, are formed by gene interaction ity, and hierarchical control in biologi- gists and Nobel laureate, Sidney Bren- and a new kind of network theory that cal systems. Because most of the kinetic ner, urged the development of a novel allows for asking the appropriate ques- rate parameters of cellular processes are quantitative and comprehensive theo- tions, is required. For a moment, let us unknown and their determination retical biology in an interview that he imagine the complexity of a mamma- through measurements is difficult, ex- gave for the German magazine Labor- lian cell: 30,000 genes have the potential pensive, and often almost impossible, journal. [7]. to produce the same number of gene the solution of the inverse mathemati- Why are the biologists now calling products, but, in every cell the majority cal problem of reaction kinetics consist- for the return of quantitative aspects? of gene activities have to be down reg- ing of the determination of parameters Have qualitative thinking and the con- ulated to leave us with a few thousand from measured data is a great challenge

© 2004 Wiley Periodicals, Inc. COMPLEXITY 15 (The forward problem is the computa- computer science, physics, electrical en- date education of biology students. Uni- tion of solution curves from rate param- gineering, and chemistry, into modern versity curricula have to be adapted to eters and initial conditions). biology. In reality, to achieve in such a these new developments. Fortunately, It is already commonplace to say that great synthesis toward the life sciences of this fact has already been appreciated understanding complexity in biology and the future is a different story, but inter- and has reached current awareness in the in other disciplines will not be possible disciplinary research is no longer placed United States [11, 12] and in other coun- without a joint effort integrating experi- at the side-table of funding agencies. The tries. With very few exceptions, however, ence from many branches of science reorientation of molecular life sciences the universities in continental Europe are and engineering, including mathematics, has clear-cut consequences for an up-to- still lagging behind.

REFERENCES 1. Tewari, M. et al. Systematic interactome mapping and genetic perturbation analysis of a C. elegans TGF-␤ signaling network. Mol Cell 2004, 13, 469–482. 2. Hasty, J.; McMillen, D.; Isaacs, F.; Collins, J.J. Computational studies of gene regulatory networks: In numero molecular biology. Nat Rev Genet 2001, 2, 268–279. 3. Kitano, H. Computational systems biology. Nature 2002, 420, 206–210. 4. Strogatz, S.H. Exploring complex networks. Nature 2001, 410, 268–276. 5. Friedman, N. Inferring cellular networks using probabilistic graphical models. Science 2004, 303, 799–805. 6. Almaas, E.; Kova´cs, B.; Vicsek, T.; Oltvai, Z.N.; Baraba´si, A.-L. Global organization of metabolic fluxes in the bacterium Escherichia coli. Nature 2004, 427, 839–843. 7. im Gespra¨ch, Eine einsame Stimme aus der Pra¨genomik A¨ ra´ (An interview with Sydney Brenner: A lonely voice from the pre-genomic era). Laborjournal 2002, Heft 4, 28–33. 8. Fagerstro¨m, T.; Jagers, P.; Schuster, P.; Szathmary, E. Biologists put on mathematical glasses. Science 1996, 274, 2039–2040. 9. Sague´s, F.; Epstein, I.R. Nonlinear chemical dynamics. J Chem Soc Dalton Trans. 2003, 1201–1207. 10. Csete, M.E.; Doyle, J. C. Reverse engineering of biological complexity. Science 2002, 295, 1664–1669. 11. Bell, E. The future of education in the molecular life sciences. Nat Rev Mol Cell Biol 2001, 2, 221–225. 12. Ideker, T. Systems biology 101—What you need to know. Nat Biotech 2004, 22, 473–475.

16 COMPLEXITY © 2004 Wiley Periodicals, Inc.