Site map > Home page > CV > Life > Systems in Biology 1

Putting the Systems back into Systems Biology

This page is the first of a series containing the article Putting the Systems back into Systems Biology by Athel Cornish-Bowden, published in Perspectives in Biology and Medicine (2006) 49, 475–489. A PDF file is also available.


In recent years the term systems biology has become widespread in the biological literature, but most of the papers in which these words appear have surprisingly little to do with older notions of biological systems: they often seem to imply little more than reductionist biology applied on a very large scale with a little attention to interactions between some of the components, but with minimal attention to the kinetic properties of enzymes, which supplied much of the reductionist foundation of biochemistry. A systemic approach to biology ought in reality to put the emphasis on the entire system; insofar as it is concerned with components at all it is to explain their roles in meeting the needs of the system as a whole. Genuinely systemic thinking allows us to understand how biochemical systems are regulated, and why clumsy attempts to manipulate them for biotechnological purposes may fail. At a more abstract level, it is necessary for understanding the nature of life, because as long as an organism is treated as no more than a collection of components one cannot ask the right questions, and certainly cannot answer them.


NOw after all this that hath been said, I cannot but hope that Those many False and Ignorant Outcries against the Lute will be laid aside, and deem’d (as indeed they are) False.
I will here Name some of Them;
First, That it is the Hardest Instrument in the World.
Secondly, That it will take up the time of an Apprenticeship to play well upon It.
Thirdly, that it makes Young People grow awry.
Fourthly, That it is a very Chargeable Instrument to keep; so that one has as good keep a Horse as a Lute, for Cost.
Fifthly, that it is a Womans Instrument.
Sixthly, and Lastly, (which is the most Childish of all the rest) It is out of Fashion.
I will give here a short (but True) Answer to each of These Aspersions.

Thomas Mace (1676)

Mace, T. (1676) Musick’s monument, London; facsimile reprint 1958, Paris: Éditions du Centre National de la Recherche Scientique

Writing in 1676, Thomas Mace was concerned with the playing of the lute, not the kinetic study of enzymes, but most of the objections that he lists (apart from the tendency to make young people grow awry) have their parallels in the modern habit of thinking that biochemistry can be taught and practiced without giving serious attention to metabolism and enzymology. The study of kinetics has always been regarded as difficult, there are those who think that it takes too much student time to be taught properly, and in recent years it has come to be unfashionable. Mace described the idea that the lute is difficult to play as that Flim-Flam-Ignorant saying of the Vulgar; anyone who has seriously studied the kinetic behavior of enzymes will feel, similarly, that the difficulties are greatly exaggerated. In any case, one can hardly expect to make significant advances in any domain of knowledge if its classical foundations are ignored. Over the past decades the effects of ignoring classical enzymology in the stampede into molecular biology can be seen in the failure to achieve the promised benefits of biotechnology as easily and as rapidly as many early enthusiasts had expected.

Genome studies and biotechnology

The disappointing results of efforts in biotechnology can be attributed to basing them on a false premise, and the main purpose of this essay will be to discuss this false premise. First, however, I should mention the failure of genome sequencing to provide benefits commensurate with the enormous financial investment. Although it might be argued that there has not been enough time to see the benefits, this argument must eventually wear thin. In any case, a long list of gene sequences is barely more useful in itself than a list of all the telephone numbers in a city would be if it contained no information about which telephone corresponded to any given number. A telephone directory that contained no information at all about the telephones corresponding to half of the numbers, together with plausible and mainly correct guesses about the other half, would be better than nothing, but it would be far from an ideal tool for making the best use of a telephone. This analogy describes quite well the genome data for organisms that have hardly been studied at all in biochemical terms, such as Treponema pallidum, the organism responsible for syphilis. The functions of perhaps half of the genes of Treponema pallidum can be plausibly guessed by comparing their sequences with those of genes identified in other bacteria that have been more thoroughly studied from a biochemical point of view, such as Escherichia coli. This is useful, of course, but not as useful as it may appear at first sight, because it tells us more about the similarities between Treponema pallidum and Escherichia coli than it does about their differences, whereas it is the differences that make Treponema pallidum worthy of sequencing in the first place; after all, it causes syphilis, and Escherichia coli does not.

Cárdenas, M. L. (1995) Glucokinase: its regulation and role in liver metabolism, Austin: R. G. Landes

The point here is that when we want to understand why two systems behave differently we need to study the differences between them, but the futility of trying to understand differences by looking at similarities is perhaps better illustrated with a kinetic example. A major point of interest of hexokinase D, the predominant isoenzyme of hexokinase in the mammalian liver, is that its kinetic properties are very different from the isoenzymes characteristic of brain, muscle and other tissues: it is insensitive to its product glucose 6-phosphate at physiological concentrations, and instead of being saturated at physiological concentrations of glucose it responds with kinetic cooperativity to variations in the physiological range (Cárdenas, 1995). These properties suit it well for its role as a glucose sensor, and also for regulating the rate of glycogen synthesis in the liver according to supply, unlike many biosynthetic processes, which are typically regulated according to demand for the end-product. They also explain why mutations of the relevant gene are dominant in maturity-onset diabetes of the young (MODY), in contrast to the usual recessive character of most genes for the enzymes of primary metabolism. Until recently, however, the only information about the three-dimensional structure of hexokinase D came from studies of yeast hexokinase, an enzyme with completely different kinetic properties. Such comparisons were useful for indicating that the chemical mechanism of phospho transfer is essentially the same in all hexokinases—something worth knowing, certainly, but not in any way surprising—but they explained none of the properties that made hexokinase D interesting in the first place.

Cornish-Bowden, A. and M. L. Cárdenas (2000) From genome to cellular phenotype—a role for metabolic flux analysis? Nat. Biotechnol. 18, 267—268

Returning to genome sequences, the problem is not so much that they contain no phenotypic information but that we do not have reliable methods for undertaking all of the steps involved in deducing a phenotype from them (Cornish-Bowden and Cárdenas, 2000). Present methods of sequence analysis allow a genome sequence to be converted into a list of genes without much difficulty, but transforming the list of genes into a list of enzymes has a high rate of failure, perhaps 50%. In any case, however, this is only the beginning of the process. A list of putative gene products, or even a list of putative enzymes, is not a phenotype, and converting it into a phenotype requires construction of a plausible metabolic map, which then needs further work to convert it into a possible phenotype. Finally, the possible phenotype can only become a real phenotype when all relevant kinetic and regulatory properties are taken into account, together with information about how all the components are organized into a three-dimensional whole, even a four-dimensional whole, given that the times when different components are made may be just as important as where they are placed.

Schuster, S., D. A. Fell, and T. Dandekar (2000) A general definition of metabolic pathways useful for systematic organization and analysis of complex metabolic networks, Nat. Biotechnol. 18, 326–332

Current methods of sequence analysis are useful for beginning this process, and knowledge of classical biochemistry is useful for finishing it, but there are several intermediate stages that present more difficulty. This is especially true when dealing with an organism for which no classical biochemical measurements have been made, so that kinetic modeling is virtually impossible. This problem has led to the development of methods of stoichiometric analysis that seek to deduce as much as possible from the structure of a metabolic map alone, making no use of kinetic or regulatory information. In an application of this approach, Stefan Schuster, David Fell, and Thomas Dandekar (2000) noted that the genome of Treponema pallidum contains genes for two different kinds of transketolase, an enzyme needed for energy management in many organisms, but no recognizable gene for a related enzyme known as transaldolase. This could be taken as evidence that the bacterium does not use the second kind of enzyme, except that the two reactions always occur together in other organisms, so that in terms of known metabolism neither can have any role in the absence of the other. This led Schuster and colleagues to conclude that the failure to identify genes for both enzymes in Treponema pallidum reflected incomplete information and not a genuinely missing enzymic activity.