The disaster of central control

Abstract

I t is commonplace to say that central control of large complex entities is doomed to fail. We have numerous examples from economies and societies of the past and the present demonstrating inefficiency when the systems exceed a certain size. In the old days, wise emperors were well aware of the problem and answered appropriately by the principle of divide et impera. Nature seems to have an elegant solution the problem too: Modular structure and partial autonomy of modules. The best example is the multicellular organism where the individual cell retains as much autonomy in metabolism as can be tolerated without endangering the whole system: A little more independence of somatic cells, for example, leads to tumor formation. Efficient division of labor is observed in bacterial cells too. No wonder I have thought that we can always learn from biology how to manage successfully the most sophisticated situations and how to handle and control complexity. It was a shocking experience therefore when I read a recent preprint and previous articles by John Mattick and his colleagues [1–3]. They present a plausible interpretation of the limitation in bacterial genome sizes based on DNA sequences: The number of genes in prokaryotes is bounded by an unaffordable regulatory overhead in genomes that are too large. This view is supported by an empirical fit of a power law to the data derived from some 90 fully sequenced prokaryotic genomes. The number of regulatory genes grows approximately with the square of the total number of genes, nr 1.63 10 5 n 1.2 10 5 n, where nr and n are the numbers of regulatory genes and all genes, respectively. Thus, complete regulation of ribosomal protein synthesis—when centrally organized on the DNA level in the spirit of the elegant operon mechanism as discovered by Jaques Monod, Francois Jacob, and Andre Lwow—falls into an inefficiency trap when genomes become too large. The guess is that the critical genome size for prokaryotes is in the range of 10,000 genes and indeed, this appears to be the size limit of bacterial genomes. Eukaryotes—these are all higher organisms from yeast to man—make use of other control mechanism in addition to genetic control at the DNA level. Examples are alternative splicing of precursormessenger RNAs (for reviews see [4, 5]), small interfering RNAs [6], genomic imprinting [7, 8], and other forms of epigenetic regulation of gene expression and silencing. For the purpose of illustration we shall consider here an idealized—and perhaps not yet fully accepted—model view of alternative splicing: The translation product of the same DNA stretch yields slightly or substantially different proteins depending on the individual organism and the particular tissue in which it is expressed. Diverse proteins are obtained by cutting out different sections called introns from the precursor RNA sequence. A well-known example is the protein that controls sex PETER SCHUSTER