Supra-domains: Evolutionary Units Larger than Single Protein Domains Christine Vogel 1 *, Carlo Berzuini 2,3 , Matthew Bashton 1 , Julian Gough 4,5 and Sarah A. Teichmann 1 * 1 MRC Laboratory of Molecular Biology, Hills Road, Cambridge CB2 2QH, UK 2 MRC Biostatistics Unit Institute of Public Health Cambridge CB2 2SR, UK 3 Dipartimento di Informatica e Sistemistica, University of Pavia, 27100 Pavia, Italy 4 Genome Exploration Research Group, RIKEN Genomic Sciences Centre, W121 1-7-22 Suehiro-cho, Tsurumi-ku Yokohama 230-0045, Japan 5 Department of Structural Biology, Fairchild bldg, D109 Stanford, CA 94305-5126 USA Domains are the evolutionary units that comprise proteins, and most proteins are built from more than one domain. Domains can be shuffled by recombination to create proteins with new arrangements of domains. Using structural domain assignments, we examined the combinations of domains in the proteins of 131 completely sequenced organisms. We found two-domain and three-domain combinations that recur in different protein contexts with different partner domains. The domains within these combinations have a particular functional and spatial relationship. These units are larger than individual domains and we term them “supra-domains”. Amongst the supra-domains, we identified some 1400 (1203 two-domain and 166 three-domain) combinations that are statistically significantly over-represented relative to the occurrence and versatility of the individual component domains. Over one-third of all structurally assigned multi-domain proteins contain these over- represented supra-domains. This means that investigation of the structural and functional relationships of the domains forming these popular combinations would be particularly useful for an understanding of multi-domain protein function and evolution as well as for genome annotation. These and other supra-domains were analysed for their versatility, duplication, their distribution across the three kingdoms of life and their functional classes. By examining the three-dimensional structures of several examples of supra-domains in different biological processes, we identify two basic types of spatial relationships between the component domains: the combined function of the two domains is such that either the geometry of the two domains is crucial and there is a tight constraint on the interface, or the precise orientation of the domains is less important and they are spatially separate. Frequently, the role of the supra-domain becomes clear only once the three-dimensional struc- ture is known. Since this is the case for only a quarter of the supra-domains, we provide a list of the most important unknown supra- domains as potential targets for structural genomics projects. q 2004 Elsevier Ltd. All rights reserved. Keywords: domain combination; protein family; domain architecture; multi-domain protein; functional annotation *Corresponding authors Introduction Domains are the evolutionary and structural units that form proteins: they can occur on their own in single-domain proteins, or in combination with different partner domains making multi- domain proteins. The domains within a protein are often also structurally and functionally inde- pendent. Domains that are related to each other by descent from a common ancestor, are members of the same superfamily as described in the struc- tural classification of proteins database, SCOP. 1 The SCOP definition of a domain is evolutionary: SCOP domains exist either on their own and/or in combination with other domains. 0022-2836/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. E-mail addresses of the corresponding authors: cvogel@mrc-lmb.cam.ac.uk; sat@mrc-lmb.cam.ac.uk Abbreviations used: HMM, hidden Markov model; SCOP, structural classification of proteins. doi:10.1016/j.jmb.2003.12.026 J. Mol. Biol. (2004) 336, 809–823