chains that form one of the PH domains of phospholipase C are separated by an insert of about 300 residues containing one SH3 and two SH2 domains (see Figure 5.11, page 151).

Although some bacteria have acquired domains from eukaryotic species by horizontal gene transfer, it is through the duplication, insertion, and deletion of genes that proteins principally acquire or lose domains and by which new combinations are generated. The process is termed domain shuffling. One way in which this occurs is through the shuffling of exons (again through insertion, duplication, or deletion). For this to work effectively there should be a correspondence between exons and domains; for example, a single exon coding for a single complete domain. Insertion of new exons requires recombination within introns and this has to be achieved without creating frameshifts that would cause misreading of the downstream sequence. To avoid disruption of the reading frame, only symmetrical exons can be inserted into introns with matching phase. Although this might seem unlikely, analysis of the human and other genomes indicates that there has been extensive exon shuffling in the evolution of eukaryotic genomes.2

After the emergence of the eukaryotes, development of domains accelerated and their inclusion into proteins by exon shuffling led to a rapid increase

in diversity. In addition, a particular feature of multicellular organisms is the need for communication between cells. This became possible with the emergence of molecules having extracellular domains able to take part in intercellular signalling processes. All of this may have contributed to the

formation of complex life forms during the period of the metazoan radiation.

In summary, the acquisition of new modules has expanded protein function in a combinatorial fashion, generating diversity without requiring new genes. It may help to account for the relatively small number of genes possessed by such complex species as humans. Indeed, the human genome seems to be the most diverse in terms of domain combinations.

Sequence homology and the acquisition of function

Some types of structural domain show high levels of sequence homology while others are very limited in this respect. Where sequence homology exists, it has tended to preserve basic properties, such as the core structure; where the sequence is variable, as at the loops between secondary structural elements, it can allow variations of function. Thus, differences in sequence may reflect adaptations to meet special requirements or they may just be the result of mutations at non-critical residues. An eroded sequence homology hampers the analysis of genetic history and when structural data are not available, it hinders the recognition of domains.

Intron phase. Introns that lie between complete codons have phase 0. Those that divide codons between the first and second nucleotides are phase 1 and those between the second

and third nucleotides are phase 2.

Symmetrical exons are bordered by introns with phases 0–0, 1–1, or 2–2.

If frameshifts are to be avoided, these exons can only be introduced into introns of phase 0, 1, or 2, respectively.3

The human genome appears to possess 20 000–25 000 genes (www.ensembl.org), far fewer than originally anticipated.