Location of domains in globular proteins

Abstract

Although it has become widely accepted that domains are the basic units of structure, function and evolution in proteins¹ and it is thought that proteins with complex functions evolve by fusion of genes coding for individual domains^2,3, the domains are not uniformly defined. Most commonly, domains are simply the compact and more or less loosely connected areas^1,4 apparent from a visual inspection of protein models; to avoid subjectivity and ambiguities inherent in visual inspection, certain computer algorithms for location of these ‘structural’ domains have recently been proposed^5–7. An alternative interpretation is that domains are stable protein fragments found in biochemical experiments^8–17. I regard them as ‘globular fragments’ which may refold autonomously and carry specific functions¹, and I propose here a method for location of these globular fragments based on surface area¹⁸ measurements. Applied to several proteins the globular fragments found often coincide with structural domains or are contained within them. In particular the globular fragments found in globins correlate with the two structural domains proposed previously^4,19, and do not correlate with the three coding sequences separated by introns in the haemoglobin genes^20–22.