Exon Shuffling
 
When intervening sequences were discovered to split eukaryotic genes into segments coding for protein, the intervening sequences were named introns and the coding regions, exons (1). Since then, heated discussion has ensued concerning the evolutionary origin and biological significance of introns in eukaryotes, because almost all of the prokaryotic genes lack introns, and the introns have no known role.
Gilbert (1) proposed a scenario in which introns simply connect neighboring exons to each other in the genome; consequently, genetic recombination would not harm the coding portions of genes if it took place within the intron regions. Such recombination within introns would facilitate the shuffling of exons, to create new exon combinations and lead to the emergence of new genes with new functions. This is the “exon shuffling” theory.
Exon shuffling is related to the present controversy regarding the “early” versus “late” intron theories. The early-intron theory maintains that introns existed in the ancestor genomes of prokaryotes and eukaryotes, but that all those in prokaryotes were deleted by some unknown mechanism. On the other hand, the late-intron theory contends that introns could have been inserted, similar to transposable elements, into eukaryotic genomes quite recently, at least after the divergence of eukaryotes and prokaryotes. This theory is based on the observation that there are many genes in which the locations of introns are not conserved among vertebrates, invertebrates, and plants (2).
If the early-intron theory is correct, exon shuffling probably had a significant role in the creation of new genes. If the late-intron theory is right, however, exon shuffling might have been significant only after the introns were inserted. The controversy remains to be resolved.
 The domain shuffling theory is similar to the exon shuffling theory, but in sharp contrast, maintains that the unit of shuffling during evolution is a functional protein domain, not an exon. The functional domain does not necessarily correspond to the exon. Although there are some cases where it does, most functional domains consist of more than one exon or occupy only part of a large exon. For example, the Kringle domain is known to have been shuffled as a unit during evolution. The Kringle domain is a characteristic supersecondary structure frequently found in the serine proteinases involved in the blood clotting system. Even though the Kringle domain is, in most cases, split into three exons by two introns, those found in various proteins are almost always complete forms and never found as parts corresponding to the exons. At present, however, there is no known genetic mechanism for facilitating domain shuffling, and domain shuffling seems to have taken place in prokaryotes as well (3).
 
References
1. W. Gilbert (1978) Nature 271, 501. 
2. J. D. Palmer and J. M. Logsdin (1991) Curr. Opin. Gen. Devel. 1, 565–570. 
3. T. Gojobori and K. Ikeo (1994) Phil. Trans. Roy. Soc. Lond. B 344, 411–415.