The highly interrupted structure of eukaryotic genes  suggests a picture of the eukaryotic genome as a sea of introns (mostly but not  exclusively unique in sequence), in which islands of exons (sometimes very  short) are strung out in individual archipelagoes that constitute genes.
A test of the models is to ask whether the difference  between eukaryotic and prokaryotic genes can be accounted for by the acquisition  of introns in the eukaryotes or by the loss of introns from the  prokaryotes.
The introns early model suggests that the mosaic structure  of genes is a remnant of an ancient approach to the reconstruction of genes to  make novel proteins. Suppose that an early cell had a number of separate  protein-coding sequences. One aspect of its evolution is likely to have been the  reorganization and juxtaposition of different polypeptide units to build up new  proteins.
If the protein-coding unit must be a continuous series of  codons, every such reconstruction would require a precise recombination of DNA  to place the two protein-coding units in register, end to end in the same  reading frame. Furthermore, if this combination is not successful, the cell has  been damaged, because it has lost the original protein-coding units.
But if an approximate recombination of DNA could place the  two protein-coding units within the same transcription unit, splicing patterns  could be tried out at the level of RNA to combine the two proteins into a single  polypeptide chain. And if these combinations are not successful, the original  protein-coding units remain available for further trials. Such an approach  essentially allows the cell to try out controlled deletions in RNA without  suffering the damaging instability that could occur from applying this procedure  to DNA. This argument is supported by the fact that we can find related exons in  different genes, as though the gene had been assembled by mixing and matching  exons (see 2.10 Some exons can be  equated with protein functions).

Figure 2.21 illustrates the outcome  when a random sequence that includes an exon is translocated to a new position  in the genome. Exons are very small relative to introns, so it is likely that  the exon will find itself within an intron. Because only the sequences at the  exon-intron junctions are required for splicing, the exon is likely to be  flanked by functional 3
![]()
 and  5
![]()
 splice junctions,  respectively. Because splicing junctions are recognized in pairs, the 5
![]()
 splicing junction of the original  intron is likely to interact with the 3
![]()
 splicing junction introduced by the new exon, instead  of with its original partner. Similarly, the 5
![]()
 splicing junction of the new exon will interact with  the 3
![]()
 splicing junction of  the original intron. The result is to insert the new exon into the RNA product  between the original two exons. So long as the new exon is in the same coding  frame as the original exons, a new protein sequence will be produced. This type  of event could have been responsible for generating new combinations of exons  during evolution. Note that the principle of this type of event is mimicked by  the technique of exon trapping that is used to screen for functional exons (see  Figure 2.12).
Alternative forms of genes for rRNA and tRNA are sometimes  found, with and without introns. In the case of the tRNAs, where all the  molecules conform to the same general structure, it seems unlikely that  evolution brought together the two regions of the gene. After all, the different  regions are involved in the base pairing that gives significance to the  structure. So here it must be that the introns were inserted into continuous  genes.
Organelle genomes provide some striking connections between  the prokaryotic and eukaryotic worlds. Because of many general similarities  between mitochondria or chloroplasts and bacteria, it seems likely that the  organelles originated by an endosymbiosis in which an early bacterial  prototype was inserted into eukaryotic cytoplasm. Yet in contrast with the  resemblances with bacteria?for example, as seen in  protein or RNA synthesis?some organelle genes  possess introns, and therefore resemble eukaryotic nuclear genes.
Introns are found in several chloroplast genes, including  some that have homologies with genes of E. coli. This suggests that the  endosymbiotic event occurred before introns were lost from the prokaryotic line.  If a suitable gene can be found, it may therefore be possible to trace gene  lineage back to the period when endosymbiosis occurred.
The mitochondrial genome presents a particularly striking  case. The genes of yeast and mammalian mitochondria code for virtually identical  mitochondrial proteins, in spite of a considerable difference in gene  organization. Vertebrate mitochondrial genomes are very small, with an extremely  compact organization of continuous genes, whereas yeast mitochondrial genomes  are larger and have some complex interrupted genes. Which is the ancestral form?  The yeast mitochondrial introns (and certain other introns) can have the  property of mobility?they are self-contained  sequences that can splice out of the RNA and insert DNA copies elsewhere?which suggests that they may have arisen by insertions  into the genome (see 26.5 Some group  I introns code for endonucleases that sponsor mobility and 26.6 Some group II introns code for  reverse transcriptases).