1052 Introns and Exons
Introns and Exons
A Stoltzfus
doi: 10.1006/rwgn.2001.0708
An intron (or `intervening sequence') is a segment of
RNA excised from a gene transcript, with concomi-
tant ligation of flanking segments called `exons.' This
process of excision and ligation, known as `splicing,'
is one of several posttranscriptional processing steps
that may occur prior to translation. Although `intron,'
in the strict sense, refers only to segments excised
from RNA (and, by extension, the DNA segments
that encode them), there exist developmental analogs
of introns that are excised from DNA (the ciliate
IES elements) or from protein (the printrons or
inteins).
Diversity and Distribution
Introns of some type are found in every kingdom of
cellular life, and also in viruses, bacteriophages, and
plasmids. Different types of introns have different
splicing mechanisms and distinctive patterns of dis-
tribution with respect to gene families, subcellular
compartments, and taxonomic groups (e.g., protein-
spliced tRNA introns are known only from tRNA
genes in archaebacterial genomes or eukaryotic nuclear
genomes). A single gene may have multiple introns
and, rarely, introns of multiple types (e.g., some fungal
mitochondrial genes have both group I and group II
introns).
The most familiar introns are the `spliceosomal'
introns, which are excised by a ribonucleoprotein
`spliceosome,' and which typically have the sequence
GU...AG. Spliceosomal introns are known only from
genes in the eukaryotic nucleus (or nucleomorph) and
in eukaryotic viruses. They range in length from less
than 20 nt (nucleotides) to over 200 kilo-nt, while
exons range in length from less than 10 nt to over 3
kilo-nt. The mean density of introns varies widely,
from over 4 introns per kilo-nt of protein-coding
sequence in the most intron-dense nuclear genomes
(including those of vertebrates and vascular plants), to
0.04 in the yeast Saccharomyces cerevisiae.
Group I and group II introns are collectively
known as `self-splicing' introns, because the intron
RNA plays a primary role in the biochemistry of
splicing, in some cases being sufficient for splicing in
vitro. Group I introns are the most broadly distri-
buted mobile elements known, being found in the
genomes of eubacteria and their phages, as well as in
the nuclear, mitochondrial, and chloroplast genomes
of eukaryotes. Group II introns are known in eukary-
otic organellar (but not nuclear) genomes, as well as
in eubacterial chromosomes and plasmids. Though
common in some organellar genomes, self-splicing
introns are extremely rare elsewhere, and seem to be
entirely absent from most prokaryotic genomes as
well as many eukaryotic nuclear genomes.
Role in Gene Expression
In most cases, introns appear to be dispensable.
Introns can be removed entirely from mitochondria
of S. cerevisiae without obvious ill effect. Neverthe-
less, in a variety of cases, introns and splicing figure
importantly in development. The delay caused by the
transcription and splicing of a gene with many long
introns can be important (e.g., the knrl gene of Dros-
ophila). The intron may contain within itself some
other feature: a DNA regulatory site (e.g., a promoter
or enhancer), a structural RNA (e.g., intron-encoded
snoRNAs in eukaryotic nuclear genomes), or a pro-
tein-coding region (e.g., intron-encoded maturases in
organellar group I and II introns and homing endonu-
cleases in bacteriophage introns). Splicing may join
parts of two different RNA transcripts, a process
known as `trans-splicing' that is common in trypano-
somes but rare or absent in most other organisms.
Finally, the pattern of splicing of a single transcript
may be variable, such that different mRNAs, and
different protein products, are produced from the
same pre-mRNA. Regulation of such `alternative
splicing' schemes plays a crucial role in sex determin-
ation in Drosophila. The frequency and importance of
alternative splicing in most species is not well under-
stood.
Mutation and Evolution
Introns are passively subject to the same mutational
lesions that affect other genomic sequences; in some
cases they contribute actively to the mutational pro-
cess as mobile elements. Nucleotide substitutions that
alter splicing have been implicated in many heritable
diseases in humans. Such changes usually map to
within a few nt of a splice junction. Over evolutionary
time-scales, the internal sequences of spliceosomal
introns diverge rapidly (by nucleotide substitutions
as well as by short insertions and deletions), presum-
ably because the demands of splicing impose no con-
straint on most internal sites. By contrast, group I and
II introns evolve more slowly, and are densely packed
with sequences that participate in splicing and mo-
bility.
Rearrangement mutations involving introns also
occur, sometimes based on recombination between

repetitive elements within introns. In animal genomes,
intron-mediated rearrangements have contributed
importantly to the evolution of novel chimaeric
genes by so-called `exon shuffling.' On the scale of
millions to hundreds of millions of years, homologous
genes may diverge by loss and gain of introns. Loss of
an intron may occur by way of reverse transcription
and recombinational reincorporation of a spliced gene
product. Insertion of introns by transposition has
been observed experimentally for group I, group II,
and spliceosomal introns. For group I and II introns,
`homing' to (intronless) allelic sites is also observed.
See also: Eukaryotic Genes; Pre-mRNA Splicing
Invariants, Phylogenetic
W Fitch
Copyright ß 2001 Academic Press
doi: 10.1006/rwgn.2001.0710
Phylogenetic invariants is a method first proposed
by Lake (1987). The `invariants' derive from the fact
that the addition and subtraction of the numbers of
certain nucleotide distribution patterns are expected
to remain constant (at zero) for all incorrect phylo-
genies. And thus can be used to distinguish among
alternative phylogenetic trees. It is a property that is
used on nucleotide sequences taken four at a time. For
example, suppose that we had four such sequences
that are homologously aligned from left to right, one
under the other:
...AGA...
...AGT...
...C T T...
...C TA...
so that for any position in the alignment the four
nucleotides produce a (vertical) pattern such as
AACC. This might suggest that the first two se-
quences are sister sequences meaning that they are
more closely related to each other than either of them
is to the second two sequences (see Figure 1A). There
1A 4C 1A 4A 1A
A
A C A A
C
2A 3C 2C 3C 2C
A B
Figure 1
Invariants, Phylogenetic 1053
are 256 possible patterns but some of them carry the
same information. For example, the same relationship
would be inferred if the pattern were GGTT. The
method restricts itself to positions that have exactly
two purines (A and/or G) and two pyrimidines (C
and/or T) in their pattern as all the examples used
here do. Their relationship is shown by the tree in
Figure 1A where the arrowhead indicates that only a
single transversion mutation is required to explain the
observed nucleotides at the tips of this tree. (A trans-
version is the historical change from (or to) a purine
to (or from) a pyrimidine; all other interchanges are
called transitions.)
On the other hand, a pattern such as ACCA would
suggest that sequences 1 and 4 were sisters rather than
sequences 1 and 2 (see Figure 1B). The two relation-
ships (trees) cannot both be true, but if sequences 1
and 2 really are the true sister sequences, then this
third pattern can only have arisen by virtue of two
transversions having occurred during the history of
these sequences (see Figure 1C).
However, we can estimate how often the mislead-
ing case in Figure 1C arises. Note that in Figure 1D
we have shown only three of the four nucleotides in the
pattern. What could the fourth nucleotide be? As we
only consider those patterns with two purines and two
pyrimidines, there must be a pyrimidine. Which one?
If we assume that there is no bias as to which nucleo-
tide the mutation is to, then it can be either C (as in
Figure 1C) or T (as in Figure 1E) with equal prob-
ability. But that means that, for the wrong tree, the
number of occurrences of a pattern like that in Figure
1C should be the same as the number for the pattern
like that in Figure 1E. Hence, subtracting those two
numbers should give an number not statistically differ-
ent from zero for the two tree structures that are wrong.
(The third possible tree is for the pattern ACAC
which suggests that sequences 1 and 3 are sisters.)
There are more details to the method but the pre-
ceding gives the spirit of the method. It is a method
that is guaranteed to give the correct answer given
sufficient lengths of the sequences being compared.
This virtue, however, is more than offset by the answer
to the question of how long the sequences must be to
get that correct answer. It turns out that the sequences
4A 1A 4A 1A 4A
A A A A
3C 2C 3? 2C 3T
C D E