Origin of Spliceosomal Introns and Alternative Splicing - CSH ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
Origin of Spliceosomal Introns and
Alternative Splicing
Manuel Irimia1 and Scott William Roy2
1
The Donnelly Centre, University of Toronto, Toronto, Ontario M5S3E1, Canada
2
Department of Biology, San Francisco State University, San Francisco, California 94132
Correspondence: mirimia@gmail.com
In this work we review the current knowledge on the prehistory, origins, and evolution of
spliceosomal introns. First, we briefly outline the major features of the different types of
introns, with particular emphasis on the nonspliceosomal self-splicing group II introns,
which are widely thought to be the ancestors of spliceosomal introns. Next, we discuss
the main scenarios proposed for the origin and proliferation of spliceosomal introns, an
event intimately linked to eukaryogenesis. We then summarize the evidence that suggests
that the last eukaryotic common ancestor (LECA) had remarkably high intron densities and
many associated characteristics resembling modern intron-rich genomes. From this intron-
rich LECA, the different eukaryotic lineages have taken very distinct evolutionary paths
leading to profoundly diverged modern genome structures. Finally, we discuss the origins
of alternative splicing and the qualitative differences in alternative splicing forms and func-
tions across lineages.
SURPRISES AND MYSTERIES OF INTRONS had to be removed to generate intact protein-
AND INTRON EVOLUTION coding messenger RNAs. The initial findings in
viruses (Berget et al. 1977; Chow et al. 1977)
ith mid-20th century breakthroughs, mo- were soon extended to many cellular genes.
W lecular cell biology finally seemed to obey
a relatively simple logic. Genetic information
With the advent of large-scale sequencing proj-
ects, it became clear that one kind of intron
was encoded in DNA genes (Avery et al. 1944; (the spliceosomal introns), as well as the cellu-
Watson and Crick 1953), which were tran- lar machinery that removes them (the spliceo-
scribed into RNA and subsequently translated some), are ubiquitous in eukaryotic genomes.
into functional proteins (Crick 1958). However, For example, the average human transcript
a most unexpected finding “interrupted” this contains 9 introns, totaling several hundred
logic. The coding information of DNA genes thousand introns across the genome and com-
was sometimes broken into pieces separated prising a quarter of the DNA content of each
by sequences whose sole apparent purpose was cell (Lander et al. 2001; Venter et al. 2001).
to generate an extra RNA sequence that then Moreover, functions for some introns began to
Editors: Patrick J. Keeling and Eugene V. Koonin
Additional Perspectives on The Origin and Evolution of Eukaryotes available at www.cshperspectives.org
Copyright # 2014 Cold Spring Harbor Laboratory Press; all rights reserved; doi: 10.1101/cshperspect.a016071
Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071
1
Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
M. Irimia and S.W. Roy
emerge. In particular, by regulating the remov- of eukaryotes (and thus are quite ancient even if
al of introns and subsequent rejoining of exons not primordial), and originated from preexist-
(so-called intron splicing), eukaryotic genes can ing self-splicing introns, which were likely pres-
generate multiple transcripts, vastly expanding ent at very early stages of life.
molecular diversity. First hypothesized by Gil-
bert (1978), this process, known as alternative
TYPES OF LARIAT INTRONS: SELF-SPLICING
splicing (AS), appears to be widespread in eu-
INTRONS AND THE PREHISTORY OF THE
karyotes, seemingly reaching its apex in mam-
SPLICEOSOMAL SYSTEM
mals (Barbosa-Morais et al. 2012), in which 95%
of multiexon genes undergo AS (Pan et al. 2008; Spliceosomal introns are just one of the four
Wang et al. 2008). major classes of introns found in nature, to-
Nearly four decades after the discovery of gether with group I and group II self-splicing
introns, many questions remain unanswered. introns, and tRNA introns. Intron types are de-
The most fascinating questions are still the fined based on various structural and mechanis-
most fundamental ones: Why do introns exist? tic features, and they have distinct phylogenetic
When and how did they arise? These questions distributions. In this section we discuss different
were first formulated as part of the exciting and intron types and compare crucial aspects of the
contentious “introns early/late” debate. Introns evolutionarily related spliceosomal and group II
early held that introns were very ancient struc- introns, collectively known as lariat introns.
tures that predated cellular life, and that mod-
ern organisms with few or no introns had lost
Spliceosomal Introns
them secondarily (in particular, all prokaryotic
genomes contain either no introns or only a few Spliceosomal introns are found in all studied
nonspliceosomal introns) (see below) (Darnell eukaryotic nuclear genomes (with two possible
1978; Gilbert 1987; Long et al. 1995; de Souza exceptions) (Andersson et al. 2007; Lane et al.
et al. 1996). The related “introns first” hypoth- 2007) and only in eukaryotic nuclear genomes,
esis holds that exons emerged from noncoding although the total number of introns in each
regions between RNA genes in the RNA world species varies by orders of magnitude. They
(Poole et al. 1998; Penny et al. 2009). On the are characterized by their mechanism of splic-
other hand, introns late counters that spliceo- ing, which is catalyzed by the spliceosome, a
somal introns arose later, at some point dur- complex ribonucleoprotein machinery formed
ing eukaryotic evolution (Cavalier-Smith 1985, by five small RNAs (the U1, U2, U4, U5, and U6
1991; Dibb and Newman 1989; Stoltzfus 1994; snRNAs) and more than 200 proteins (Wahl
Logsdon et al. 1995; Logsdon 1998). This debate et al. 2009). Although there are also marked
spanned over 20 years despite (or perhaps ow- lineage-specific differences, the sequence struc-
ing to) the scarcity of directly relevant data, ture of most spliceosomal introns consists of a
finding resolution only with the availability of short 50 splice site (50 ss) boundary, a minimal
many whole genome sequences. Although ad- AG dinucleotide 30 splice site (30 ss) boundary, a
herents to both perspectives remain, the introns catalytic adenosine (the branch point [BP]),
early perspective has been weakened by the find- and a polypyrimidine tract between the BP
ing of low or zero intron density in all prokary- and the 30 ss (Fig. 1). These sequences are recog-
otic lineages, and the gradual weakening of (and nized and bound by the core components of the
emergence of potential alternative explanations spliceosome and are crucial for the splicing
for) statistical signals suggestive of early introns. reaction (Ruskin and Green 1985; Chiara and
However, although, formally, the data tipped Reed 1995; Umen and Guthrie 1995; Chiara
the scales in favor of introns late, the current et al. 1997; Du and Rosbash 2002). The rest of
consensus may be seen as a mixture of the two the intronic sequence, which in some instances
perspectives (Koonin 2006): spliceosomal in- in vertebrates may reach up to a million nucle-
trons appeared abruptly at the time of the origin otides, is generally evolutionarily unconstrained
2 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
Origin of Introns and Alternative Splicing
A
2
Bits
1
0
Major/U2 (H. sapiens)
B
2
Bits
1
0
Major/U2 (S. cerevisiae)
C EXON INTRON EXON INTRON
2
Bits
1
2–5 nt
0
Minor/U12 (H. sapiens)
Figure 1. Consensus sequences of spliceosomal intron core splicing signals for (A) human major/U2 introns,
(B) yeast major/U2 introns, and (C ) human minor/U12 introns. The branch point (BP) adenosine is indicated
by a red arrow.
and highly variable (Irimia and Roy 2008b). the major or U2 subtype described above). The
The splicing process consists of two consecu- mechanism of splicing is nearly identical, but
tive transesterification reactions, catalyzed by the minor spliceosome is assembled around
the snRNAs of the spliceosome (Domdey et al. four distinct snRNAs—U11, U12, U4atac, and
1984; Padgett et al. 1984; Ruskin et al. 1984; Lin U6atac—and a few specific accessory proteins
et al. 1985). First, the BP adenosine performs a (Hall and Padgett 1996; Tarn and Steitz 1996;
nucleophilic attack on the 50 end of the intron, Will et al. 1999); the rest of the protein machin-
resulting in the cleavage of the 50 nucleotide of ery and the U5 snRNA are shared with the U2
the intron (generally the “G” in the GT) from spliceosome. Although the sequence structure
the upstream exon. Second, the intron is fully of both subtypes is similar overall, U12 introns
excised from the mRNA by another nucleo- have a different and stricter consensus sequence
philic attack by the upstream exon, which is at the 50 ss and near the BP, and the distance be-
ligated to the 50 of the downstream exon, gen- tween the BP and the 30 ss is highly constrained
erating a precise exon –exon junction and liber- (typically 10 – 15 nucleotides) (Fig. 1C) (Jack-
ating the intronic sequence in a characteristic son 1991; Hall and Padgett 1994). Although
lariat structure. the much more limited phylogenetic distribu-
Interestingly, there is another subtype of tion of U12 introns initially suggested they
spliceosomal introns in eukaryotes, the so- might have arisen more recently in evolution,
called minor or U12 introns (in contrast to U12 introns and spliceosomal components have
Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071 3Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
M. Irimia and S.W. Roy
since been discovered in diverse eukaryotic lin- liferated from a single intron copy, providing
eages, implying the existence of the U12 sys- a common splicing apparatus, and allowing
tem in the last eukaryotic common ancestor some IEP copies to degenerate (Dai and Zim-
(LECA) (Russell et al. 2006; Dávila López merly 2003; Meng et al. 2005). In addition to the
et al. 2008). role in assisting intron splicing, IEPs are multi-
functional proteins that also have reverse tran-
scriptase activity, which allows group II introns
Group II Self-Splicing Introns
to reverse transcribe into DNA, conferring them
Found in bacterial genomes and in chloroplast their nature of mobile genetic elements (Cous-
and mitochondrial genomes of widely diverged ineau et al. 2000; Lambowitz and Zimmerly
eukaryotes (Ferat and Michel 1993; Lambowitz 2011).
and Zimmerly 2011; Candales et al. 2012), There are at least three major subgroups
group II self-splicing introns are believed to pre- of group II introns, with further subdivisions,
date the origin of eukaryotes, perhaps even pre- distributed among eight phylogenetically sup-
ceding the origin of cellular life (Koonin 2006). ported lineages (Zimmerly et al. 2001; Simon
They have been identified, always in low num- et al. 2008). Each group is characterized by spe-
bers, in around a quarter of the sequenced cific RNA secondary structures, and by pecu-
bacterial genomes (Lambowitz and Zimmerly liarities of splicing and mobility (Lambowitz
2011). On the other hand, they are rare in ar- and Zimmerly 2011). In particular, although
chaea (only described, so far, in two related spe- bacterial introns are usually highly active and
cies) (Dai and Zimmerly 2003; Rest and Mindell functionally complete, introns in eukaryotic
2003; Doose et al. 2013). The mechanism of organelles often lack important secondary do-
splicing of group II introns is very similar to mains and/or encode degenerate IEPs (Coper-
that of spliceosomal introns—and likely evolu- tino and Hallick 1993; Michel and Ferat 1995;
tionarily related—involving two transesterifi- Bonen 2008; Barkan 2009). Importantly, the
cation reactions and the release of an excised splicing of these degenerate introns likely relies
intron lariat. However, in this case, these reac- on the action of trans-acting RNAs from other
tions are catalyzed by the intronic RNA itself, introns and/or host-encoded proteins (Jarrell
which has ribozymatic activity (Peebles et al. et al. 1988; Goldschmidt-Clermont et al. 1991;
1986; Schmelzer and Schweyen 1986; van der Suchy and Schmelzer 1991; Lambowitz and
Veen et al. 1986). Accordingly, group II intronic Zimmerly 2011), potentially mirroring inter-
RNA sequences show very complex and con- mediate steps hypothesized for the origin of
served secondary structures that span 400– 800 spliceosomal introns.
nucleotides (Lambowitz and Zimmerly 2011).
Group II intron sequences are organized in six
Other Types of Introns
main domains (DI– DVI) consisting of various
functional “loops” and “bulges,” which are the (i) Group I introns: present in representatives of
basis of the ribozymatic activity (Michel and most eukaryotic supergroups, either in nuclear
Ferat 1995; Qin and Pyle 1998; Lambowitz ribosomal RNA genes or in mitochondrial and/
and Zimmerly 2011). Although group II introns or plastid genes, as well as in some bacteria and
are capable of self-splicing in vitro, efficient viruses (Haugen et al. 2005). They also catalyze
splicing in vivo requires specific proteins, which their own splicing reaction by a very different
are usually encoded by the intron, but can also mechanism, utilizing an exogenous guanosine
be recruited from the host (Solem et al. 2009). (exoG) as a cofactor that does not release a lariat
The protein encoded by the intron (the “intron- (Cech et al. 1981; Bass and Cech 1984; van der
encoded protein,” IEP) usually acts in cis, assist- Horst and Tabak 1985).
ing the splicing only of its own intron. However, (ii) Introns in nuclear and archaeal transfer
in some cases, an IEP can evolve the ability to RNA genes: present in tRNA genes of eukary-
aid splicing of multiple related introns that pro- otic nuclear or archaeal genomes (Marck and
4 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
Origin of Introns and Alternative Splicing
Grosjean 2003; Randau and Söll 2008), and ilar boundary sequences (GT-AY in group II in-
also in some coding genes in archaea (Yokobori trons and usually GT-AG in spliceosomal in-
et al. 2009; Doose et al. 2013). They do not share trons, although some U12 introns are AT-AC)
functional or structural similarities with the (Lambowitz and Zimmerly 2011), and there
other types of introns, as they are typically are striking structural similarities between key
very short and their splicing is fully catalyzed regions of group II intron domains and spli-
by protein enzymes (rather than ribozymes) ceosomal snRNAs (Lambowitz and Zimmerly
(Randau and Söll 2008). 2011). These include at least (i) domain DV
and U6 snRNA, with divalent metal-ion binding
sites involved in catalysis and similar base-pair-
ORIGIN AND ESTABLISHMENT OF THE
ing interactions (Jarrell et al. 1988; Peebles et al.
SPLICEOSOMAL SYSTEM DURING
1995; Yu et al. 1995; Abramovitz et al. 1996;
EUKARYOGENESIS
Konforti et al. 1998; Yean et al. 2000; Shukla
In this section, we discuss the major steps lead- and Padgett 2002), further supported by crystal
ing to the origin and establishment of the structure (Toor et al. 2008; Keating et al. 2010);
spliceosomal system in eukaryotes. Despite the (ii) ID3 subdomain and d –d0 motifs and U5
persistence of disagreement on some important snRNA stem loop, involved in the recognition
aspects, there is a general consensus about how of 50 and 30 exons (Hetzer et al. 1997); and (iii)
this process may have unfolded (Fig. 2). Accord- DVI and the U2-intron pairing that include the
ing to this general model, spliceosomal introns BP adenosine (Schmelzer and Schweyen 1986;
evolved from invading group II introns, perhaps Parker et al. 1987; Li et al. 2011). In addition, it
derived from the early mitochondrion (thought has also been shown that extracts of snRNAs can
to be descended from an engulfed member of catalyze both splicing reactions without proteins
the a-proteobacteria, whose modern members in vitro (Valadkhan et al. 2007, 2009), in a sim-
contain group II introns). For some reason(s), ilar manner to complete group II introns.
these introns then proliferated to an unprece- Whereas it is theoretically possible that
dented level in the host genome. Over time, the group II introns evolved from spliceosomal in-
self-splicing activities of these many intron cop- trons (or that both share a distinct common
ies degenerated, which was associated with the ancestor [Vesteg et al. 2012]), it seems much
increase of trans-encoded RNAs and proteins more likely that group II introns gave rise to
that promoted efficient intron splicing, setting spliceosomal introns (Cech 1986). The presence
the basis of the protospliceosomal machinery, of spliceosomal introns in all extant eukaryotic
and further releasing selective pressure on cis- supergroups indicates that this transformation
intronic splicing elements. As this protospliceo- occurred before LECA. The most favored hy-
somal machinery recruited more proteins and pothesis is that spliceosomal introns originated
became more efficient, introns became increas- from a-proteobacterial group II introns trans-
ingly reliant on the emerging spliceosome for ferred from the protomitochondria to the host
proper splicing. genome soon after the endosymbiotic event
(Cavalier-Smith 1991), along with other parts
of the protomitochondrial genome (Thorsness
Transfer of Group II Introns to the Host
and Weber 1996; Adams and Palmer 2003). Sup-
Genome
porting this view, complete and mobile group
Structural and functional evidence suggests that II introns are relatively common in modern
spliceosomal and group II self-splicing introns a-proteobacteria, but (nearly) absent in the
are evolutionarily related. Both types of introns archaeal domain (Lambowitz and Zimmerly
are spliced through a similar two-step catalytic 2011), to which the preendosymbiotic protoeu-
reaction that relies on an endogenous adenosine karyote is thought to be more closely related
(the BP), and releases the excised intron as a (see Guy et al. 2014). In addition, few intron
lariat structure. The two intron types have sim- positions are shared between ancient paralogs
Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071 5Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
M. Irimia and S.W. Roy
α-Proteobacteria with
Endosymbiosis group II introns
and
Intron transfer
Intron proliferation Sexual host? Low Ne? Nucleus?
Establishment of Intronic sequence degeneration
the spliceosomal Splicing trans-complementation
system
Protein recruitment
Intron-rich
Weak splice site signals
LECA Complex spliceosomes
IR-dominated AS
Intense intron loss
Evolutionary history of
Intense intron gain
spliceosomal introns
Figure 2. Steps leading to the origin and establishment of the complex spliceosomal system of LECA during
eukaryogenesis (see text).
(i.e., gene duplicates originated before eukaryo- from the endosymbiont to the host genome,
genesis), suggesting that at least most introns at least one event of massive intron prolifera-
arose during or after the advent of eukaryotes tion took place (Fig. 2). From a number perhaps
(Sverdlov et al. 2007). similar to modern a-proteobacteria—usual-
ly ,30 introns per genome (Lambowitz and
Zimmerly 2004)—the ancestors of LECA ex-
Unprecedented Group II Intron Proliferation
perienced an expansion of introns that populat-
in the Host Genome
ed their genomes with thousands of elements,
It is also widely accepted that, following the estimated at 2 introns per kbp of coding
transfer of ( presumably few) group II introns sequence, and hypothesized to account for
6 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
Origin of Introns and Alternative Splicing
.70% of the ancestral protoeukaryotic genome 2006) invokes a very different causality, drawing
(Koonin 2009). This proliferation is orders of on the classic arguments on transposable ele-
magnitude larger than any known expansion ments by Hickey (1982). Hickey showed that
in prokaryotes, suggesting exceptional circum- frequent sexual reproduction is crucial to the
stances. Although most researchers agree on the success of transposable elements: whereas with
evolutionary singularity of this event, the na- no or infrequent sex, the fitness costs of aggres-
ture of these circumstances is strongly debated. sively propagating elements will doom them
Three (nonmutually exclusive) major explana- to extinction, in the presence of frequent sex,
tions have been proposed: (i) the archaeal host even highly deleterious elements can spread
lacked unknown defense mechanism against re- through the genome. This suggests the possibil-
troelement multiplication (Koonin 2006); (ii) ity that the advent of frequent sexual reproduc-
an extremely low effective population size (Ne) tion could have brought an unprecedented
after an evolutionary bottleneck allowed the spread of type II introns. Consistent with this
protoeukaryote to fix thousands of intron cop- notion, meiosis and sexual reproduction appear
ies by simple genetic drift, despite the selective to have been well established by LECA (Ramesh
disadvantage (Koonin 2006); and (iii) the oc- et al. 2005), although, crucially, the order of
currence of frequent meiosis or meiosislike sex- appearance in the evolution of frequent sexual
ual reproduction in the eukaryotic ancestor al- reproduction and high intron density remains
lowed large-scale spreading of introns at the unknown.
expense of the host’s fitness (Poole 2006), as
previously proposed by Hickey more generally
Need and Benefits of
for any class of mobile genetic element (Hickey
Trans-Complementation:
1982).
Emergence of the Spliceosome
The discovery of two archaea species with
a few (4 and 21) group II introns, likely later- Whatever the cause, the effect of this intron
ally transferred from bacteria (Dai and Zim- propagation was a genome newly riddled with
merly 2003; Rest and Mindell 2003), for which thousands of elements interrupting most cod-
no dramatic proliferation has occurred, argues ing genes that required removal from pre-
that the first explanation is insufficient (the lack mRNAs. Although intact group II introns are
of an unknown defense system against group II capable of self-splicing and thus may be (near-
intron expansion in this lineage). The second ly) functionally neutral, the need to maintain
proposed explanation, which can be integrated a high number of constrained functional se-
within a more general hypothesis of the origin quences and tertiary structures in every intron
and evolution of eukaryotes and their genomes implies strong mutational pressure. Although
(Lynch and Conery 2003; Lynch 2006), current- the number of intronic and exonic sites re-
ly has considerable support. This hypothesis quired for splicing of spliceosomal introns has
posits that many eukaryotic genomic features been estimated to be around 20– 40 per intron
initially proliferated owing to selection being (Lynch 2002, 2006), the number of potentially
too inefficient to purge them from the popula- constrained sites in a group II intron seems to be
tion, which situation might arise if the effective at least one order of magnitude higher (Lambo-
population size of these ancestral organisms was witz and Zimmerly 2011). Therefore, a spliceo-
very small (a conjecture for which direct tests somal system that made use of only a few trans-
are very difficult). The presence of these slight- acting factors to replace hundreds of thousands
ly deleterious elements is nonetheless argued of constrained sequences in cis may have signif-
to have driven the emergence of profound and icantly relieved much of the mutational pres-
defining eukaryotic characteristics, such as sure associated with an extremely high number
the nucleus, as a defense mechanism (Koonin of introns.
2006; López-Garcı́a and Moreira 2006; Martin There is also compelling mechanistic evi-
and Koonin 2006). The third hypothesis (Poole dence that the transition from a system of self-
Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071 7Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
M. Irimia and S.W. Roy
spliced introns to one based on trans-comple- distribution ( present in all eukaryotic groups,
mentation may have evolved gradually. The first and thus in LECA) and the presence of shared
step would involve the partial degeneration of intron positions among paralogs, can be traced
a number of introns, which would lose impor- to the period between the initial intron prolif-
tant functional structural elements and/or their eration and the completion of the fully modern
IEPs. As in many of modern organelle’s group II spliceosome (Veretnik et al. 2009). By the time
introns (Jarrell et al. 1988; Goldschmidt-Cler- of LECA, a very complex, modern-looking spli-
mont et al. 1991; Suchy and Schmelzer 1991; ceosome, composed of at least 78 proteins and
Lambowitz and Zimmerly 2011), the splicing all the snRNAs, had already been established
of these partially degenerated introns would (Collins and Penny 2005).
have been assisted in trans by ( partial) RNA
sequences of other introns. Presumably, some
The Origins of the Minor and Major
of these sequences were more efficient than oth-
Spliceosomal Systems
ers at assisting exogenous splicing, and thus
conferred a benefit and became fixed in the pop- In fact, LECA had not only one complex spli-
ulations, eventually giving rise to the snRNAs ceosome, but two. Comparative genomics have
of the two types of modern spliceosomes (Sharp revealed that both U2/major and U12/minor
1991). Various pieces of evidence support this spliceosomal introns, as well as RNA and pro-
idea. In addition to their structural similarities, tein components associated with their distinct
some domains of group II introns can directly splicing machineries, were present in LECA.
replace snRNAs in spliceosomal splicing in vitro These two subtypes are quite similar, and they
(e.g., group II domain DV RNA can act instead share most of the protein machinery involved in
of U6atac snRNA [Shukla and Padgett 2002]). their splicing; however, each subtype uses a spe-
Conversely, U5 snRNA can complement group cific subset of snRNAs (with the exception of
II introns missing the ID3 region (Hetzer et al. U5), which are likely the most ancestral part of
1997). the spliceosome. These subsets of snRNAs are
These ( probably quite inefficient) initial nonetheless evolutionarily related to one anoth-
protospliceosomes could then have evolved in- er and probably originated by gene duplication
creased efficiency by recruiting auxiliary pro- some time before LECA (Russell et al. 2006).
teins. Some of these factors could have been de- Thus, debate has emerged as to which subtype
rived from the multiple available IEP copies, originated first in evolution and which one is
which can also act in trans in the splicing of derived. Several nonconclusive arguments can
related introns (Dai and Zimmerly 2003; Meng be put forward for each case (Roy and Irimia
et al. 2005). Others could have been coopted 2009; Rogozin et al. 2012). Supporting U12 in-
from various other processes of RNA metabo- trons’ ancestrality is their higher structural sim-
lism (e.g., Lsm/Sm proteins, whose homologs ilarity with group II introns. The BP in both
play diverse roles in prokaryotes) (Wilusz and types typically consist of an A within a polypy-
Wilusz 2005; Mura et al. 2013), or recruited rimidine tract located at a constant, short dis-
from sequences of retrotransposons or other tance from the 30 ss (domain DIV in group II
mobile elements (e.g., Prp8, the largest protein introns) (Bonen and Vogel 2001; Lambowitz
of the spliceosome and an integral part of its and Zimmerly 2011); on the other hand, U2
catalytic center, is thought to have evolved introns’ BPs are usually located further away
from a retroelement-encoded reverse transcrip- from the 30 ss, and at a much wider range of
tase) (Dlakić and Mushegian 2011). Gene du- distances. In fact, U2 introns ancestrally have a
plication and functional specialization are likely polypyrimidine tract between the BP and the
to have also played a role. For example, the 30 ss (Irimia and Roy 2008a), which has been
aforementioned Sm/Lsm proteins experienced hypothesized to derive from the ancestral BP
several waves of early gene duplication (Veretnik sequences of U12 introns (Burge et al. 1998).
et al. 2009), which, based on their phylogenetic Along these lines, U12 introns can be viewed
8 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
Origin of Introns and Alternative Splicing
in some sense as a transitional form between LECA. The time point corresponding to LECA
the highly constrained structures of group II represents a quite clear divide in our confidence
introns (Bonen and Vogel 2001; Lambowitz about evolutionary inferences. Reconstruction
and Zimmerly 2011) and the diffuse splicing of steps leading up to LECA (origins of spli-
signals of most U2 spliceosomal introns (Irimia ceosomal-like introns, reasons for the unprece-
et al. 2007b; Irimia and Roy 2008a; Schwartz dented spread of introns, and transfer of splic-
et al. 2008). Consistent with such a directional ing from cis- to trans-encoded elements, etc.)
process, U12 introns are known to often convert remains uncertain because of a lack of close
into U2 introns, whereas the reverse process has protoeukaryotic-like relatives. In contrast, infer-
never been described (e.g., Burge et al. 1998; ences of features of LECA are relatively much
Alioto 2007). Second, positions of U12 introns more straightforward because of the possibil-
may be more conserved between Arabidopsis ity of comparing lineages that diverged from
and humans than those of U2 introns, and LECA. That is, whereas the former by necessity
they are also more often localized at the 50 of relies on largely indirect evidence, the latter can
the gene—a signature of ancestral introns that draw on direct (not to say incontrovertible) ev-
have resisted intron loss (Basu et al. 2008a). idence.
On the other hand, phylogenetic reconstruc- Comparative studies of modern genomes
tion of the insertion sites of ancient ( pre-LECA) have drawn a surprising picture of the intron –
introns revealed a U2-like insertion bias (remi- exon structures of genes in the genome of LECA
niscent of the protosplice site model of intron (Roy and Irimia 2009; Rogozin et al. 2012; Koo-
insertion) (Dibb and Newman 1989). This result nin et al. 2013), all of them pointing to a com-
suggests that U2 introns dominated at least by plex spliceosomal system. First, comparisons of
the time of LECA (Basu et al. 2008c). However, intron positions across orthologous genes from
this leaves open the possibility that this prolifer- dozens of eukaryotic genomes have shown that
ation postdates the secondary advent of U2 in- LECA was remarkably intron rich. At its sim-
trons, because group II introns may not have plest, the logic is that if modern introns were
an insertion bias similar to U2 introns. Finally, created after the various species diverged, the
neither type may be truly ancestral: The two positions of these introns in homologous genes
types may have emerged semi-independently might be expected to be quite different across
from a larger pool of primitive snRNA dupli- species, whereas if ancestral genes already con-
cates with a shared protein machinery. tained many introns and those introns have
been retained in modern species, intron posi-
tions in homologous genes might largely coin-
HISTORY OF THE SPLICEOSOMAL SYSTEM
cide across species (Fig. 3). This general strategy
In this section we will describe the evolutionary can be modified to account for the possibil-
history of introns since LECA. First, we dis- ity that general preferences in intron insertion
cuss attempts to reconstruct LECA’s intron – (e.g., into protosplice sites) and/or intron
exon structures using comparisons of modern phase biases (reviewed in Rogozin et al. 2012)
genomes. Then, we summarize the evolutionary could lead to independent intron insertions oc-
trajectories taken by different eukaryotic lineag- curring at homologous positions in different
es, ranging from near complete loss and trans- lineages (Sverdlov et al. 2005; Yoshihama et al.
formation of their ancestral introns, to the in- 2006). Avariety of studies have shown that genes
tronic stasis observed in many modern lineages. from intron-rich species from any major eu-
karyotic supergroup share a significant number
of intron positions with other groups (Fedorov
Reconstruction of Intron– Exon
et al. 2002; Rogozin et al. 2003; Csurös et al.
Structures in LECA
2008, 2011). Different investigators have used
The evolutionary steps described in the previ- different methods of evolutionary inference,
ous section were largely complete by the time of ranging from parsimony to maximum likeli-
Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071 9Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
M. Irimia and S.W. Roy
A Loss Gene A B Gain Gene A
Inferred ancestor
Figure 3. Scenarios for intron evolution: shared ancestral introns versus independent gain. Hypothetical com-
parisons of intron positions across orthologous genes of distantly related species are shown. (A) Many introns
are in homologous positions, suggesting that these introns represent ancestral introns that have been retained.
(B) Lack of correspondence between intron positions in different lineages suggest that most introns have been
independently gained within each lineage.
hood, to attempt to reconstruct ancestral intron same 50 ss, GTATGT. Put another way, the prob-
densities (Rogozin et al. 2003; Qiu et al. 2004; ability that two randomly chosen Saccharomyces
Csurös 2005, 2008; Nguyen et al. 2005; Roy and cerevisiae introns have the same hexamer splice
Gilbert 2005; Carmel et al. 2007, 2009). With site is 58%, whereas two human 50 ss’s will cor-
increasing taxonomic sampling usually yield- respond only 6% of the time. Following the
ing increasingly higher estimates for LECA’s in- intuitive sense that (i) genome complexity will
tron density, the latest reconstructions (Csurös reflect organismal complexity; and (ii) more
et al. 2011) used a Markov chain Monte Carlo “highly evolved” organisms have more trans-
approach to investigate 99 genomes from five formed genomes, it was initially thought that
eukaryotic supergroups, and estimated that ancestral unicellular eukaryotes would have
LECA had 53% – 74% of modern human intron had “simple” yeastlike intron sequences, with
density, or 4.5 – 6.3 introns per gene (assuming regular and predictable splice signals, and that
similar average protein-coding lengths). That is, “complex” diverse signals arose by sequence di-
available evidence suggests that the intron –exon vergence in multicellular lineages ( perhaps re-
structures of ancestral eukaryotes were exceed- lated to AS, which makes extensive use of splice
ingly complex, with intron densities compara- site heterogeneity) (Ast 2004). However, com-
ble to the most intron-dense modern densities, parisons across species revealed that in fact yeast
and more intron dense than, for instance, mod- is the exception, with the majority of unicellular
ern plants and insects (Fig. 4). and multicellular species alike (and thus likely
The second unexpected characteristics of LECA) utilizing heterogeneous splicing signals,
LECA’s intron – exon structures involve the in- a finding that holds both for the 50 ss and the BP
tron sequences themselves. Nearly all studied (Irimia et al. 2007a; Irimia and Roy 2008a;
eukaryotic species show similar consensus se- Schwartz et al. 2008; Keren et al. 2010). More-
quences for core splicing sequence motifs—the over, LECA’s introns also likely harbored a poly-
50 splice site (ss), BP, and 30 ss (Fig. 1)—reflecting pyrimidine tract between the BP and the 30 ss—
largely similar sequence preferences across spe- similar to modern animals and plants, but not
cies determined by a shared spliceosomal ma- yeasts—as inferred by the general presence of
chinery. However, the strength of these prefer- this signal in representatives of most eukaryotic
ences—the extent to which individual introns’ supergroups (Irimia and Roy 2008a). This se-
specific motifs adhere to this consensus—varies quence is bound in the first steps of spliceosome
dramatically (Fig. 4). For instance, whereas hu- assembly by the U2AF65 (Zamore et al. 1992),
man introns use a vast array of 50 ss, 75% of all a factor already present in the ancestral spli-
introns in the model yeast S. cerevisiae have the ceosome (Collins and Penny 2005). In addition
10 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
Origin of Introns and Alternative Splicing
LECA
A
op ter
ns s
us
a
rm as
P. s mbe um
T. t elan NM
ga an
nh us
hil
r vu e
P. t assa m
ca
C. neof llanii
lis
z a lia
C. astensis
ce NM
hr2 rin
t ii
a
he og
me alis
luc ca
ele orm
ae
ru
na
s
ns
po ut
m
ico
ad
C. evisi
li
y ti
C. agin lia
ory ure
P. t ayd ri
is
e
C. iculo
m s
s
n-c ma
icu
N. lcipa
O. olyti
S. corn
U. urbe
rol
tha ii
D. atan
tan
pie
A. vecte
lia
ta
tol
T. vlamb
P. f jae
)
v
A. ond
R. etra
un
bre
i
the
r
his
pa
rei
sa
na
sil
cr
o
lip
m
ri
a
g
n
c
c
T. g
M.
G.
G.
(no
N.
N.
H.
E.
E.
E.
S.
B.
B.
Y.
Introns/geneDownloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
M. Irimia and S.W. Roy
For example, nonsense-mediated decay (NMD), For some lineages, phylogenetic reconstruction
a mechanism to eliminate spurious or unwant- of intron loss and gain suggest that modern high
ed transcripts, is found in diverse eukaryotes intron density is almost completely owing to
(Amrani et al. 2004; Chen et al. 2008; Jaillon retention of ancestral introns (e.g., in the case
et al. 2008; Kerényi et al. 2008; Roy and Irimia of vertebrates, intron number appears to have
2009). Indeed NMD and intron spread are likely decreased somewhat since the ancestor of ani-
closely linked, because frequent splicing errors mals 1 billion years ago) (Sullivan et al. 2006;
would necessitate NMD, whereas the presence Srivastava et al. 2008; Csurös et al. 2011). In
of NMD would decrease the costs of intron other cases, many modern introns appear to
spread (Koonin 2006; Lynch et al. 2006). Other have been gained through a variety of processes,
“policing” processes, such as the ubiquitin sig- most dramatically creation of hundreds or thou-
naling at the proteomic level, could also have sands of introns through propagation of intron-
evolved as a defense mechanism against massive creating elements (Worden et al. 2009; Roy and
intron presence (Koonin 2006). The timing of Irimia 2012; van der Burgt et al. 2012). Similar-
emergence of other secondary aspects of the ly, organelle-derived or horizontally transferred
spliceosomal system, most notably the function- genes sometimes acquire similar densities to
al associations between chromatin and splicing those of more ancestral nuclear host genes (Basu
seen in animals (Braunschweig et al. 2013), still et al. 2008b; Ahmadinejad et al. 2010; Clarke et
remain to be elucidated. al. 2013, although see Roy et al. 2006 and Flot
et al. 2013). Interestingly, both nearly complete
loss of introns and dramatic intron prolifera-
Diverging from LECA: Different Modes
tion have occurred several times independently
of Spliceosomal Intron Evolution across
across the eukaryotic tree. For instance, nearly
Eukaryotes
complete loss has occurred in groups as di-
From these intron-rich ancestors with hetero- verged as fungi (at least twice), green and red
geneous splicing signals and complex spliceo- algae, apicomplexa, and excavates (three times)
somes, the full diversity of modern eukaryotic (Irimia and Roy 2008a). The reasons for this are
intron – exon structures has emerged over the still unclear, and the proposed explanations
past 1.5 billion years of evolution. The results range from selection for genome streamlining
of this evolutionary process have been dramatic. to runaway intron loss mutation (reviewed in
Most noticeably, intron densities vary by orders Maeso et al. 2012b). In lineages that have expe-
of magnitude, from several per gene and thou- rienced many new intron gains, new introns
sands or tens of thousands per genome in many appear not to have been created at constant rates
unicellular and multicellular lineages (Mer- over millions of years of evolution, but instead
chant et al. 2007; Roy and Irimia 2009; Curtis to be concentrated in episodes of gain (Roy
et al. 2012), to a handful per genome in the 2006; Roy and Penny 2006; Csurös et al. 2011).
genomes of some other species (Fig. 4) (Matsu- Over recent times, many studied modern line-
zaki et al. 2004; Vanacova et al. 2005; Morrison ages appear to be under a sort of intronic evo-
et al. 2007). Notably, these densities do not re- lutionary stasis, experiencing low (or very low)
spect our intuitive sense of biological simplicity numbers of intron gain and loss (Roy et al. 2003,
and complexity. Many parasitic lineages have 2006; Nielsen et al. 2004; Roy and Hartl 2006;
intron densities comparable to those of most Coulombe-Huntington and Majewski 2007a,b;
intron-rich multicellular lineages, and at least Roy and Penny 2007a; Stajich et al. 2007; Loh
one multicellular lineage has intron densities et al. 2008; Zhang et al. 2010). On the contrary,
comparable to some of the most intron-poor only a few known exceptions scattered across the
known unicellulars (Collén et al. 2013). Al- eukaryotic tree show significant ongoing intron
though the origins of nearly intronless lineag- gain-dominated evolution (Carmel et al. 2007;
es—massive intron loss—are clear, the origins Roy and Penny 2007b; Worden et al. 2009; De-
of the highly intron-dense lineages are less so. noeud et al. 2010; van der Burgt et al. 2012).
12 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
Origin of Introns and Alternative Splicing
In addition to intron densities, splicing sig- though it is difficult to confidently estimate
nals also show marked differences across ge- average and median intron length in LECA, they
nomes (Fig. 4). In particular, 50 ss consensus in may have been similar to those in most eukary-
a given species may be either very strict, with otic genomes studied so far, perhaps around
nearly all introns in the genome having an iden- 150 nucleotides on average and with a median
tical or similar sequence, or highly heteroge- centered around 70 – 90 nucleotides. It should
neous, as in the case of the human genome (Iri- be noted, however, that introns were probably
mia et al. 2007b). Heterogeneous 50 ss signals significantly longer at least in earlier ancestors,
are observed in most lineages, whereas the few as group II introns from which they derived are
lineages with strict signals are scattered across at least 400– 800 nucleotides long (Lambowitz
the phylogenetic tree, representing exceptions and Zimmerly 2011), and they may reach 2.5
to the general pattern in nearly all major eukary- kbp when they encode their own IEP (Lambo-
otic supergroups (Irimia et al. 2007b). This dis- witz and Zimmerly 2004; Koonin 2009). Sur-
tribution strongly suggests that these exceptions veys of intron lengths across eukaryotes reveal
have evolved independently within each lineage, striking deviations in intron length distributions
echoing the case for lineages that have under- in both directions. First, tiny introns (18 – 36
gone nearly complete intron loss. Unexpectedly, nucleotides) have independently evolved in a
these two processes were found to have a very few species scattered across the eukaryotic tree,
clear correspondence: In nearly intronless spe- in groups as diverged as animals (Ogino et al.
cies, the few remaining introns have highly sim- 2010; Tsai et al. 2013), green algae (Gilson et al.
ilar and constrained 50 ss signals (Fig. 4) (Irimia 2006), and ciliates (Russell et al. 1994). More-
et al. 2007b). A similar pattern of strict signals over, in all of these species the distributions
has been observed for the BP (Irimia and Roy of lengths are very sharp—suggesting strong
2008a; Schwartz et al. 2008), although with a few pressure to maintain a seemingly optimal small
exceptions involving species with few introns length—and their splicing signals are highly dif-
but heterogeneous BPs (Fig. 4). Finally, in a few fuse (with the noted exception of microspori-
instances, a further constraint on splicing sig- dians) (Irimia and Roy 2008a; Lee et al. 2010).
nals has evolved, in which the BP is “anchored” On the opposite side, introns from certain lin-
to the 30 ss, always present at exactly or almost eages are unusually long. These include some
the same number of nucleotides away (Irimia plants (Jiang and Goertzen 2011), brown algae
and Roy 2008a). Thus, in these extreme cases, (Cock et al. 2010), and, especially, vertebrates,
the remaining U2 introns show similarities to with averages one order of magnitude higher
U12 introns, which are also found in low num- than most species, and with some introns reach-
bers per genome: extended constrained 50 ss and ing up to one megabase. These expansions seem
BP signals and BP position. Although several likely to be associated with insertion of trans-
explanations have been proposed to explain the posable elements into intronic sequences (Jiang
coevolution between splicing signals and intron and Goertzen 2011), although, in some cases,
numbers (Irimia et al. 2007b, 2009), the cause exceptionally long introns may be associated
remains obscure. It seems likely that an impor- with the presence of large numbers of cis-regu-
tant component is changes in the spliceosome: latory elements (Irimia et al. 2011).
Massive introns loss is often accompanied by
loss of some auxiliary spliceosomal factors
FUNCTIONS OF INTRONS: ORIGINS AND
largely involved in recognition of auxiliary splic-
EVOLUTION OF ALTERNATIVE SPLICING
ing sequence signals such as ESEs (Plass et al.
2008), presumably leading to a greater emphasis Given the fascinating and complex evolutionary
on core splicing signals, and thus requiring in- paths that intron – exon structures have taken
creased information content in those signals. from LECA to each extant eukaryotic lineage,
Intron lengths have also experienced con- perhaps the major paradox is that, for most cas-
siderable divergence across eukaryotes. Al- es, introns are not known to play a general func-
Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071 13Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
M. Irimia and S.W. Roy
tional role in the biology of the cell: They are in intracellular transport, either at the mRNA
simply intervening sequences that must be re- (Buckley et al. 2011) or protein level (Freitag
moved for proper gene expression to occur. Sev- et al. 2012; Kabran et al. 2012). Nevertheless,
eral evolutionary functions have been proposed despite a plethora of described examples of
for introns, including enhancing within-gene each kind, a major unanswered question is still
homologous recombination (Comeron and to what extent AS is functional or simply splic-
Kreitman 2000), and facilitation of exon shuf- ing “noise” (Sorek et al. 2004; Irimia et al. 2008;
fling and thus the evolution of new genes (Gil- Roy and Irimia 2008).
bert 1987; Liu and Grigoriev 2004). In addition, AS was initially believed a relatively recent
certain intron sequences are known to play reg- innovation, largely associated with the emer-
ulatory roles in transcription, either by acting gence of multicellularity (Ast 2004). However,
as a limiting factor in cotranscriptional splic- the wealth of genomes and transcriptomes ac-
ing (Patel et al. 2002), or indirectly by harbor- cumulated over the past decade has turned
ing DNA regulatory elements (e.g., Epstein et al. around this view. First, significant amounts of
1999; Irimia et al. 2012; Maeso et al. 2012). AS (usually intron retention) have now been
Similarly, other introns harbor nested bona described in nearly all well-studied species.
fide protein-coding or noncoding RNA genes These include representatives of all eukaryotic
(e.g., Assis et al. 2008; Kumar 2009; Chorev supergroups (McGuire et al. 2008; Labadorf
and Carmel 2013). However, the most wide- et al. 2010; Otto et al. 2010; Rhind et al. 2011;
spread function of introns is probably through Shen et al. 2011; Curtis et al. 2012; Sebé-Pedrós
AS, which has been estimated to occur in nearly et al. 2013), and even some intron-poor species
all human multiexonic genes (Pan et al. 2008; (Pleiss et al. 2007; Kabran et al. 2012). Second,
Wang et al. 2008). By differentially processing LECA’s intron – exon structures seem to have
introns (and exons), eukaryotic cells can gener- met all classic requirements for AS to occur
ate multiple mRNA and protein isoforms from (Roy and Irimia 2009; Koonin et al. 2013): It
a single gene. This differential processing may was intron-rich (Irimia et al. 2007b) and had
consist of nonsplicing (or retention) of introns, heterogeneous splice signals, which are associ-
in the simplest case, but also of exclusion (or ated with splicing variation within and across
skipping) of an entire exon, alternative choice of genomes (Stamm et al. 2000; Ast 2004; Baek
multiple 50 ss, and/or 30 ss within an intron and Green 2005). Moreover, functional and evo-
(which would produce exon truncations/ex- lutionary analyses of alternatively spliced genes
pansions), or more complex splicing patterns showed that ancient eukaryotic genes (i.e., those
(Irimia and Blencowe 2012). present in LECA) show high levels of AS in dif-
AS may itself perform different functions. ferent eukaryotic lineages, suggesting that they
First, AS can generate protein isoforms with could have been amenable to AS also in LECA
highly distinct sequences that may have dramat- (Irimia et al. 2007b). Thus, all these lines of
ically different activities and biological roles evidence strongly suggest that LECA could
(e.g., Boise et al. 1993; Gabut et al. 2011); but have had at least a simple program of AS, per-
also proteins with only slight variations that haps comparable to those observed in most
contribute to fine-tuning of cellular processes modern unicellular eukaryotes, dominated by
(Lopez 1998). Second, AS can produce non- intron retention and perhaps acting mainly as
functional transcript variants, serving as an ex- an extra layer of gene regulation. However, as for
tra layer of down-regulation of gene expression. any modern eukaryote, it is not clear what frac-
This may be achieved by on/off protein switches tion of that transcriptional diversity played any
(Bingham et al. 1988), or by coupling missense relevant biological role.
AS to NMD (Lewis et al. 2003; Lareau et al. Despite the similar patterns of AS in most
2007; Yap et al. 2012), leading to transcript deg- eukaryotes, AS has also undergone remarkable
radation before translation. Third, inclusion diversification in some specific lineages. For in-
of alternative sequences may lead to differences stance, and most unexpectedly (Ast 2004), the
14 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071Downloaded from http://cshperspectives.cshlp.org/ on October 9, 2021 - Published by Cold Spring Harbor Laboratory Press
Origin of Introns and Alternative Splicing
intron-poor yeast S. cerevisiae may have evolved This species has intron densities comparable
regulated, functional AS for the majority, if not to humans (eight to nine introns per gene),
all, of its intron-containing genes. S. cerevisiae and displays extremely high levels of splicing
has 260 introns, with a strong bias toward variation, including both intron retention and
ribosomal genes (Juneau et al. 2006). In-depth exon skipping. RNAseq quantification of exon
transcriptomic analyses have shown that splic- skipping shows levels comparable only to the
ing of specific sets of introns can be down-reg- human cortex (Curtis et al. 2012), the highest
ulated in response to different growth condi- AS levels described so far in any tissue and or-
tions (Pleiss et al. 2007; Parenteau et al. 2011). ganism (Barbosa-Morais et al. 2012). Whereas
For example, only a few minutes after amino much of B. natans AS may represent noisy splic-
acid starvation, splicing of most ribosomal in- ing, a few potential cases in which AS drives
trons is down-regulated, thereby reducing pro- alternative subcellular targeting were identified
duction of ribosomal proteins, thus presumably (Curtis et al. 2012), suggesting that AS may play
reducing global translation (Pleiss et al. 2007). important and diverse roles in the biology of
Nonetheless, the most famous example of B. natans. Surveys of diverse eukaryotes to iden-
extreme AS diversification is found in animals, tify additional cases of independent evolution
particularly in primates (Barbosa-Morais et al. of widespread functional exon skipping (or oth-
2012). Not only is AS much more common in er forms of protein-diversifying AS) are crucial
metazoans, but the mode of AS is also qualita- to a full understanding of the functional impact
tively different from that of other eukaryotic of the spliceosomal system in eukaryotes.
lineages. Unlike most eukaryotes, in which in-
tron retention dominates, animals show fre- CONCLUDING REMARKS
quent usage of exon skipping (McGuire et al.
Evidence from biochemistry and comparative
2008). In exon skipping, AS leads to the inclu-
genomics strongly suggests that spliceosomal
sion/exclusion of full exons that behave as “cas-
introns originated and proliferated during the
settes” of sequence and that can be deployed in
origin of eukaryotes, in a process intimately
a tissue-specific manner. Consistent with the
linked to other evolutionary events of eukar-
cassette idea, most regulated alternative exons
yogenesis. The resulting portrait of the LECA
are multiple of three nucleotides (Xing and Lee
spliceosomal system was comparable to that
2005), thereby preserving the reading frame
of intron-rich modern eukaryotic organisms,
when included/excluded. Thus, cassette exons
with myriad introns, complex splicing mach-
can have a much broader impact on proteome
inery, and coupling of splicing to other cellular
diversity, by introducing or disrupting func-
processes. From that complex ancestor, intron –
tional protein domains, or by regulating disor-
exon structures have been highly plastic
dered regions often involved in protein – protein
throughout eukaryotic evolution, with cases of
interactions (Buljan et al. 2012; Ellis et al. 2012;
extreme divergence from their ancestors that
Colak et al. 2013). Intriguingly, the reasons for
shaped the genomic landscapes of eukaryotes
this transition from intron retention (mostly
probably like no other. Particularly remarkable
impacting gene regulation) to exon skipping
are the multiple instances of recurrent evolution
(increasing proteomic diversity) during the
of multiple features, including massive intron
evolution of early animals are still unknown.
loss, constraint of splicing signals, restriction
Given the potential centrality of AS for the or-
of BP-30 ss distance, loss of polypyrimidine tract,
igin of animal complexity, understanding this
and complete loss of U12 introns.
crucial transition is a central priority for future
studies.
ACKNOWLEDGMENTS
Strikingly, another eukaryotic lineage has
recently been reported to have mammalian- We thank Eesha Sharma for critical reading of
like levels of exon skipping, the chlorarachnio- the manuscript. M.I. holds a PDF from the Hu-
phyte Bigelowiella natans (Curtis et al. 2012). man Frontiers Science Program Organization.
Cite this article as Cold Spring Harb Perspect Biol 2014;6:a016071 15You can also read
