nature reviews genetics https://doi.org/10.

1038/s41576-022-00556-8

Review article Check for updates

Regulation of pre-mRNA splicing:

roles in physiology and disease,
and therapeutic prospects
Malgorzata Ewa Rogalska    1,5, Claudia Vivori    1,2,4,5 & Juan Valcárcel    1,2,3 
Abstract Sections

The removal of introns from mRNA precursors and its regulation by Introduction

alternative splicing are key for eukaryotic gene expression and cellular 5′ splice site recognition
function, as evidenced by the numerous pathologies induced or modified 3′ splice site recognition
by splicing alterations. Major recent advances have been made in
Splice site communication
understanding the structures and functions of the splicing machinery,
Role of regulatory sequences
in the description and classification of physiological and pathological
isoforms and in the development of the first therapies for genetic Co-transcriptional regulation

diseases based on modulation of splicing. Here, we review this progress Epitranscriptomic regulation
and discuss important remaining challenges, including predicting Regulation by RNA structure
splice sites from genomic sequences, understanding the variety of
Conclusions and future
molecular mechanisms and logic of splicing regulation, and harnessing perspectives
this knowledge for probing gene function and disease aetiology and
for the design of novel therapeutic approaches.

Genome Biology Program, Centre for Genomic Regulation (CRG), The Barcelona Institute of Science and
1

Technology, Barcelona, Spain. 2Department of Medicine and Life Sciences, Universitat Pompeu Fabra (UPF),
Barcelona, Spain. 3Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Spain. 4Present address:
The Francis Crick Institute, London, UK. 5These authors contributed equally: Malgorzata Ewa Rogalska, Claudia Vivori.
 e-mail: [email protected]

Nature Reviews Genetics

Review article

Introduction In higher eukaryotes, accurate splicing is a formidable challenge

Most primary RNA transcripts produced by eukaryotic RNA polymerase II because, whereas intron removal requires single-nucleotide precision
(RNA Pol II) contain introns that need to be removed by the process of to preserve coding information in exons, the sequences delineating
splicing (Box 1, see the figure), to generate functional mRNAs and long exon–intron boundaries are remarkably diverse, often resembling other
non-coding RNAs. Although many aspects of the origin and function of sequences that are not splice sites21 (Figs. 2 and 3). Testifying to the high
introns remain uncertain (Box 1), they likely originated from autocata- level of specificity required, mutations in pre-mRNAs or in splicing fac-
lytic RNAs that spread through the genomes of primitive eukaryotes as tors lead to alterations in the splicing process that are associated with
retrotranscribed DNA transposable elements1. During evolution, the human pathologies ranging from genetic diseases to neurodegenera-
strict sequence and 3D structure required for self-removal of intronic tion or cancer13,14,18. Here, we provide examples of the clinical impact
RNAs were progressively relaxed as their excision became dependent of splicing alterations, and discuss how a deeper understanding of
upon an increasingly complex cellular machinery known as the spliceo- the regulatory mechanisms of splicing can help design therapies that
some (Fig. 1; see Supplementary Tables 1–3). The function of the spli- counteract these detrimental effects of splicing alterations or even
ceosome depends on the recognition of intronic boundaries by small exacerbate them to facilitate immune responses against tumour cells22.
nuclear ribonucleoprotein (snRNP) complexes, followed by a series
of conformational transitions that involve remodelling of numerous 5′ splice site recognition
RNA–RNA, RNA–protein and protein–protein interactions to enable Initial recognition of the 5′ss of an intron is carried out by the U1 snRNP
protein-assisted, RNA-based catalysis of intron removal2–5. A general complex of the spliceosome through base-pairing interactions involv-
principle of spliceosome assembly is that intron boundaries are recog- ing the 5′ end of U1 small nuclear RNA (snRNA) (Figs. 1 and 2). Given the
nized multiple times, ensuring accuracy in the splicing process4,6. Two variability of 5′ss sequences, base pairing is key to defining the effi-
classes of introns and spliceosomes coexist in complex organisms, the ciency (or ‘strength’) with which a particular 5′ss is used23. 5′ss mutations
minor class being spliced with slower kinetics but being nevertheless that weaken base pairing with U1 snRNA can cause defective splice site
essential for the expression of genes involved in multiple processes, recognition and disease. For example, a single mutation at position +6
including early development7. of intron 20 of the IKBKAP gene causes exon skipping, introducing
Variations in the patterns of intron removal (known as alternative a premature termination codon that reduces expression of functional
splicing) occur in the majority of genes in multicellular organisms8–10, protein24 (Fig. 4a). This leads to the autosomal recessive condition
contributing to proteome diversification as well as to the regulation familial dysautonomia, a neurodegenerative disorder that often causes
of gene expression by the degradation of transcripts containing premature death owing to cardio-respiratory arrest25. Conversely,
premature termination codons (Box 2). There are numerous examples mutations in the 5′ end sequence of one of the multiple gene copies
of alternative splicing events that are important for cell identity, pluripo- of U1 snRNA, which are observed in patients with chronic lymphocytic
tency and organismal physiology, or that contribute to various patholo- leukaemia and sonic hedgehog medulloblastoma, can result in the
gies (reviewed in refs. 11–15). How widespread is the functional relevance recognition of novel 5′ss that have base-pairing complementarity with
of alternative splicing, however, remains an open question (Box 2). the mutated U1 snRNA, and more generally alter patterns of splicing of
Owing to recent developments in the methods, software and data multiple genes. These genes include known cancer drivers — for example,
sets available to study alternative splicing (Supplementary Box 1), resulting in inactivation of tumour suppressor genes or activation of
now is a particularly exciting time for studies of RNA splicing and proto-oncogenes — and correlate with worse disease prognosis26,27.
its regulation. For the first time, detailed cryogenic electron micro­ However, it is striking that most 5′ss that are efficiently recognized
scopy structures of the spliceosome at various steps of its assembly do not form perfect base-pairing interactions with the 5′ end of U1
and catalysis have been determined at unprecedented resolution snRNA, even after considering non-canonical base-pairing schemes
(Fig. 1; see Supplementary Table 4), providing a structural framework (such as bulged or other unpaired nucleotides28–30). It is also striking
to interpret decades of previous biochemical and genetic studies2–5. that mutations of the 5′ss that are not predicted to significantly dis-
There is also an unprecedented wealth of transcriptome data showing rupt base pairing with U1 snRNA nevertheless alter 5′ss recognition,
the large diversity of transcript isoforms, with profound implications being associated with diseases such as Fanconi anaemia, haemophilia,
for understanding basic biology and disease outcomes. In parallel, neurofibromatosis and phenylketonuria29. Massively parallel splic-
strong evidence has accumulated indicating that mutations in fac- ing assays assessing all of the 32,768 possible 5′ss sequences — NNN/
tors involved in post-transcriptional regulation contribute to cancer GYNNNN (N = any nucleotide, maintaining at position +1 the G required
and neuro­degeneration16. Finally, the success of splicing-modulating for catalysis; Y = pyrimidine at position +2, present in 99.6% of 5′ss) —
therapies for the treatment of spinal muscular atrophy17, as well as unex- in three different minigene contexts confirmed the relevance of base
pected applications of splicing inhibitors as cancer therapies or for pairing between U1 snRNA and the 5′ss for splicing efficiency but also
the maintenance of cell totipotency18,19, bring hope that our increasing revealed marked context-related differences31. This suggests that addi-
understanding of splicing mechanisms will provide a new generation tional nearby sequences, their cognate factors and interactions between
of therapeutics. Alternatively spliced isoforms should also be given these and U1 snRNP components can aid the efficient use of 5′ss that have
careful consideration in the design of mRNA-based therapeutics20. suboptimal base pairing with U1 snRNA32–34; for example, such additional
This Review focuses on the most recent advances in our under- sequences and factors have been shown to have a role in alternative
standing of the molecular mechanisms that help distinguish between splicing that regulates the physiological shift in energy metabolism from
introns and exons to enable accurate splicing and to regulate alter- glycolysis to oxidative phosphorylation35. Systematic approaches (such
native splicing. These include the interplay between components as those described in ref. 31) can help predict the pathogenic effects of
of the splicing machinery that recognize the 5′ splice site (5′ss) and 5′ss mutations or of natural sequence variation. For example, they veri-
3′ splice site (3′ss), their coupling with the process of transcription, fied that nearly 90% of the 5′ss mutations found in BRCA2 in breast cancer
the role of RNA structures and the contribution of RNA modifications. samples do affect splicing, potentially facilitating genetic screening31.

Nature Reviews Genetics

Review article

Therapeutic targeting
Blocking the inhibitory effect of an intronic splicing silencer (ISS) on
the recognition of a 5′ss using an antisense oligonucleotide known as
Box 1
nusinersen (approved for clinical use in 2016) has provided a major
therapeutic breakthrough for patients with spinal muscular atrophy, The function of introns
which is a leading genetic cause of infant mortality17,36–38 (Fig. 4a). Spinal
muscular atrophy is caused by mutations that inactivate SMN1, which Introns are internal sequences Exon 1 BP Exon 2
encodes a protein important for snRNP assembly. For reasons that that are removed from GU A AG
5′ss Intron 3′ss
remain unclear, loss of SMN1 function mainly affects the function of precursor mRNA transcripts
motor neurons, leading to progressive muscle weakness and in the by a two-step splicing process
most severe cases to death within the first 2 years of life38. Quarterly (see the figure). The first First step
intrathecal injection of nusinersen increases the levels of protein step involves cleavage of

AG
generated from a second gene, SMN2, which under normal circum- the phosphodiester bond A2′OH
stances fails to produce functional protein owing to limited inclusion between the upstream exon
P GU
of exon 7 (refs. 17,39). (exon 1) and the intron, and
Facilitated by the early success of nusinersen, an orally available, concomitant formation of a
small-molecule, pyrido-pyrimidinone drug known as risdiplam that has 2′–5′ phosphodiester bond
similar effects on SMN2 exon 7 inclusion obtained US Food and Drug between the 5′ guanosine and U
Administration (FDA) approval in 2020 and has shown promising clini- an internal adenosine (the PG
A AG
cal results40,41. One mode of action proposed for risdiplam analogues branch point (BP)), generating 3′OH
is to ‘repair’ the bulge formed by the lack of base pairing between a lariat intermediate. The Second step
U1 snRNA and the last nucleotide of SMN2 exon 7, thus stabilizing second step involves cleavage
U1 snRNP recruitment42 (Figs. 2b and 4a). This is achieved, at least of the phosphodiester bond

AG GU
in part, by facilitating an interaction between the zinc finger of the between the 3′ end of the P P
A
U1 snRNP protein U1C and the minor groove of the U1–5′ss helix42, intron and the downstream 3′OH
although other mechanisms may also be involved43. It is quite remark- exon (exon 2), concomitant
able that a compound that modulates structural features of particular with ligation of the two exons
U1–5′ss configurations has therapeutic properties, paving the way and release of the intron in a
to a new generation of compounds targeting 5′ss recognition. lariat configuration.

GU
P P
Interestingly, engineering U1 snRNA such that its 5′ end can base Despite great progress in A
pair to intronic sequences downstream of a 5′ss activates the use of understanding the splicing
the bona fide upstream 5′ss (refs. 44,45) (Fig. 4a). This suggests that process and its regulation, mRNA
an increase in the local concentration of and/or cooperativity between important fundamental
Intron lariat
U1 snRNP complexes can enhance 5′ss recognition, perhaps by propa- questions remain. One
gating complexes that change the physical behaviour of the exon. such question is whether
This approach has been used in vitro to correct exon skipping events intronic sequences can have
associated with various pathologies including spinal muscular atrophy, functions of their own or Linearized intron
cystic fibrosis or neurological disorders such as CDKL5-deficiency are simply by-products of a
disorder46–48 (Fig. 4a). However, the oncogenic properties of U1 snRNA process designed to eliminate snoRNAs
mutations mentioned above26,27 bring a note of caution when considering ancient transposon insertions. Degradation microRNAs
the general applicability of this approach. Recent work has shown
Although not directly related to 5′ss recognition, another major that particular introns in the yeast Saccharomyces cerevisiae
function of the U1 snRNP complex is to bind to 3′ untranslated regions accumulate as linear RNA species under various stress conditions,
of mRNAs and inhibit the use of proximal 3′ end formation sites, which with their accumulation contributing to stress responses through
are often used in actively proliferating cells, including cancer cells49,50. a regulatory network that involves the target of rapamycin complex
Consistent with this, inhibition of U1 snRNP using antisense morpholino (TORC), a key integrator of growth signalling295. Indeed, systematic
oligonucleotides increases cancer cell migration and invasion, whereas deletion of introns in S. cerevisiae led to an impaired response to
increased levels of U1 snRNP inhibit these phenotypes51. starvation, which was linked to the function of intronic sequences
Similar approaches have been proposed for the therapeutic cor- as repressors of ribosomal protein synthesis296. These observations
rection of 5′ss recognition in familial dysautonomia (Fig. 4a). These are consistent with the long-standing proposal that introns
include antisense oligonucleotides targeting ISSs downstream from might provide, via autonomous functions, an additional layer of
the 5′ss of IKBKAP exon 20 (refs. 52,53), modified U1 snRNAs54 and small genetic information to that provided by mature mRNAs and long
molecules such as kinetin, a plant cytokinin that enhances the recogni- non-coding RNAs297. It remains unclear, however, whether the
tion of 5′ss flanked by a particular sequence motif55,56, or RECTAS, which functions documented above for certain yeast introns, as well
enhances the phosphorylation of SRSF6, a splicing regulatory factor as other examples of non-coding RNAs located within introns
that functions through an intronic splicing enhancer (ISE) located in (such as small nucleolar RNAs (snoRNAs)298 and microRNAs), are
intron 20 (refs. 57,58). important exceptions to a general lack of function for introns.
Other examples of 5′ss recognition that have significant therapeu- 3′ss, 3′ splice site; 5′ss, 5′ splice site.
tic potential for modulation include the induction of a pro-apoptotic

Nature Reviews Genetics

Review article
Complexes RNA interactions Cryogenic electron microscopy structures
Exon 1 BP Exon 2 BP Exon 2
A Exon 1 A
GU AG GU Intron AG
5′ss Intron 3′ss 3′ss
5′ss
U1 SF1 U2 AF U1 snRNP
5′ss recognition S. cerevisiae
6n7p
U1 U1 snRNP
U1 SF1 U2 AF
E GU A AG GU A AG 5′ss

Prp5/DDX46ATP U2 17S U2 snRNP

UAP56/DDX39BATP SF3 BP recognition
SF1 U1 S. cerevisiae U2 snRNP
U2 6g90
U2
U1 SF3 U2 AF 3′ss
GU AG U1 snRNP
Assembly

A GU A AG A

5′ss and 3′ss U4/U6-U5 snRNP

U4 U6 U5
proofreading 5′ss
U6 H. sapiens
U1 6qx9 U2 snRNP
U2 U4
U6 SF3 U2 AF U2 U5 3′ss
pre-B U4GUU1 A AG AG
U4/U6–U5
U5 GU A snRNP 5′ss

Prp28/DDX23 ATP 5′ss transfer U1 snRNP U1 snRNP

U1 U4 H. sapiens
U6 AG 6ahd U2 snRNP
U2 U2
U4 U6 SF3 U2 AF A U4/U6–U5
B GU AG GU
U5 A snRNP
U5
Brr2/SNRNP200ATP
Snu114/EFTUD2GTP NTC NTR 3′ss
5′ss
U4 Formation of
active site U4 snRNP
H. sapiens U2 snRNP
AG

U2 6ff7
AG
Activation

A
U2 A U6 NTC
Bact U6 SF3
NTC GU
GU NTR
U5 U5 U6–U5 snRNP NTR
Prp2/DHX16ATP
Branching BP positioning *snRNAs
SF3 factors not shown S. cerevisiae
(YJU2, ISY1) First step
6j6n NTC
AG

AG A A2′OH
U6 U2 NTC NTR
B* GU NTRA P GU
U6–U5 snRNP
U5 U2 snRNP
EJC
SNRNP200
H. sapiens
Branching 5yzg
EJC
Catalysis

U5 EJC
U6 NTRNTC U
PG U2 snRNP
C A U2 AG A AG
3′OH 3′OH U6–U5 snRNP NTC
NTR
Prp16/DHX38ATP Exon ligation
Branching factors (CACTIN,
SDE2, NKAP) H. sapiens
factors 3′ss–5′ss docking 5mqf
NTC Second step U6–U5 snRNP NTR
EJC NTR
AG GU

U5 U6 U2 P P
C* A EJC
AG

3′OH
A 3′OH NTC
U2 snRNP

H. sapiens
Exon ligation 6qdv EJC
EJC U6–U5 snRNP
NTC
P U5 U6 U2 NTR
GU
Release

P P
A NTC
U2 snRNP
Prp22/DHX8ATP Intron release
NTR

ILS U2 NTC
U6 NTR Prp43/DHX15ATP mRNA Intron lariat mRNA
U5

Nature Reviews Genetics

Review article
Fig. 1 | The splicing mechanism and the spliceosome. Pre-mRNA splicing complexes so far available in different organisms2–4,6,34,95,261–267. The structures
involves the identification of intron–exon boundaries (splice sites) and two with the highest resolution available are shown (Protein Data Bank (PDB) codes
successive transesterification reactions (catalytic steps) (see Box 1 for details). indicated at the top left of each). Structures from Saccharomyces cerevisiae
The spliceosome comprises 5 small nuclear ribonucleoprotein (snRNP) are shown for complexes that have not yet been determined for Homo sapiens
complexes (U1, U2, U4, U5 and U6) and more than 150 additional proteins, (for example, the E and A complex structures). For simplicity and owing to space
which together recognize the splice sites, bring them together and catalyse constraints, a recently described pre-Bact complex has not been included268
intron removal4,6. Left-hand column: the dynamics of spliceosome assembly and additional snapshots are likely to emerge from future work, including those
and the exchanges of snRNPs and other factors, which are driven by ATP- corresponding to spliceosome proofreading mechanisms. See Supplementary
consuming RNA helicases — (yeast/human) Prp5/DDX46, Uap56/DDX39B, Tables 1–3 for a full list of spliceosome components in H. sapiens, S. cerevisiae
Prp28/DDX23, Brr2/SNRNP200, Snu114/EFTUD2, Prp2/DHX16, Prp16/DHX38, and Schizosaccharomyces pombe, and see Supplementary Table 4 for details of all
Prp22/DHX8 and Prp22/DHX15 — that can resolve kinetic traps along the published cryogenic electron microscopy structures of spliceosome complexes.
pathway to spliceosome activation, catalysis and product release. Central BP, branch point; EJC, exon junction complex; NTC, Prp19 (NineTeen) complex;
column: key RNA–RNA interactions that occur during the process. Right-hand NTR, ntc-related complex; SF1, splicing factor 1; snRNA, small nuclear RNA;
column: snapshot cryogenic electron microscopy structures of the different 3′ss, 3′ splice site; 5′ss, 5′ splice site; ILS, intron lariat spliceosome.

isoform of BCL-X in cancer cells59 or the repression of a cryptic 5′ss in partially retained introns78. The activation of cryptic 3′ss has been
lamin A, which becomes activated in Hutchinson–Gilford’s progeria60. linked to a reduced interaction of mutant SF3B1 with the splicing
factor SUGP1 (ref. 79).
3′ splice site recognition One key question relates to how transcriptome changes induced
Recognition of the 3′ end of introns in higher eukaryotes is initiated by mutations in factors that recognize 3′ss can influence tumour
by the cooperative binding of three interacting proteins — splicing progression, particularly considering that the same mutation, for
factor 1 (SF1; also known as branch point-binding protein (BBP)) and example K700E in SF3B1, correlates with worse prognosis in chronic
the U2AF heterodimer (U2AF1–U2AF2) — to three adjacent sequence lymphocytic leukaemia but with better prognosis in myelodysplastic
motifs, namely the branch point (BP), polypyrimidine tract (PPT) and syndrome72–74, or with shorter or longer overall survival depending
3′ss (Figs. 1 and 2). Although distance constraints determine the use of on the melanoma class75,76. Pan-cancer splicing analysis and positive-
a particular 3′ss upon recognition of the BP, a given 3′ss is frequently enrichment CRISPR screening showed that the effects of various SF3B1
associated with more than one functional BP61. Mutations in U2AF1 mutations converge on repression of BRD9, a core component of the
that have been identified in various types of cancer, including myeloid non-canonical BAF chromatin remodelling complex that is a potent
malignancies and lung adenocarcinomas, alter the specificity of 3′ss tumour suppressor for uveal melanoma, through activation of a
recognition such that different mutations enhance binding to and ‘poison exon’ that introduces a premature termination codon and leads
selection of 3′ss that have specific nucleotides flanking the conserved to degradation of BRD9 mRNA80 (Fig. 4b). Other splicing alterations
3′ss AG62–64. For example, whereas S34F/Y mutants of U2AF1 favour associated with SF3B1 mutations are also likely to contribute, including
the inclusion of exons harbouring CAG 3′ss and disfavour the inclu- for example increased expression of telomerase RNA (and telomerase
sion of exons harbouring UAG 3′ss, Q157P/R mutants of U2AF1 favour activity) or decreased expression of the MAP3K7 kinase (which is
the inclusion of exons with AG/G 3′ss and promote skipping of exons related to increased NF-κB signalling) or of the haem transporter ABCB7
with AG/A 3′ss63. One study found that U2AF1 mutations directly affect (which is relevant to sideroblastic anaemia)81–83.
stress granule components and responses65. Mutations in U2AF2 have
also been found in cancer samples and correlate with reduced bind- Therapeutic targeting
ing to PPTs66. These results illustrate how modulating the binding of Surprisingly, various families of small molecules that inhibit the confor-
core splicing factors, which are generally required for splicing of most mational change of SF3B1 during BP recognition at the 3′ end of introns
introns, can be rate-limiting for splice site selection. have anti-proliferative effects in vitro and inhibit tumour growth in vari-
SF1 is subsequently replaced by U2 snRNP at the BP (Fig. 1) and ous mouse cancer models18,67–69,84. One of these (H3B-8800) is currently
the U2 snRNP proteins SF3B1 and PHF5A have a key role in BP recog- in clinical trials for myelodysplastic syndrome71 (Fig. 4b). Different
nition (Fig. 2a). A major rearrangement of SF3B1, from an open to a BP sequences have differential sensitivity to these compounds, with BP
closed conformation, is triggered by recognition of the pre-mRNA, with sequences that have more extensive base pairing with U2 snRNA being
the HEAT repeats domain of SF3B1 establishing specific contact with the more resistant to their effects. As a result, these compounds can induce
adenosine at the BP sequence, which is sandwiched between SF3B1 and changes in splice site selection, rather than general splicing inhibition,
PHF5A (refs. 67–69) (Fig. 2b). As occurs for multiple other transitions in at concentrations that have cytostatic rather than cytotoxic effects85–88.
the spliceosome cycle, an RNA-dependent helicase (PRP5; associated Treatment with one such compound, pladienolide B, reprogrammes
with U2 snRNP) provides a mechanism for proofreading, ensuring mouse pluripotent cells into totipotent blastomere-like cells that
that proper recognition of the BP has been achieved within the closed can be cultured stably in vitro19, suggesting that SF3B1 inhibitors can
conformation of SF3B1 (ref. 70). exert specific effects on various biologically relevant programmes of
Mutations in SF3B1 are common in various types of tumour62,71. post-transcriptional regulation. Different structural variants of these
They occur in 81% of patients with a class of myelodysplastic syn- small molecules induce alternative splicing changes that are only par-
drome having perinuclear iron accumulations known as ring sidero- tially overlapping89, which suggests that slight modifications to their
blasts72; the SF3B1 K700E mutation is among the most common single chemical structures might generate drugs of improved specificity.
mutations detected in any gene in patients with chronic lymphocytic Cancer cells, particularly those with mutations in SF3B1, seem to
leukaemia73,74; and SF3B1 mutations are detected in 15–36% of eye be more sensitive to compounds that inhibit SF3B1 than are non-cancer
melanomas75,76. SF3B1 mutations are associated with changes in alterna- cells71. These observations led to the concept that although cancer cells
tive splicing of numerous genes, and involve a characteristic pattern can tolerate marked alterations in their transcriptomes (for example,
of activation of cryptic 3′ss 10–30 nucleotides upstream of canonical induced by mutations in splicing factors) that contribute to tumour
3′ss (at least in some cases associated with the use of an alternative progression, as a result they become more susceptible to further per-
BP77), intron retention and, intriguingly, enhanced splicing of certain turbations of the splicing process. A similar synergistic effect (known

Nature Reviews Genetics

Review article

Box 2

Functions of alternative splicing

Different combinations of binary splice site choices produce mRNAs is translated to different protein isoforms303 (see the figure,
various classes of alternative splicing events, which are observed part b). Systematic analyses of the effects of alternative splicing
at different frequencies in the human transcriptome. These on protein–protein interaction networks are also more compatible
include cassette exons or mutually exclusive exons that can be with alternative splicing having widespread effects on protein
included or skipped, the use of alternative splice sites associated and cellular function304–309. Alternative splicing tends to affect
with the use of alternative promoters or polyadenylation sites disordered protein domains, which are often involved in functionally
(giving rise to alternative first or last exons, respectively), the use important protein–protein interactions304,310,311. Protein isoforms
of alternative 5′ or 3′ splice sites (5′ss or 3′ss) within exons, the can also have differences in stability, localization, enzymatic
retention of certain introns and reverse splicing reactions (back activity and protein–nucleic acid interactions. Also of relevance
splicing) that generate circular RNA molecules (see the figure, to the functions of alternative splicing, 30% of alternative splicing
part a). However, the extent to which alternative splicing affects events introduce premature termination codons that can trigger
protein and/or cell function or, rather, represents a by-product nonsense-mediated decay (NMD) and other mechanisms of RNA
of transcriptome noise is unclear. There are many examples of degradation and, therefore, alternative splicing often functions to
alternative splicing generating protein isoforms that are relevant control mRNA abundance312–315. The exon junction complex (EJC),
for cellular or organismal phenotypes, disease progression or which is deposited on the mRNA on completion of splicing, has a
the ecology of organisms, ranging from apoptotic switches to role in translation-coupled nonsense-mediated decay as well as
opioid analgesia, from neural function to sexual behaviour or seed impairing cryptic splice site usage316. A pooled CRISPR–Cas9 screen
edibility (reviewed in refs. 299,300). The question is whether these assessing the relevance of such ‘poison exons’ that cause premature
functional examples are the norm or, rather, are exceptions among termination of translation showed that, for 50% of the tested exons,
the hundreds of thousands of alternatively spliced transcripts that deletion had effects on cancer cell viability and xenograft growth,
exist. Whereas some proteomic studies detect mainly a single with a subset of these exons having tumour suppressor activity317.
isoform expressed at the protein level in most tissues301, or a limited These results suggest the widespread functionality of splicing
number of tissue-specific isoforms in specific protein families302, regulation, even in cases where alternative splicing does not
other studies suggest that a major fraction of alternatively spliced generate alternative protein products.

a Cassette exons Mutually exclusive exons b

Gene

Alternative first exons Alternative last exons Ribosome

mRNA
isoforms
Alternative 5′ splice sites Alternative 3′ splice sites

Premature termination codon

Retained introns Back splicing

NMD Protein
isoforms

Effects on stability, localization,

enzymatic activity and interactions
with other proteins or nucleic acids

Nature Reviews Genetics

Review article

as synthetic lethality) applies to the cell cycle inhibitor indisulam and Recent results suggest an interesting additional use of splicing factor
related compounds, which enhance degradation of the U2AF2-related inhibitors such as pladienolide B or indisulam in oncology. Previous work
factor RBM39, resulting in alternative splicing changes that have cyto- has shown that splicing alterations in cancer can lead to the production of
toxic effects in haematological malignancies, particularly in those that tumour-associated neoantigens, for example through the activation
already have mutations in other splicing factors such as SF3B1 or U2AF of cryptic splice sites that introduce in-frame or out-of-frame novel amino
(refs. 90,91) (Fig. 4b). acid sequences in protein-coding genes92,93. This, in turn, can enhance the

a Interactions mediating exon definition

?

U1 snRNP
UAP56
SF3A1
U2 snRNP RBM39
U2AF2 U2AF1
SF3B1 PHF5A ESE/ESS
ISE/ISS

BP PPT 3′ss 5′ss

+/– +/–
+/– +/–

6qx9 Sm ring 6qx9 6qx9 6qx9

SF3B complex AdML pre-mRNA U1A
AdML pre-mRNA U1C
U2 SF3B1
snRNA
U1 snRNA

5′ss
Bulged out 5′ss U1-70K
SF3A complex Sm ring
BP region U2 snRNA BP adenosine U1 snRNA
U2 snRNP on BP region (pre-B complex) U2 snRNA base pairing to BP region U1 snRNA base pairing to 5′ss U1 snRNP on 5′ss (pre-B complex)

b
7q4o 7onb 6hmi 6hmo
-1 A -1 A
SMN2 5′ss SMN2 5′ss
SF3B1 SF3B1 region region
PHF5A
PHF5A

Risdiplam
Bulged out SSA
U2 snRNA BP adenosine U2 snRNA 5′ end of U1 snRNA 5′ end of U1 snRNA

Recognition of BP–U2 snRNA by PHF5A–SF3B1 (closed conformation) and Interaction between U1 snRNA and SMN2 exon 7 5′ss and enhancement
inhibition by SSA (open conformation) by risdiplam
Fig. 2 | Early splice site recognition and exon definition. a, Base-pairing between U1 and U2 snRNP complex components have been proposed to mediate
interactions between pre-mRNA sequences (the branch point (BP) and intron (and possibly also exon) definition and splice site pairing (see main
5′ splice site (5′ss)) and small nuclear RNA (snRNA) components of small nuclear text for details); additional stabilizing interactions across exons and introns
ribonucleoprotein (snRNP) complexes (U2 and U1, respectively) are crucial for are likely. The four structural snapshots illustrate the principles of 3′ss and
the definition of intron–exon boundaries. These interactions are assisted by 5′ss recognition by U2 and U1 snRNPs, respectively, and the role of RNA–RNA
proteins of the snRNP complexes and other auxiliary factors — such as U2AF2, interactions in these processes. b, Structural snapshots showing the changes
which binds to the polypyrimidine tract (PPT), and U2AF1, which binds to the in conformation of U2 and U1 snRNP complex components upon binding to
3′ splice site (3′ss) adenosine–guanosine (AG)4,6. Various sequences can function spliceostatin A (SSA) and a risdiplam analogue. SSA belongs to a family of
as splice sites (represented by sequence logos), their strength generally splicing inhibitors with antitumour properties that prevent the transition of the
correlating with their potential to base pair with U2 or U1 snRNAs and with the SF3B1–PHF5A complex that mediates BP recognition from an open to a closed
length and uridine-richness of the PPT. Intronic and exonic splicing enhancers (ISEs conformation. Risdiplam is a small molecule that stabilizes the interaction
and ESEs) and intronic and exonic splicing silencers (ISSs and ESSs) are recognized between U1 snRNA and the 5′ss of SMN2 exon 7, facilitating exon inclusion and
by regulatory factors (shown in orange), such as SR proteins, heterogenous the production of functional SMN2 protein as a therapy for spinal muscular
nuclear ribonucleoproteins (hnRNPs), RBM or CELF proteins, that enhance or atrophy. Protein Data Bank (PDB) codes for each structure are indicated at the
inhibit the association of U1 and U2 snRNPs with the splice sites. Interactions top left of each box.

Nature Reviews Genetics

Review article

Consensus splice sites delineated Computational assessment of splicing regulation from primary
by SpliceSiteFinder algorithm269 1987
RNA sequence: early algorithms and classical machine
learning tools
Improved splice site identification Computational assessment of splicing regulation from primary
by HMM NNSPLICE270 1997
RNA sequence: deep learning networks and related methods
High-throughput mutagenesis assays
Dependencies between splice site
bases captured by GeneSplicer271 2001
High-throughput assessment of natural and disease-
associated sequence variants
Discrimination of bona fide splice Splice site identification
sites by maximum entropy using idlBNs273 2004 Benchmarking studies
modelling272

2009

2011 Saturation mutagenesis reveals dense splicing

regulatory content of exons114,139,140,284
First sequence-based model for
tissue-specific splicing prediction274 2010

Impact of human genetic variants 2 million reporters reveal effects of SNPs in

assessed by Bayesian neural network275 2015
exons and sequences flanking splice sites138

2016
Complete genotype–phenotype landscape
of alternative exons reveals scaling law for
the effects of mutations126–128
2017
Exon skipping predicted by combining Bayesian and
deep neural networks with experimental data276,277 Splicing defects of disease-causing exonic
mutations revealed by MapSy131,144
Splice sites predicted by Splice site competition
convolutional neural networks modelled by four deep Quantification of Library of random Splicing effects
and biological information learning architectures 2018 >32,000 5′ splice sites mutations reveals of >2,000 human
(SpliceRover)278 (COSSMO)279 reveals effects of the effects of genetic variants
context and of cancer revealed by
Critical Assessment MMSplice (neural Genetic variant- pathogenic variants31 mutations130 Vex-seq286
of Genome network modules) induced cryptic
Interpretation wins CAGI5 exon splicing predicted
challenge (CAGI5) skipping by 32 layer-deep Saturation Library of >32,000 Splicing effects
on splicing impact of prediction neural network 2019 mutagenesis of splicing events reveals of 27,000
sequence variants287 challenge143 SpliceAI141 SMN1 exon 7 cell to cell variability genetic variants
reveals a wide and interplay between revealed by
range of splicing primary sequence and MFASS146
Deep learning-derived splice sores (CADD-Splice) alterations129 secondary structures125
predict variant effects genome-wide281,282

Comparative evaluation of algorithms to identify splicing 2020

regulatory elements in two independent data sets288

Benchmarking of prediction tools Tissue-specific effects of 20,000 intron variants in yeast

on real-time clinical settings reveals genetic variants in GTEx 2021 reveal effects on splicing efficiency
need for improved algorithms145 predicted by MTSplice280 and evolutionary adaptation285

Tissue-specific splicing and effects of mutations 9,000 synthetic introns reveal Mutagenesis reveals CD19
predicted by neural network Pangolin283 2022 features required for cancer splicing variants relevant
cell-specific splicing260 for immunotherapy289

immune surveillance of cancer cells. Thus, small-molecule splicing mod- of splicing can therefore be combined with therapies that prevent the
ulators can enhance the production of tumour neoepitopes that trigger inhibition of T cell-mediated immune responses — ‘checkpoint blockers’
effective antitumour immune responses. Pharmacological modulation such as antibodies to PD1 — to enhance their effects22 (Fig. 4b).

Nature Reviews Genetics

Review article

Fig. 3 | A timeline of key events in cracking the ‘splicing code’. How does a cell In the past four decades, several computational and high-throughput
distinguish between exons and introns? How does one cell type decide that a experimental approaches have been developed to crack the ‘splicing code’ — in
particular sequence should be included in the mature mRNA whereas another other words, to identify from genomic sequences alone bona fide splice sites
cell type decides to skip it? How is this achieved given that the sequences at and their differential use31,114,125–131,138–141,143–146,260,269–289. See Supplementary Table 5
intron boundaries (5′ splice site (5′ss) and 3′ splice site (3′ss)) are highly diverse for further details of these studies. GTEx, Genotype-Tissue Expression database;
in multicellular organisms — with the exception of GU/C at the 5′ end of the idlBNs, inclusion-driven learned Bayesian networks; SNP, single nucleotide
intron and adenosine–guanosine (AG) at the 3′ end of the intron (Fig. 2)? polymorphism.

Similarly to the regulation of 5′ss recognition, antisense oligonu- Role of regulatory sequences
cleotides targeting 3′ss also have potential therapeutic applications. Exon sequences themselves can also contribute to splice site recog-
For example, blocking activation of the poison exon in BRD9 that is nition and exon definition through the function of exonic splicing
activated by SF3B1 mutations suppresses tumour growth80, and tar- enhancers (ESEs) and exonic splicing silencers (ESSs) (Fig. 2). Together
geting a 3′ss in the oncogene ERG inhibits the proliferation of prostate with ISEs and ISSs, these regulatory sequences are thought to nucle-
cancer cells94 (Fig. 4b). ate the assembly of complexes of regulatory factors that promote or
inhibit splice site recognition by the core splicing machinery108,114,115
Splice site communication (Fig. 2). This can be achieved through various mechanisms, including
5′ss and 3′ss necessarily need to pair (‘commit’) to each other for the the recruitment of core splicing factors through direct interactions with
splicing process to occur. Assembly of the U4/U6–U5 tri-snRNP on these regulatory complexes116,117, the establishment of exclusion zones
pre-mRNAs at positions where 5′ss and 3′ss are recognized by U1 and through cooperative coating of the RNA by RNA binding proteins118,119,
U2 snRNPs, respectively, establishes such pairing through multiple interference with specific interactions mediating exon and/or intron
RNA–RNA, RNA–protein and protein–protein interactions with the definition120 or the formation of higher-order assemblies (possibly even
pre-mRNA and/or early splicing factors6,95 (Fig. 1). Connections between experiencing local phase transitions) involving tyrosine-rich intrinsi-
5′ss and 3′ss can, however, also occur before tri-snRNP assembly and cally disordered protein domains that are themselves regulated by
such interactions are thought (although not yet fully proven) to stabi- alternative splicing121,122. Classical examples include proteins of the
lize complexes formed on the splice sites across the intron. Examples arginine–serine-rich (SR) family, which have positive effects on splic-
of such connections involve interactions between U1 snRNP and the ing from exonic enhancers, and proteins of the heterogenous nuclear
3′ and 5′ domains of U2 snRNP34; the interaction of stem–loop IV of U1 ribonucleoprotein (hnRNP) family, which inhibit spliceosome assembly
snRNA (which is essential for splicing) with a non-canonical RNA bind- from intronic silencers108,115. However, often the same sequence motif
ing domain in the U2 snRNP protein SF3A1 (refs. 96,97); the interaction and cognate factors can have positional effects, for example promot-
of stem–loop III of U1 snRNA with the U2AF-associated RNA helicase ing exon skipping when bound upstream of an alternative exon but
UAP56 (ref. 98); and the interaction of SF1 with the U1 snRNP-associated promoting exon inclusion when bound downstream108,115,123–125.
protein Prp40p in yeast99,100 (Fig. 2). It is currently unclear whether these Not unexpectedly, exon mutations have broad effects on alterna-
are the main contacts for early splice site communication or whether tive splicing. Recent high-throughput, saturation mutagenesis studies
multiple other molecular bridges can be formed on different introns. have shown that two thirds of all possible mutations in an alternative
An important complementary concept posits that some of these exon can affect its inclusion, whereas this was not the case for constitu-
interactions might occur not only between splice sites across introns tive exons126–130. This has obvious implications for the joint evolution
but also across internal exons through a process known as exon of splicing and protein codes131,132 and for understanding the effects of
definition101,102 (Fig. 2). The mutual stabilization of splice site recognition synonymous mutations in natural genetic variation associated with
complexes across internal exons can explain various long-standing genetic diseases133–135 as well as in cancer136,137.
observations, including the upper (250 nucleotides) and lower Several recent studies have aimed to systematically assess the
(50 nucleotides) length constraints of these exons (which define the effects of exonic or proximal intronic sequence motifs on splice site
boundaries for optimal exon definition interactions) and the fact that selection using high-throughput read-outs (Fig. 3; see Supplementary
genetic mutations in splice sites can induce exon skipping (as a con- Table 5). For example, one study assessing and modelling the effects of
sequence of the failure of exon definition) instead of intron retention random libraries of 25 nucleotides flanking 5′ss or 3′ss (involving more
(which would be the consequence of a failure of intron definition)103,104. than 2 million synthetic minigenes) showed that the vast majority of
It might also explain the results of saturation mutagenesis of 5′ss, possible hexamer sequence motifs influence splice site selection, hav-
whereby the strength of the upstream 3′ss was found to influence the ing similar effects in 5′ss or 3′ss competition assays138. This large vari-
effects of 5′ss mutations31. ability of sequence motifs is consistent with results from other reports,
An important exception to the length constraints of internal exons although these studies found differential positional effects of ESSs but
are microexons, which are ~3–27 nucleotides in length but have impor- not ESEs139,140. Other studies have also highlighted the importance of
tant regulatory effects on protein functions in nervous system devel- genomic context and of starting levels of exon inclusion on the effects
opment, synaptic transmission and autism spectrum disorder105–107. of mutations and in the generation of transcriptome complexity during
Microexons have evolved specific mechanisms of recognition coordi- evolution127,141,142, as well as the contribution of various inputs to splice
nated by the neuron-specific regulatory protein SRRM4, which functions site selection in most native contexts125 (Fig. 3; see Supplementary
as a master regulator for this programme108–110. Master regulatory factors Table 5).
have been described to coordinate other programmes of splicing regula- These efforts have obvious relevance for predicting the effects
tion, for example for sex determination in fruit flies, shaping synapses of potentially pathogenic mutations, and various strategies have
or coordinating epithelial–mesenchymal transition in vertebrates111–113. been envisioned to assist in genetic counselling by modelling and/or

Nature Reviews Genetics

Review article

a b
Spinal muscular atrophy Engineered Myelodysplastic syndrome and cancer
U1 snRNA
U1 snRNP Indisulam

6 7 8 SMN2 6 7 8 U2 snRNP
RBM39
ISS Production of
Risdiplam Production of neoantigens and
Nusinersen functional protein increased anti-
SF3B1
tumour response
Cystic fibrosis
12 13 14 12 13 14 Pladienolide B SSA/Sudemycins H3B-8800 + Anti-PD1
CFTR
15 16 17 15 16 17

Rescue of functional Antisense oligonucleotide

Engineered
U1 snRNA protein production
STOP codon
14 14a 15 BRD9 14 15
CDKL5-deficiency disorder Inhibition of cancer
2 3 4 CDKL5 2 3 4 cell growth
Antisense oligonucleotide

Engineered Rescue of functional

U1 snRNA protein production 3 4 5 ERG 3 5
Inhibition of
Familial dysautonomia cancer cell growth
c
Kinetin RECTAS
Duchenne muscular dystrophy
SRSF6
48 49 50 51 52 Dystrophin 48 52
19 20 21 IKBKAP 19 20 21
ISS ISE ESE
Production of a
Engineered Antisense Rescue of functional functional shorter
oligonucleotide protein production Eteplirsen protein isoform
U1 snRNA

Fig. 4 | Therapeutic targeting of splice site recognition and exon definition. interventions relevant in oncology include the production of neoantigens by
Examples of how small-molecule drugs, antisense oligonucleotides or inducing inhibition of 3′ss recognition using the indicated SF3B1-targeting
engineered U1 small nuclear ribonucleoprotein (snRNP) complexes can be used or RBM39-targeting drugs; neoantigens in combination with immune
as potential or current therapies for genetic diseases or cancer. a, Targeting checkpoint blockade (such as anti-PD1 therapy) can elicit immunotherapy
5′ splice site (5′ss) recognition. Examples are provided of four diseases in which responses. Other examples include blocking 3′ss recognition by antisense
promoting 5′ss recognition can enhance the inclusion of exons and lead to the oligonucleotides, leading to exon skipping events that inhibit cancer cell growth.
synthesis of functional proteins whose production was disrupted in the indicated c, Targeting regulatory sequences. Antisense oligonucleotides (including
pathologies. This is achieved by using engineered U1 snRNP (in which the 5′ eteplirsen) targeting exonic enhancers induce skipping of exons containing
end sequence of U1 small nuclear RNA (snRNA) has been modified to target the inactivating mutations in the Dystrophin gene that cause Duchenne muscular
complex to specific locations within a transcript), antisense oligonucleotides dystrophy, leading to in-frame deletion and production of a shorter protein
(including nusinersen) targeting intronic silencers or small molecules (see text isoform that rescues function. ESE, exonic splicing enhancer; ISE, intronic
for details). b, Targeting 3′ splice site (3′ss) recognition. Examples of therapeutic splicing enhancer; ISS, intronic splicing silencer; SSA, spliceostatin A.

experimentally assessing splicing perturbations in genes of inter- for interrogating gene and protein function without requiring complex
est141,143–145. For example, a high-throughput system designed to test genome engineering.
the effects of more than 27,000 variants annotated in the Genome
Aggregation Database found that very rare variants had large effects Co-transcriptional regulation
on splicing, mostly located outside the splice sites themselves146. Long-standing evidence ranging from electron microscopy148 to tran-
scriptome analyses of RNAs physically associated with transcribing
Therapeutic targeting RNA Pol II149 indicates that the removal of introns from pre-mRNAs can
The antisense oligonucleotide eteplirsen, which targets ESE sequences, occur co-transcriptionally, even almost immediately after the 3′ss exits
has been approved as a therapy for Duchenne muscular dystrophy. from the polymerase tunnel (reviewed in refs. 150,151) (Fig. 5). Although
Eteplirsen promotes skipping of an exon in the dystrophin gene that in vitro-transcribed model pre-mRNAs can be spliced in nuclear extracts
harbours inactivating mutations, leading to in-frame production of a or upon transfection or injection (for example, in the nucleus of
shorter, but still functional, protein that restores muscle function147 Xenopus oocytes)152, which indicates that splicing can be uncoupled from
(Fig. 4c). More generally, approaches targeting exon sequences may be transcription, co-transcriptional splicing has important mechanistic
used to generate truncated protein variants lacking specific domains, implications, as functional connections between the transcription and

Nature Reviews Genetics

Review article

splicing machineries can enhance splicing efficiency and influence has been questioned163. Results from long-read sequencing of nascent
splice site choice153,154. Transcription elongation rates determine the pre-mRNA transcripts indicate that splicing in human and Drosophila
time window during which alternative splice sites enter into competi- cells typically occurs after RNA Pol II has transcribed several kilobases
tion (reviewed in refs. 153,155,156), which can influence splice site choice of pre-mRNA, with the order of intron removal not following strictly
via a kinetic model that has been shown to operate both in vitro and the order of transcription164. Another study combining similar tech-
in vivo, in animals and in plants157–159 (Fig. 5a). Conversely, splice site nologies with precision run-on sequencing found that, during mouse
recognition can influence promoter choice, transcription elongation erythropoiesis, although introns are often spliced during the time of
and proper 3′ end formation154,160–162. transcription of the downstream intron, nascent transcripts with a high
However, more recently, the extent to which co-transcriptional proportion of unspliced introns are also detected162. Along the same
splicing occurs and/or contributes to the majority of splicing events lines, a three-pronged methodology to characterize nascent RNAs

b c
a
Chromatin-mediated Preferential nucleosome
Faster rate of elongation allows for stronger Slow rate of elongation allows recruitment of positioning at internal
splice site to be transcribed and used for use of weaker splice site splicing factors exons

PTB

MRG15 H3K36me3
U1 snRNP U1 snRNP
RNA Pol II RNA Pol II

Nucleosomes

U2 U2
SF3 SF3
U1 Weaker
splice site d
Stronger U1 CTD-mediated recruitment
splice site of splicing factors

CTD
Weaker PRP19 YSPTSPS
splice site heptad repeats
U2AF
U2
U1 RNA Pol II
SF3

U1 snRNP

5′ss Pre-mRNA
U2
SF3
U2 3′ss
U1 SF3 U1
5′ss 3′ss 5′ss

Fig. 5 | Co-transcriptional splicing. Intron removal can occur on nascent mRNA exons may favour exon definition291–293. d, There is evidence to suggest that
transcripts during transcription by RNA polymerase II (RNA Pol II), which has splicing factors (such as U2AF–PRP19)294 may be recruited through direct
important mechanistic implications for splice site selection. a, The elongation interactions with RNA Pol II, including its carboxy-terminal domain (CTD).
rates of RNA Pol II can affect the competition between alternative splice sites. The CTD is composed of multiple repeats of the heptad amino acid sequence
For example, slow rates of elongation facilitate the use of weaker splice sites YSPTSPS harbouring potential phosphorylation sites that are linked to different
before stronger splice sites are transcribed155. b, Chromatin can recruit splicing elongation states of the enzyme156 and to a switch between transcriptional and
regulators (such as PTB) through proteins that recognize epigenetic marks, splicing condensates178. Importantly, the interaction between U1 small nuclear
such as MRG15, which binds both PTB and the trimethylation of histone H3 on ribonucleoprotein (snRNP) and RNA Pol II requires a 5′ splice site (5′ss) being
lysine 36 (H3K36me3)186,290. c, Preferential nucleosome positioning at internal present in the nascent transcript253. 3′ss, 3′ splice site.

Nature Reviews Genetics

Review article

also documented examples of co-transcriptional splicing occurring at the adenosine–guanosine (AG) dinucleotide of 3′ss inhibits its recog-
immediately after 3′ss transcription, as well as other examples in nition by U2AF1, leading to the retention of an intron in the pre-mRNA
which splicing was delayed until RNA Pol II had transcribed sequences encoding S-adenosylmethionine (SAM) synthetase191. This results in
much further downstream165. Nevertheless, clear examples of post- the downregulation of SAM synthetase expression as part of a negative
transcriptional splicing do exist. These include examples of coordi- feedback loop by which an excess of methionine triggers accumulation
nately regulated distant alternative splicing events within the same of methylated SAM, which itself functions as the methyl donor for the
transcript in different cell types166,167, and the recently reported category enzymatic deposition of m6A on pre-mRNA. A similar mechanism oper-
of detained introns, excision of which can be induced by signalling cues ates in human cells192. In this case, under conditions of abundant SAM,
during development or meiosis, or in certain cancers168–171. The otherwise the N6-adenosyl-methyl transferase METTL16 methylates a loop struc-
fully processed polyadenylated transcripts containing detained introns ture at the 3′ untranslated region of SAM synthetase pre-mRNA. Under
are retained at the gene locus until splicing of the detained intron allows conditions of low levels of SAM, METTL16 binds to the unmethylated
for mRNA export to the cytoplasm and translation, thus allowing for loop and enhances splicing of SAM synthetase pre-mRNA, facilitating
rapid changes in protein expression in response to external cues169–172. expression of the enzyme192. Thus, in both C. elegans and humans, m6A
A long-standing issue related to co-transcriptional splicing is functions as a sensor of the availability of metabolites to switch off the
whether splicing is subject to some form of compartmentalization expression of SAM synthetase via regulation of splicing.
in the nucleus. For example, splicing has been proposed to occur in Another mechanism by which m6A can regulate splicing is via
the vicinity of nuclear speckles, regions of the nucleus that have an proteins that recognize this modification. For example, hnRNPG (also
accumulation of splicing factors, contain active spliceosomes and known as RBMX) recognizes m6A modifications at exonic positions
have phase transition properties (reviewed in refs. 168,173–177). This close to regulated splice sites and, through interactions with RNA
is particularly relevant considering that different phosphorylation polymerase, modulates alternative splicing193. Another example is
states of the intrinsically disordered carboxy-terminal domain (CTD) the nuclear m6A reader YTHDC1, which recruits splicing regulatory
of RNA Pol II can drive an exchange from condensates involved in tran- factors of the SR protein family (such as SRSF3), but antagonizes the
scription initiation to condensates in nuclear speckles178. Two recent binding of other factors such as SRSF10, with the result of promoting
studies reported two distinct areas of the nucleus, one associated inclusion of alternative exons194. The effects of YTHDC1 on splicing and
with nuclear speckles and the other with peripheral nuclear lamina, polyadenylation may underlie its essential function in germline devel-
that had characteristic patterns of splicing regulation179,180. Such pat- opment195. m6A-mediated regulation of the splicing kinetics of multiple
terns correlate with differences in the levels of splicing factors and/or introns, associated with positional effects of m6A deposition on nascent
genomic guanosine–cytosine (GC) content in these areas, the latter transcripts196, is an attractive proposed mechanism for the coordinated
being associated with distinct genomic architectures that have been control of developmental programmes. However, the extent to which
linked to exon versus intron definition179–181. m6A modifications generally regulate splicing programmes remains
Chromatin organization, including nucleosome positioning and unclear, with one study arguing that their major functional effect is
epigenetic marks such as histone modifications and DNA methylation, on cytoplasmic mRNA stability197.
has also been associated with splice site recognition and regulation Recent work highlights the regulatory potential of pseudouridine
(reviewed in refs. 182,183) (Fig. 5b,c). Proposed mechanistic models modifications at alternatively spliced regions of pre-mRNAs and their
involve proteins that recognize methylated DNA and slow down RNA regulatory sequences, with direct effects on splicing efficiencies198.
Pol II, indirectly affecting splice site selection184, or adaptor proteins such The tissue-specific expression of pre-mRNA pseudouridine synthases
as MRG15 that recognize both histone tail modifications and splicing thus offers another potential mechanism for the control of alternative
factors, thus increasing the local concentration of splicing regulators splicing and 3′ end formation198.
such as PTB185 (Fig. 5b). Proving causality, a recent study used genome Functionally important RNA modifications also occur in snRNAs,
editing tools to introduce histone modifications that are observed dur- which have 2′-O-methyl and pseudouridylated residues at phylo-
ing epithelial–mesenchymal transition at specific genomic locations, genetically conserved positions, in addition to characteristic cap
which induced corresponding splicing modifications and concomitant structures at their 5′ ends (2,2,7-trimethyl-guanosine for U1, U2,
biological effects186. U4 and U5 snRNAs and γm-guanosine for U6 snRNAs) (reviewed in
It is clear that splicing regulation can be influenced by chromatin ref. 199). Some of these modifications have been shown to be important
spatial organization and epigenetic modification, as further illustrated for snRNP biogenesis and/or for efficient splice site recognition, and
by the recent report of enhanced effects of combining the splicing an additional m6A modification in U2 snRNA has been proposed to
modu­latory drug nusinersen with a histone deacetylase inhibitor187. modulate 3′ss choice199,200.
Future work will establish how general and diverse are co-transcriptional
mechanisms influencing the splicing code. Regulation by RNA structure
One difficulty in assessing the functional effects of RNA structure on
Epitranscriptomic regulation alternative splicing is that RNAs exist, almost invariably from birth,
More than 70 chemical modifications of RNA molecules have been as RNP complexes in which the associated proteins strongly influ-
described in eukaryotes and, for some of these, dedicated protein ence the conformation(s) that RNAs adopt during or after folding.
factors involved in their deployment, reading or erasing are known188. Although methods for the high-throughput analysis of higher-order
Such epitranscriptomic modifications might establish a regulatory transcriptome structure in living cells (reviewed in ref. 201) remain to
code on RNA189, resembling the epigenetic code on DNA and his- be fully exploited in investigating splicing regulation, recent studies
tones that functions to recruit or inhibit enzymatic complexes that argue that introns may be more highly structured than exons and that
modulate transcription, replication or DNA repair190. For example, distinct RNA folding around alternative exons, depending on RNA
in Caenorhabditis elegans, N6-methyladenosine (m6A) modification Pol II elongation rates, influences splicing outcomes202,203. Additional

Nature Reviews Genetics

Review article

Glossary

Alternative splicing Exon definition MicroRNAs Small nucleolar RNAs

The process by which intron and/or A model for the mutual stabilization of A class of small regulatory RNAs whose (snoRNAs). A class of small regulatory
exon sequences are differentially splicing factors recognizing splice sites function is to induce the degradation RNAs whose function is to guide the
recognized in different cell types flanking internal exons in multicellular or repress the translation of mRNAs addition of chemical modifications
or biological conditions to generate organisms. with which they have full or partial at specific residues in other RNAs,
distinct mRNAs and long non-coding complementarity, respectively. including ribosomal, transfer or small
RNAs from the same primary transcript. Intron definition They are often transcribed as part of nuclear RNAs (snRNAs). They are
A model for the mutual stabilization intronic sequences, from which they often transcribed as part of intronic
Back splicing of splicing factors recognizing the are released to be assembled with sequences, from which they are
The process by which a 5′ splice site splice sites across an intron, which likely specific proteins on microRNA-induced released to be assembled with specific
is spliced to a 3′ splice site located has a major role in the efficient silencing complexes. proteins on small nucleolar RNP
upstream in the same pre-mRNA co-transcriptional splicing of many complexes.
molecule, leading to the generation introns. Premature termination codons
of a circular RNA, typically spanning Translation termination codons in Spliceosome
one or a few exons. Introns mRNA arising from single-nucleotide The molecular machinery involved in
Internal sequences within primary mutations or from alternative intron removal, composed of 5 small
Branch point transcripts produced by eukaryotic splicing events that disrupt an open nuclear ribonucleoprotein (snRNP)
(BP). An intronic adenosine nucleotide, RNA polymerase II (RNA Pol II) that are reading frame, often leading to complexes (U1, U2, U4, U5 and U6)
typically located 15–45 nucleotides 5′ removed through the process of pre- mRNA degradation by the process and more than 150 accessory proteins.
of the 3′ end of introns, which engages mRNA splicing, allowing their flanking of nonsense-mediated decay.
in formation of a 2′–5′ phosphodiester sequences (exons) to be spliced
bond with the 5′ end of the intron after together and thus generate functional Recursive splicing
the first catalytic step of the splicing mRNAs and long non-coding RNAs. The sequential excision of shorter
reaction. pieces of a long intron, each piece
being separated from the next
by a zero-length exon.

evidence has accumulated supporting roles for secondary RNA struc- synthesis of peptides or proteins, the general functional relevance
tures in splice site recognition (reviewed in refs. 204,205), explaining of which is under intense debate228–232.
for example temperature-sensitive splice site selection206, or bring-
ing together distant splice sites through long-range base pairing207. Conclusions and future perspectives
In mammalian cells, stem–loop structures involving splice sites or The progress reviewed above provides a valuable framework to under-
regulatory sequences, chaperoned by various RNA binding proteins, stand how the spliceosome has evolved various mechanisms to regulate
modulate alternative splicing decisions during development and are splice site selection. These range from the control of RNA structure or
of relevance for potential therapeutic approaches in spinal muscular RNA modifications to the spatial organization of genes in the nucleus,
atrophy and tauopathies (neurodegenerative disorders characterized from tight coupling between transcription, chromatin and RNA pro-
by the deposition of abnormal Tau protein in neurons)208–211. Links have cessing to the complex interactions of regulatory sequences and fac-
also been found between the function of RNPs containing structured tors that modulate exon and intron definition. However, important
RNAs and tumour biology, including G-quadruplexes recognized by challenges to our understanding of pre-mRNA splicing remain.
hnRNPF that enhance exon inclusion events212 and are relevant for A comprehensive and quantitative assessment of the isoform
cancer progression213, and a promestastatic splicing programme that is structure of individual full transcript molecules using long-read
regulated by interactions between the protein SNRPA1 and structured sequencing in single cells remains challenging233, but this will be nec-
splicing enhancers214. essary to reconstruct the spatial regulation of alternative splicing
Base-pairing interactions involving sequences flanking internal (spatial transcriptomics) and to understand precisely the contributions
exons have been shown to contribute to the production of circular of alternative splicing to tissue development and homeostasis. Linked
RNAs by facilitating back splicing between the 3′ss and 5′ss associated to this is the need for high-resolution methods to reliably assess cell to
with the looped-out exon215–220. Such sequences are often associated with cell variability in splice site selection and in the levels and/or activity of
repetitive DNA elements and their limited conservation has been regulatory factors234–237 (Supplementary Box 1). If individual cells
viewed as evidence against the general functionality of at least some of the same type diverge markedly in their alternative splicing deci-
families of circular RNAs221–223. There is, however, evidence for the func- sions125,238–244, this would call for major revision of our understanding
tional relevance of specific circular RNAs, for example as ‘sponges’ for of the molecular mechanisms of splicing regulation, which are cur-
proteins or microRNAs224–226, and a functional CRISPR screen showed rently mostly based upon the study of cell populations. For example,
that a group of circular RNAs are important for cell growth, mostly in a recent study showed that the regulation of intron excision in certain
a cell type-specific manner, or for the preimplantation development yeast ribosomal protein genes can be used to induce phenotypic het-
of mouse embryos227. Furthermore, some circular RNAs can direct the erogeneity that facilitates population adaptation to starvation or high

Nature Reviews Genetics

Review article

levels of sugar availability245. Similarly important will be to assess the 9. Wang, E. T. et al. Alternative isoform regulation in human tissue transcriptomes. Nature
456, 470–476 (2008).
levels, origins, proofreading mechanisms and potential biological 10. Nilsen, T. W. & Graveley, B. R. Expansion of the eukaryotic proteome by alternative
functions of ‘noise’ in the splicing process, for example by observing splicing. Nature 463, 457–463 (2010).
synthesis and processing kinetics of single nascent RNA molecules in 11. Baralle, F. E. & Giudice, J. Alternative splicing as a regulator of development and tissue
identity. Nat. Rev. Mol. Cell Biol. 18, 437–451 (2017).
real time246,247. In this regard, recursive splicing246,248,249 and variations 12. Zavolan, M. & Kanitz, A. RNA splicing and its connection with other regulatory layers
in the order of intron removal164,250 can have important roles in the in somatic cell reprogramming. Curr. Opin. Cell Biol. 52, 8–13 (2018).
kinetics of RNA processing and in splice site selection. 13. Scotti, M. M. & Swanson, M. S. RNA mis-splicing in disease. Nat. Rev. Genet. 17, 19–32
(2016).
Our knowledge of rate-limiting, regulatable steps in spliceosome 14. Manning, K. S. & Cooper, T. A. The roles of RNA processing in translating genotype
assembly and catalysis remains mostly based upon (painstaking) efforts to phenotype. Nat. Rev. Mol. Cell Biol. 18, 102–114 (2017).
to understand the biochemical process as it occurs in a limited number 15. Wright, C. J., Smith, C. W. J. & Jiggins, C. D. Alternative splicing as a source of phenotypic
diversity. Nat. Rev. Genet. 23, 697–710 (2022).
of pre-mRNA substrates that are detectably spliced in cell or nuclear 16. Gebauer, F., Schwarzl, T., Valcárcel, J. & Hentze, M. W. RNA-binding proteins in human
extracts, which is uncoupled from transcription and chromatin, and genetic disease. Nat. Rev. Genet. 22, 185–198 (2021).
17. Finkel, R. S. et al. Nusinersen versus sham control in infantile-onset spinal muscular
neither recapitulates the complexity of splicing decisions nor reflects
atrophy. N. Engl. J. Med. 377, 1723–1732 (2017).
cell type-specific variations in splicing, epigenetic or epitranscrip- This study provides evidence for the clinical benefit of splicing modulation in the
tomic factors. It is conceivable that cell type-specific or even substrate- treatment of a human genetic disorder.
18. Bonnal, S. C., López-Oreja, I. & Valcárcel, J. Roles and mechanisms of alternative splicing
specific spliceosomes exist, characterized by different composition,
in cancer — implications for care. Nat. Rev. Clin. Oncol. 17, 457–474 (2020).
stoichiometry and/or modifications of their components, in addition 19. Shen, H. et al. Mouse totipotent stem cells captured and maintained through
to the modulation of their function by master regulatory factors that spliceosomal repression. Cell 184, 2843–2859 (2021).
This study reveals an unexpected link between splicing activity and cell totipotency,
respond to environmental cues to shape tissue-specific transcriptomes
with potential applications in regenerative medicine.
during development251. Recent efforts to develop cryogenic electron 20. To, K. K. W. & Cho, W. C. S. An overview of rational design of mRNA-based therapeutics
microscopy methods to visualize the complexes involved in coupling and vaccines. Expert. Opin. Drug. Discov. 16, 1307–1317 (2021).
21. Black, D. L. Finding splice sites within a wilderness of RNA. RNA 1, 763–771 (1995).
between transcription and splicing252,253, in early steps of splice site 22. Lu, S. X. et al. Pharmacologic modulation of RNA splicing enhances anti-tumor immunity.
communication34,254 and in the function of higher-order suprasplice- Cell 184, 4032–4047 (2021).
osomes255, as well as tomography-based visualization of spliceosomes This study illustrates the potential of splicing inhibitors to enhance the generation
of neoantigens expressed in cancer cells.
in situ, should pave the way to a better understanding of the mole­ 23. Kondo, Y., Oubridge, C., van Roon, A. M. M. & Nagai, K. Crystal structure of human U1
cular basis of cell type-specific splicing regulation. Such efforts can be snRNP, a small nuclear ribonucleoprotein particle, reveals the mechanism of 5’ splice site
complemented by genetic analyses aimed at reconstructing networks recognition. eLife 4, 1–19 (2015).
24. Slaugenhaupt, S. A. et al. Tissue-specific expression of a splicing mutation in the IKBKAP
of splicing regulation91,109,256 and by detailed characterization of patho- gene causes familial dysautonomia. Am. J. Hum. Genet. 68, 598–605 (2001).
genic variants in pre-mRNAs and splicing factors. Other important 25. Dietrich, P. & Dragatsis, I. Familial dysautonomia: mechanisms and models. Genet. Mol. Biol.
open questions concern the extent to and mechanisms by which long 39, 497–514 (2016).
26. Shuai, S. et al. The U1 spliceosomal RNA is recurrently mutated in multiple cancers.
non-coding RNAs can contribute to the regulation of alternative splic- Nature 574, 712–716 (2019).
ing257 and how alternative splicing of long non-coding RNAs — including 27. Suzuki, H. et al. Recurrent noncoding U1 snRNA mutations drive cryptic splicing in SHH
the combination of exons across classical transcriptional units — can medulloblastoma. Nature 574, 707–711 (2019).
Shuai et al. and Suzuki et al. show that mutations in snRNAs can promote cancer
contribute to generating a large repertoire of RNA modules with progression.
possible functions in gene regulation258. 28. Roca, X. & Krainer, A. R. Recognition of atypical 5′ splice sites by shifted base-pairing
Ultimately, integrating structural and functional information to to U1 snRNA. Nat. Struct. Mol. Biol. 16, 176–182 (2009).
29. Roca, X. et al. Widespread recognition of 5′ splice sites by noncanonical base-pairing
predict patterns of alternative splicing is likely to benefit from artificial to U1 snRNA involving bulged nucleotides. Genes. Dev. 26, 1098–1109 (2012).
intelligence methods that can generate models of splice site selection259 30. Roca, X., Krainer, A. R. & Eperon, I. C. Pick one, but be quick: 5′ splice sites and the
(Fig. 3). Such approaches will have applications for understanding problems of too many choices. Genes. Dev. 27, 129–144 (2013).
31. Wong, M. S., Kinney, J. B. & Krainer, A. R. Quantitative activity profile and context
the effects of genetic variation or pathogenic mutations, as well as for the dependence of all human 5′ splice sites. Mol. Cell 71, 1012–1026.e3 (2018).
design of novel therapies to correct splicing alterations or to eliminate This study presents a systematic assessment of the activity and context dependence
of sequence variation at 5’ splice sites.
cells, such as cancer cells, that have pathogenic splicing phenotypes260.
32. Aznarez, I. et al. A systematic analysis of intronic sequences downstream of 5′ splice sites
reveals a widespread role for U-rich motifs and TIA1/TIAL1 proteins in alternative splicing
Published online: xx xx xxxx regulation. Genome Res. 18, 1247–1258 (2008).
33. Yu, Y. et al. Dynamic regulation of alternative splicing by silencers that modulate 5′ splice
References site competition. Cell 135, 1224–1236 (2008).
1. Irimia, M. & Roy, S. W. Origin of spliceosomal introns and alternative splicing. Cold Spring 34. Plaschka, C., Lin, P. C., Charenton, C. & Nagai, K. Prespliceosome structure provides
Harb. Perspect. Biol. 6, a016071 (2014). insights into spliceosome assembly and regulation. Nature 559, 419–422 (2018).
2. Plaschka, C., Newman, A. J. & Nagai, K. Structural basis of nuclear pre-mRNA splicing: 35. Jourdain, A. A. et al. Loss of LUC7L2 and U1 snRNP subunits shifts energy metabolism
lessons from yeast. Cold Spring Harb. Perspect. Biol. 11, a032391 (2019). from glycolysis to OXPHOS. Mol. Cell 81, 1905–1919 (2021).
3. Wan, R., Bai, R., Yan, C., Lei, J. & Shi, Y. Structures of the catalytically activated yeast 36. Singh, N. N., Singh, R. N. & Androphy, E. J. Modulating role of RNA structure in alternative
spliceosome reveal the mechanism of branching. Cell 177, 339–351 (2019). splicing of a critical exon in the spinal muscular atrophy genes. Nucleic Acids Res. 35,
4. Kastner, B., Will, C. L., Stark, H. & Lührmann, R. Structural insights into nuclear 371–389 (2007).
pre-mRNA splicing in higher eukaryotes. Cold Spring Harb. Perspect. Biol. 11, a032417 37. Hua, Y. et al. Peripheral SMN restoration is essential for long-term rescue of a severe
(2019). spinal muscular atrophy mouse model. Nature 478, 123–126 (2011).
5. Tholen, J. & Galej, W. P. Structural studies of the spliceosome: bridging the gaps. 38. Jha, N. N., Kim, J. K. & Monani, U. R. Motor neuron biology and disease: a current
Curr. Opin. Struct. Biol. 77, 102461 (2022). perspective on infantile-onset spinal muscular atrophy. Future Neurol. 13, 161–172 (2018).
6. Wahl, M. C., Will, C. L. & Lührmann, R. The spliceosome: design principles of a dynamic 39. Albrechtsen, S. S., Born, A. P. & Boesen, M. S. Nusinersen treatment of spinal muscular
RNP machine. Cell 136, 701–718 (2009). atrophy — a systematic review. Dan. Med. J. 67, 1–12 (2020).
7. Turunen, J. J., Niemelä, E. H., Verma, B. & Frilander, M. J. The significant other: splicing 40. Ratni, H. et al. Discovery of risdiplam, a selective survival of motor neuron-2 (SMN2) gene
by the minor spliceosome. Wiley Interdiscip. Rev. RNA 4, 61–76 (2013). splicing modifier for the treatment of spinal muscular atrophy (SMA). J. Med. Chem. 61,
8. Pan, Q., Shai, O., Lee, L. J., Frey, B. J. & Blencowe, B. J. Deep surveying of alternative 6501–6517 (2018).
splicing complexity in the human transcriptome by high-throughput sequencing. 41. Darras, B. T. et al. Risdiplam-treated infants with type 1 spinal muscular atrophy versus
Nat. Genet. 40, 1413–1415 (2008). historical controls. N. Engl. J. Med. 385, 427–435 (2021).

Nature Reviews Genetics

Review article

42. Campagne, S. et al. Structural basis of a small molecule targeting RNA for a specific 77. Darman, R. B. et al. Cancer-associated SF3B1 hotspot mutations induce cryptic 3′ splice
splicing correction. Nat. Chem. Biol. 15, 1191–1198 (2019). site selection through use of a different branch point. Cell Rep. 13, 1033–1045 (2015).
This study provides the molecular rationale for the specific effects of a small molecule 78. Shiozawa, Y. et al. Aberrant splicing and defective mRNA production induced by somatic
modulator of 5’ splice recognition. spliceosome mutations in myelodysplasia. Nat. Commun. 9, 3649 (2018).
43. Singh, R. N., Seo, J. & Singh, N. N. RNA in spinal muscular atrophy: therapeutic implications 79. Zhang, J. et al. Disease-causing mutations in SF3B1 alter splicing by disrupting interaction
of targeting. Expert. Opin. Ther. Targets 24, 731–743 (2020). with SUGP1. Mol. Cell 76, 82–95 (2019).
44. Alanis, E. F. et al. An exon-specific U1 small nuclear RNA (snRNA) strategy to correct 80. Inoue, D. et al. Spliceosomal disruption of the non-canonical BAF complex in cancer.
splicing defects. Hum. Mol. Genet. 21, 2389–2398 (2012). Nature 574, 432–436 (2019).
45. Rogalska, M. E. et al. Therapeutic activity of modified U1 core spliceosomal particles. This study shows that mutations in SF3B1 converge on repression of the tumour
Nat. Commun. 7, 11168 (2016). suppressor BRD9 by activation of a poison exon.
46. Donegà, S. et al. Rescue of common exon-skipping mutations in cystic fibrosis with 81. Dolatshad, H. et al. Cryptic splicing events in the iron transporter ABCB7 and other key
modified U1 snRNAs. Hum. Mutat. 41, 2143–2154 (2020). target genes in SF3B1-mutant myelodysplastic syndromes. Leukemia 30, 2322–2331 (2016).
47. Donadon, I. et al. Rescue of spinal muscular atrophy mouse models with AAV9-exon- 82. Lee, S. C. W. et al. Synthetic lethal and convergent biological effects of cancer-associated
specific U1 snRNA. Nucleic Acids Res. 47, 7618–7632 (2019). spliceosomal gene mutations. Cancer Cell 34, 225–241 (2018).
48. Balestra, D. et al. Splicing mutations impairing CDKL5 expression and activity can be This study explains the synthetic lethality of splicing mutations in myelodydplastic
efficiently rescued by U1snRNA-based therapy. Int. J. Mol. Sci. 20, 4130 (2019). syndrome as a consequence of aberrant splicing and downregulation of regulators
49. Kaida, D. et al. U1 snRNP protects pre-mRNAs from premature cleavage and of haematopoietic stem cell survival and quiescence.
polyadenylation. Nature 468, 664–668 (2010). 83. Wang, L. et al. Transcriptomic characterization of SF3B1 mutation reveals its pleiotropic
50. Sandberg, R., Neilson, J. R., Sarma, A., Sharp, P. A. & Burge, C. B. Proliferating cells effects in chronic lymphocytic leukemia. Cancer Cell 30, 750–763 (2016).
express mRNAs with shortened 3′ untranslated regions and fewer microRNA target sites. 84. Desterro, J., Bak-Gordon, P. & Carmo-Fonseca, M. Targeting mRNA processing as an
Science 320, 1643–1647 (2008). anticancer strategy. Nat. Rev. Drug. Discov. 19, 112–129 (2020).
51. Oh, J. M. et al. U1 snRNP regulates cancer cell migration and invasion in vitro. Nat. Commun. 85. Corrionero, A., Miñana, B. & Valcárcel, J. Reduced fidelity of branch point recognition
11, 1–8 (2020). and alternative splicing induced by the anti-tumor drug spliceostatin A. Genes. Dev. 25,
52. Sinha, R. et al. Antisense oligonucleotides correct the familial dysautonomia splicing 445–459 (2011).
defect in IKBKAP transgenic mice. Nucleic Acids Res. 46, 4833–4844 (2018). 86. Xargay-Torrent, S. et al. The splicing modulator sudemycin induces a specific antitumor
53. Bruun, G. H. et al. Blocking of an intronic splicing silencer completely rescues IKBKAP response and cooperates with ibrutinib in chronic lymphocytic leukemia. Oncotarget 6,
exon 20 splicing in familial dysautonomia patient cells. Nucleic Acids Res. 46, 7938–7952 22734–22749 (2015).
(2018). 87. Larrayoz, M. et al. The SF3B1 inhibitor spliceostatin A (SSA) elicits apoptosis in chronic
54. Donadon, I. et al. Exon-specific U1 snRNAs improve ELP1 exon 20 definition and rescue lymphocytic leukaemia cells through downregulation of Mcl-1. Leukemia 30, 351–360
ELP1 protein expression in a familial dysautonomia mouse model. Hum. Mol. Genet. 27, (2016).
2466–2476 (2018). 88. Gao, Y. & Koide, K. Chemical perturbation of Mcl-1 pre-mRNA splicing to induce
55. Axelrod, F. B. et al. Kinetin improves IKBKAP mRNA splicing in patients with familial apoptosis in cancer cells. ACS Chem. Biol. 8, 895–900 (2013).
dysautonomia. Pediatr. Res. 70, 480–483 (2011). 89. Vigevani, L., Gohr, A., Webb, T., Irimia, M. & Valcárcel, J. Molecular basis of differential
56. Hims, M. M. et al. Therapeutic potential and mechanism of kinetin as a treatment 3′ splice site sensitivity to anti-tumor drugs targeting U2 snRNP. Nat. Commun. 8, 2100
for the human splicing disease familial dysautonomia. J. Mol. Med. 85, 149–161 (2017).
(2007). 90. Han, T. et al. Anticancer sulfonamides target splicing by inducing RBM39 degradation via
57. Yoshida, M. et al. Rectifier of aberrant mRNA splicing recovers tRNA modification recruitment to DCAF15. Science 356, eaal3755 (2017).
in familial dysautonomia. Proc. Natl Acad. Sci. USA 112, 2764–2769 (2015). This study reports a splicing-based mechanism for the anticancer activity of
58. Ajiro, M. et al. Therapeutic manipulation of IKBKAP mis-splicing with a small molecule sulfonamides and highlights the relevance of controlling RBM39 levels in
to cure familial dysautonomia. Nat. Commun. 12, 1–12 (2021). haematopoietic and lymphoid cancer cell lineages.
59. Zhang, J. et al. Correction of Bcl-x splicing improves responses to imatinib in chronic 91. Wang, E. et al. Targeting an RNA-binding protein network in acute myeloid leukemia.
myeloid leukaemia cells and mouse models. Br. J. Haematol. 189, 1141–1150 (2020). Cancer Cell 35, 369–384 (2019).
60. Osorio, F. G. et al. Hutchinson–Gilford progeria: splicing-directed therapy in a new mouse 92. Jayasinghe, R. G. et al. Systematic analysis of splice-site-creating mutations in cancer.
model of human accelerated aging. Sci. Transl. Med. 3, 1–12 (2011). Cell Rep. 23, 270–281 (2018).
61. Pineda, J. M. B. & Bradley, R. K. Most human introns are recognized via multiple and 93. Kahles, A. et al. Comprehensive analysis of alternative splicing across tumors from 8,705
tissue-specific branchpoints. Genes. Dev. 32, 577–591 (2018). patients. Cancer Cell 34, 211–224 (2018).
62. Yoshida, K. et al. Frequent pathway mutations of splicing machinery in myelodysplasia. This study reveals extensive generation of tumour-specific alternative splicing events
Nature 478, 64–69 (2011). with the potential to generate MHC class I-binding neoantigen peptides.
63. Shirai, C. L. et al. Mutant U2AF1 expression alters hematopoiesis and pre-mRNA splicing 94. Li, L. et al. Targeting the ERG oncogene with splice-switching oligonucleotides as a novel
in vivo. Cancer Cell 27, 631–643 (2015). therapeutic strategy in prostate cancer. Br. J. Cancer 123, 1024–1032 (2020).
64. Park, S. M. et al. U2AF35(S34F) promotes transformation by directing aberrant ATG7 95. Charenton, C., Wilkinson, M. E. & Nagai, K. Mechanism of 5′ splice site transfer for human
pre-mRNA 3′ end formation. Mol. Cell 62, 479–490 (2016). spliceosome activation. Science 364, 362–367 (2019).
65. Biancon, G. et al. Precision analysis of mutant U2AF1 activity reveals deployment of stress 96. Sharma, S., Wongpalee, S. P., Vashisht, A., Wohlschlegel, J. A. & Black, D. L. Stem–loop
granules in myeloid malignancies. Mol. Cell 82, 1107–1122.e7 (2022). 4 of U1 snRNA is essential for splicing and interacts with the U2 snRNP-specific SF3A1
66. Maji, D. et al. Representative cancer-associated U2AF2 mutations alter RNA interactions protein during spliceosome assembly. Genes. Dev. 28, 2518–2531 (2014).
and splicing. J. Biol. Chem. 295, 17148–17157 (2020). 97. Martelly, W., Fellows, B., Senior, K., Marlowe, T. & Sharma, S. Identification of a noncanonical
67. Cretu, C. et al. Molecular architecture of SF3b and structural consequences of its RNA binding domain in the U2 snRNP protein SF3A1. RNA 25, 1509–1521 (2019).
cancer-related mutations. Mol. Cell 64, 307–319 (2016). 98. Martelly, W. et al. Synergistic roles for human U1 snRNA stem–loops in pre-mRNA
68. Cretu, C. et al. Structural basis of splicing modulation by antitumor macrolide splicing. RNA Biol. 18, 2576–2593 (2021).
compounds. Mol. Cell 70, 265–273 (2018). 99. Abovich, N. & Rosbash, M. Cross-intron bridging interactions in the yeast commitment
69. Teng, T. et al. Splicing modulators act at the branch point adenosine binding pocket complex are conserved in mammals. Cell 89, 403–412 (1997).
defined by the PHF5A–SF3b complex. Nat. Commun. 8, 1–16 (2017). 100. Becerra, S., Andrés-León, E., Prieto-Sánchez, S., Hernández-Munain, C. & Suñé, C. Prp40
70. Zhang, Z. et al. Molecular architecture of the human 17S U2 snRNP. Nature 583, 310–313 and early events in splice site definition. Wiley Interdiscip. Rev. RNA 7, 17–32 (2016).
(2020). 101. De Conti, L., Baralle, M. & Buratti, E. Exon and intron definition in pre-mRNA splicing.
71. Seiler, M. et al. H3B-8800, an orally available small-molecule splicing modulator, Wiley Interdiscip. Rev. RNA 4, 49–60 (2013).
induces lethality in spliceosome-mutant cancers. Nat. Med. 24, 497–504 (2018). 102. Schneider, M. et al. Exon definition complexes contain the tri-snRNP and can be directly
This study reports the development of an SF3B1 inhibitor that has greater effects converted into B-like precatalytic splicing complexes. Mol. Cell 38, 223–235 (2010).
in cancer cells that have mutations in SF3B1 components. 103. Robberson, B. L., Cote, G. J. & Berget, S. M. Exon definition may facilitate splice site
72. Mangaonkar, A. A. et al. Prognostic interaction between bone marrow morphology selection in RNAs with multiple exons. Mol. Cell. Biol. 10, 84–94 (1990).
and SF3B1 and ASXL1 mutations in myelodysplastic syndromes with ring sideroblasts. 104. Berget, S. M. Exon recognition in vertebrate splicing. J. Biol. Chem. 270, 2411–2414 (1995).
Blood Cancer J. 8, 1–4 (2018). 105. Gonatopoulos-Pournatzis, T. & Blencowe, B. J. Microexons: at the nexus of nervous
73. Quesada, V. et al. Exome sequencing identifies recurrent mutations of the splicing factor system development, behaviour and autism spectrum disorder. Curr. Opin. Genet. Dev.
SF3B1 gene in chronic lymphocytic leukemia. Nat. Genet. 44, 47–52 (2011). 65, 22–33 (2020).
74. Wang, L. et al. SF3B1 and other novel cancer genes in chronic lymphocytic leukemia. 106. Gonatopoulos-Pournatzis, T. et al. Autism-misregulated eIF4G microexons control
N. Engl. J. Med. 365, 2497–2506 (2011). synaptic translation and higher order cognitive functions. Mol. Cell 77, e16 (2020).
75. Furney, S. J. et al. SF3B1 mutations are associated with alternative splicing in uveal 107. Parras, A. et al. Autism-like phenotype and risk gene mRNA deadenylation by CPEB4
melanoma. Cancer Discov. 3, 1122–1129 (2013). mis-splicing. Nature 560, 441–446 (2018).
76. Rose, A. M. et al. Detection of mutations in SF3B1, EIF1AX and GNAQ in primary orbital Gonatopoulos-Pournatzis et al. and Parras et al. reveal a link between misregulation
melanoma by candidate gene analysis. BMC Cancer 18, 1–9 (2018). of microexons in the cap-binding translation factor eIF4G and autism disorder.

Nature Reviews Genetics

Review article

108. Ule, J. & Blencowe, B. J. Alternative splicing regulatory networks: functions, mechanisms, 141. Jaganathan, K. et al. Predicting splicing from primary sequence with deep learning.
and evolution. Mol. Cell 76, 329–345 (2019). Cell 176, 535–548 (2019).
109. Gonatopoulos-Pournatzis, T. et al. Genome-wide CRISPR–Cas9 interrogation of splicing This study developed a deep neural network to predict splice site utilization, including
networks reveals a mechanism for recognition of autism-misregulated neuronal cryptic splice sites induced by genetic variants associated with autism and intellectual
microexons. Mol. Cell 72, 510–524 (2018). disability.
110. Choudhary, B., Marx, O. & Norris, A. D. Spliceosomal component PRP-40 is a central 142. Bao, S., Moakley, D. F. & Zhang, C. The splicing code goes deep. Cell 176, 414–416 (2019).
regulator of microexon splicing. Cell Rep. 36, 109464 (2021). 143. Cheng, J. et al. MMSplice: modular modeling improves the predictions of genetic variant
111. Salz, H. K. Sex determination in insects: a binary decision based on alternative splicing. effects on splicing. Genome Biol. 20, 1–15 (2019).
Curr. Opin. Genet. Dev. 21, 395–400 (2011). 144. Rhine, C. L. et al. Future directions for high-throughput splicing assays in precision
112. Ule, J. et al. Nova regulates brain-specific splicing to shape the synapse. Nat. Genet. 37, medicine. Hum. Mutat. 40, 1225–1234 (2019).
844–852 (2005). 145. Riepe, T. V., Khan, M., Roosing, S., Cremers, F. P. M. & t Hoen, P. A. C. Benchmarking deep
113. Roy Burman, D., Das, S., Das, C. & Bhattacharya, R. Alternative splicing modulates cancer learning splice prediction tools using functional splice assays. Hum. Mutat. 42, 799–810
aggressiveness: role in EMT/metastasis and chemoresistance. Mol. Biol. Rep. 48, 897–914 (2021).
(2021). 146. Cheung, R. et al. A multiplexed assay for exon recognition reveals that an unappreciated
114. Ke, S. et al. Saturation mutagenesis reveals manifold determinants of exon definition. fraction of rare genetic variants cause large-effect splicing disruptions. Mol. Cell 73,
Genome Res. 28, 11–24 (2018). 183–194 (2019).
115. Fu, X. D. & Ares, M. Context-dependent control of alternative splicing by RNA-binding 147. Lim, K. R. Q., Maruyama, R. & Yokota, T. Eteplirsen in the treatment of Duchenne muscular
proteins. Nat. Rev. Genet. 15, 689–701 (2014). dystrophy. Drug. Des. Devel. Ther. 11, 533–545 (2017).
116. Wu, J. Y. & Maniatis, T. Specific interactions between proteins implicated in splice site 148. Beyer, A. L. & Osheim, Y. N. Splice site selection, rate of splicing, and alternative splicing
selection and regulated alternative splicing. Cell 75, 1061–1070 (1993). on nascent transcripts. Genes. Dev. 2, 754–765 (1988).
117. Jobbins, A. M. et al. The mechanisms of a mammalian splicing enhancer. Nucleic Acids 149. Carrillo Oesterreich, F. et al. Splicing of nascent RNA coincides with intron exit from RNA
Res. 46, 2145–2158 (2018). polymerase II. Cell 165, 372–381 (2016).
118. Clerte, C. & Hall, K. B. Characterization of multimeric complexes formed by the human 150. Neugebauer, K. M. Nascent RNA and the coordination of splicing with transcription.
PTB1 protein on RNA. RNA 12, 457–475 (2006). Cold Spring Harb. Perspect. Biol. 11, a032227 (2019).
119. Braun, J. E., Friedman, L. J., Gelles, J. & Moore, M. J. Synergistic assembly of human 151. Custódio, N. & Carmo-Fonseca, M. Co-transcriptional splicing and the CTD code.
pre-spliceosomes across introns and exons. eLife 7, 1–18 (2018). Crit. Rev. Biochem. Mol. Biol. 51, 395–411 (2016).
120. Sharma, S., Maris, C., Allain, F. H. T. & Black, D. L. U1 snRNA directly interacts with 152. Moon, K. H., Zhao, X. & Yu, Y. T. Pre-mRNA splicing in the nuclei of Xenopus oocytes.
polypyrimidine tract-binding protein during splicing repression. Mol. Cell 41, 579–588 Methods Mol. Biol. 322, 149–163 (2006).
(2011). 153. Herzel, L., Ottoz, D. S. M., Alpert, T. & Neugebauer, K. M. Splicing and transcription touch
121. Gueroussov, S. et al. Regulatory expansion in mammals of multivalent hnRNP assemblies base: co-transcriptional spliceosome assembly and function. Nat. Rev. Mol. Cell Biol. 18,
that globally control alternative splicing. Cell 170, 324–339 (2017). 637–650 (2017).
122. Ying, Y. et al. Splicing activation by Rbfox requires self-aggregation through its tyrosine- 154. Anvar, S. Y. et al. Full-length mRNA sequencing uncovers a widespread coupling
rich domain. Cell 170, 312–323 (2017). between transcription initiation and mRNA processing. Genome Biol. 19, 1–18 (2018).
Gueroussov et al. and Ying et al. illustrate the importance of intrinsically disordered 155. Giono, L. E. & Kornblihtt, A. R. Linking transcription, RNA polymerase II elongation and
domains mediating higher order multivalent assemblies for the function of various alternative splicing. Biochem. J. 477, 3091–3104 (2020).
families of splicing regulatory factors. 156. Bentley, D. L. Coupling mRNA processing with transcription in time and space. Nat. Rev.
123. Witten, J. T. & Ule, J. Understanding splicing regulation through RNA splicing maps. Genet. 15, 163–175 (2014).
Trends Genet. 27, 89–97 (2011). 157. Dujardin, G. et al. How slow RNA polymerase II elongation favors alternative exon
124. Yee, B. A., Pratt, G. A., Graveley, B. R., van Nostrand, E. L. & Yeo, G. W. RBP-Maps enables skipping. Mol. Cell 54, 683–690 (2014).
robust generation of splicing regulatory maps. RNA 25, 193–204 (2019). 158. Godoy Herz, M. A. et al. Light regulates plant alternative splicing through the control
125. Mikl, M., Hamburg, A., Pilpel, Y. & Segal, E. Dissecting splicing decisions and cell-to-cell of transcriptional elongation. Mol. Cell 73, 1066–1074 (2019).
variability with designed sequence libraries. Nat. Commun. 10, 1–14 (2019). 159. Maslon, M. M. et al. A slow transcription rate causes embryonic lethality and perturbs
126. Julien, P., Miñana, B., Baeza-Centurion, P., Valcárcel, J. & Lehner, B. The complete local kinetic coupling of neuronal genes. EMBO J. 38, 1–18 (2019).
genotype–phenotype landscape for the alternative splicing of a human exon. Nat. Commun. 160. Chathoth, K. T., Barrass, J. D., Webb, S. & Beggs, J. D. A splicing-dependent transcriptional
7, 1–8 (2016). checkpoint associated with prespliceosome formation. Mol. Cell 53, 779–790 (2014).
127. Baeza-Centurion, P., Miñana, B., Schmiedel, J. M., Valcárcel, J. & Lehner, B. Combinatorial 161. Fiszbein, A., Krick, K. S., Begg, B. E. & Burge, C. B. Exon-mediated activation of
genetics reveals a scaling law for the effects of mutations on splicing. Cell 176, 549–563 transcription starts. Cell 179, 1551–1565 (2019).
(2019). This study provides evidence for the general mechanisms by which alternative
Deep mutagenesis of an alternatively spliced exon reveals that the effect of a mutation splicing influences alternative transcription initiation.
is maximal at intermediate initial levels of exon inclusion. 162. Reimer, K. A., Mimoso, C. A., Adelman, K. & Neugebauer, K. M. Co-transcriptional splicing
128. Baeza-Centurion, P., Miñana, B., Valcárcel, J. & Lehner, B. Mutations primarily alter the regulates 3′ end cleavage during mammalian erythropoiesis. Mol. Cell 81, 998–1012 (2021).
inclusion of alternatively spliced exons. eLife 9, 1–74 (2020). 163. Bedi, K. et al. Cotranscriptional splicing efficiencies differ within genes and between cell
129. Souček, P. et al. High-throughput analysis revealed mutations’ diverging effects on SMN1 types. RNA 27, 829–840 (2021).
exon 7 splicing. RNA Biol. 16, 1364–1376 (2019). 164. Drexler, H. L., Choquet, K. & Churchman, L. S. Splicing kinetics and coordination
130. Braun, S. et al. Decoding a cancer-relevant splicing decision in the RON proto-oncogene revealed by direct nascent RNA sequencing through nanopores. Mol. Cell 77, 985–998
using high-throughput mutagenesis. Nat. Commun. 9, 3315 (2018). (2020).
131. Soemedi, R. et al. Pathogenic variants that alter protein code often disrupt splicing. 165. Sousa-Luís, R. et al. POINT technology illuminates the processing of polymerase-associated
Nat. Genet. 49, 848–855 (2017). intact nascent transcripts. Mol. Cell 81, 1935–1950 (2021).
132. Rong, S. et al. Mutational bias and the protein code shape the evolution of splicing 166. Tilgner, H. et al. Microfluidic isoform sequencing shows widespread splicing coordination
enhancers. Nat. Commun. 11, 1–10 (2020). in the human transcriptome. Genome Res. 28, 231–242 (2018).
133. Mueller, W. F., Larsen, L. S. Z., Garibaldi, A., Hatfield, G. W. & Hertel, K. J. The silent 167. Gupta, I. et al. Single-cell isoform RNA sequencing characterizes isoforms in thousands
sway of splicing by synonymous substitutions. J. Biol. Chem. 290, 27700–27711 of cerebellar cells. Nat. Biotechnol. 36, 1197–1202 (2018).
(2015). This technique allows for high-resolution discovery and analysis of alternative
134. Dufner-Almeida, L. G., do Carmo, R. T., Masotti, C. & Haddad, L. A. Understanding splicing in individual cells from complex tissues.
human DNA variants affecting pre-mRNA splicing in the NGS era. Adv. Genet. 103, 39–90 168. Gordon, J. M., Phizicky, D. V. & Neugebauer, K. M. Nuclear mechanisms of gene expression
(2019). control: pre-mRNA splicing as a life or death decision. Curr. Opin. Genet. Dev. 67, 67–76
135. Cummings, B. B. et al. Transcript expression-aware annotation improves rare variant (2021).
interpretation. Nature 581, 452–458 (2020). 169. Boutz, P. L., Bhutkar, A. & Sharp, P. A. Detained introns are a novel, widespread class
136. Supek, F., Miñana, B., Valcárcel, J., Gabaldón, T. & Lehner, B. Synonymous mutations of post-transcriptionally spliced introns. Genes. Dev. 29, 63–80 (2015).
frequently act as driver mutations in human cancers. Cell 156, 1324–1335 (2014). 170. Naro, C. et al. An orchestrated intron retention program in meiosis controls timely usage
137. Sharma, Y. et al. A pan-cancer analysis of synonymous mutations. Nat. Commun. 10, of transcripts during germ cell differentiation. Dev. Cell 41, 82–93 (2017).
2569 (2019). 171. Mauger, O., Lemoine, F. & Scheiffele, P. Targeted intron retention and excision for rapid
138. Rosenberg, A. B., Patwardhan, R. P., Shendure, J. & Seelig, G. Learning the sequence gene regulation in response to neuronal activity. Neuron 92, 1266–1278 (2016).
determinants of alternative splicing from millions of random sequences. Cell 163, 172. Braun, C. J. et al. Coordinated splicing of regulatory detained introns within oncogenic
698–711 (2015). transcripts creates an exploitable vulnerability in malignant glioma. Cancer Cell 32,
139. Ke, S. et al. Quantitative evaluation of all hexamers as exonic splicing elements. 411–426 (2017).
Genome Res. 21, 1360–1374 (2011). This study demonstrates the relevance of a splicing regulatory programme affecting
140. Arias, M. A., Lubkin, A. & Chasin, L. A. Splicing of designer exons informs a biophysical detained introns in cancer cells and its potential targeting using inihibitors of
model for exon definition. RNA 21, 213–229 (2015). argininine methylation enzymes.

Nature Reviews Genetics

Review article

173. Galganski, L., Urbanek, M. O. & Krzyzosiak, W. J. Nuclear speckles: molecular organization, 205. Xu, B., Meng, Y. & Jin, Y. RNA structures in alternative splicing and back-splicing.
biological function and role in disease. Nucleic Acids Res. 45, 10350–10368 (2017). Wiley Interdiscip. Rev. RNA 12, 1–39 (2021).
174. Ilık, İ. A. et al. SON and SRRM2 are essential for nuclear speckle formation. eLife 9, 1–48 206. Meyer, M., Plass, M., Pérez-Valle, J., Eyras, E. & Vilardell, J. Deciphering 3′ss selection
(2020). in the yeast genome reveals an RNA thermosensor that mediates alternative splicing.
175. Ilık, İ. A. & Aktaş, T. Nuclear speckles: dynamic hubs of gene expression regulation. Mol. Cell 43, 1033–1039 (2011).
FEBS J. https://doi.org/10.1111/febs.16117 (2021). 207. Kalmykova, S. et al. Conserved long-range base pairings are associated with pre-mRNA
176. Girard, C. et al. Post-transcriptional spliceosomes are retained in nuclear speckles until processing of human genes. Nat. Commun. 12, 1–17 (2021).
splicing completion. Nat. Commun. 3, 994 (2012). 208. Warf, M. B., Diegel, J. V., Von Hippel, P. H. & Berglund, J. A. The protein factors MBNL1 and
177. Dias, A. P., Dufu, K., Lei, H. & Reed, R. A role for TREX components in the release of spliced U2AF65 bind alternative RNA structures to regulate splicing. Proc. Natl Acad. Sci. USA
mRNA from nuclear speckle domains. Nat. Commun. 1, 97 (2010). 106, 9203–9208 (2009).
178. Guo, Y. E. et al. Pol II phosphorylation regulates a switch between transcriptional and 209. Varani, L. et al. Structure of tau exon 10 splicing regulatory element RNA and
splicing condensates. Nature 572, 543–548 (2019). destabilization by mutations of frontotemporal dementia and parkinsonism linked
This study reveals the existence of functionally distinct nuclear condensates involved to chromosome 17. Proc. Natl Acad. Sci. USA 96, 8229–8234 (1999).
in transcription initiation and RNA processing and the role of protein phosphorylation 210. Kar, A. et al. RNA helicase p68 (DDX5) regulates tau exon 10 splicing by modulating
in the switch between them. a stem–loop structure at the 5′ splice site. Mol. Cell. Biol. 31, 1812–1821 (2011).
179. Barutcu, A. R. et al. Systematic mapping of nuclear domain-associated transcripts 211. Singh, N. N. & Singh, R. N. How RNA structure dictates the usage of a critical exon of spinal
reveals speckles and lamina as hubs of functionally distinct retained introns. Mol. Cell muscular atrophy gene. Biochim. Biophys. Acta—Gene Regul. Mech. 1862, 194403 (2019).
82, 1035–1052 (2022). 212. Georgakopoulos-Soares, I. et al. Alternative splicing modulation by G-quadruplexes.
180. Tammer, L. et al. Gene architecture directs splicing outcome in separate nuclear spatial Nat. Commun. 13, 2404 (2022).
regions. Mol. Cell 82, 1021–1034 (2022). 213. Huang, H., Zhang, J., Harvey, S. E., Hu, X. & Cheng, C. RNA G-quadruplex secondary
Barutcu et al. and Tammer et al. reveal that the nuclear sub-localization of transcription structure promotes alternative splicing via the RNA-binding protein hnRNPF. Genes. Dev.
influences alternative splicing decisions. 31, 2296–2309 (2017).
181. Amit, M. et al. Differential GC content between exons and introns establishes distinct 214. Fish, L. et al. A prometastatic splicing program regulated by SNRPA1 interactions
strategies of splice-site recognition. Cell Rep. 1, 543–556 (2012). with structured RNA elements. Science 372, eabc7531 (2021).
182. Iannone, C. & Valcárcel, J. Chromatin’s thread to alternative splicing regulation. Early evidence that a structural splicing enhancer coordinates an exon inclusion
Chromosoma 122, 465–474 (2013). programme relevant for highly metastatic cancer cells. This regulatory element
183. de Almeida, S. F. & Carmo-Fonseca, M. Reciprocal regulatory links between is activated by a non-canonical function of the snRNP protein SNRPA1.
cotranscriptional splicing and chromatin. Semin. Cell Dev. Biol. 32, 2–10 (2014). 215. Dubin, R. A., Kazmi, M. A. & Ostrer, H. Inverted repeats are necessary for circularization
184. Shukla, S. et al. CTCF-promoted RNA polymerase II pausing links DNA methylation of the mouse testis Sry transcript. Gene 167, 245–248 (1995).
to splicing. Nature 479, 74–79 (2011). 216. Jeck, W. R. et al. Circular RNAs are abundant, conserved, and associated with ALU
185. Luco, R. F. et al. Regulation of alternative splicing by histone modifications. Science 327, repeats. RNA 19, 426 (2013).
996–1000 (2010). 217. Ashwal-Fluss, R. et al. CircRNA biogenesis competes with pre-mRNA splicing. Mol. Cell
186. Segelle, A. et al. Histone marks regulate the epithelial-to-mesenchymal transition via 56, 55–66 (2014).
alternative splicing. Cell Rep. 38, 110357 (2022). 218. Zhang, X. O. et al. Complementary sequence-mediated exon circularization. Cell 159,
An early demonstration that modification of chromatin epigenetic marks can directly 134–147 (2014).
affect splice site choice. 219. Liang, D. & Wilusz, J. E. Short intronic repeat sequences facilitate circular RNA production.
187. Marasco, L. E. et al. Counteracting chromatin effects of a splicing-correcting antisense Genes. Dev. 28, 2233–2247 (2014).
oligonucleotide improves its therapeutic efficacy in spinal muscular atrophy. Cell 185, 220. Ivanov, A. et al. Analysis of intron sequences reveals hallmarks of circular RNA biogenesis
2057–2070 (2022). in animals. Cell Rep. 10, 170–177 (2015).
188. Helm, M. & Motorin, Y. Detecting RNA modifications in the epitranscriptome: predict and 221. Gruhl, F., Janich, P., Kaessmann, H. & Gatfield, D. Circular RNA repertoires are associated
validate. Nat. Rev. Genet. 18, 275–291 (2017). with evolutionarily young transposable elements. eLife 10, 1–33 (2021).
189. Zhao, B. S., Roundtree, I. A. & He, C. Post-transcriptional gene regulation by mRNA 222. Santos-Rodriguez, G., Voineagu, I. & Weatheritt, R. J. Evolutionary dynamics of circular
modifications. Nat. Rev. Mol. Cell Biol. 18, 31–42 (2016). rnas in primates. eLife 10, 1–22 (2021).
190. Dawson, M. A. & Kouzarides, T. Cancer epigenetics: from mechanism to therapy. 223. Xu, C. & Zhang, J. Mammalian circular RNAs result largely from splicing errors. Cell Rep.
Cell 150, 12–27 (2012). 36, 109439 (2021).
191. Mendel, M. et al. Splice site m6A methylation prevents binding of U2AF35 to inhibit RNA 224. Du, W. W. et al. Foxo3 circular RNA retards cell cycle progression via forming ternary
splicing. Cell 184, 1–18 (2021). complexes with p21 and CDK2. Nucleic Acids Res. 44, 2846–2858 (2016).
192. Pendleton, K. E. et al. The U6 snRNA m6A methyltransferase METTL16 regulates SAM 225. Hansen, T. B. et al. Natural RNA circles function as efficient microRNA sponges. Nature
synthetase intron retention. Cell 169, 824–835 (2017). 495, 384–388 (2013).
Mendel et al. and Pendleton et al. link m6A modification of pre-mRNAs with splicing 226. Piwecka, M. et al. Loss of a mammalian circular RNA locus causes miRNA deregulation
regulation events important for S-adenosylmethionine homeostasis. and affects brain function. Science 357, eaam8526 (2017).
193. Zhou, K. I. et al. Regulation of co-transcriptional pre-mRNA splicing by m6A through the 227. Li, S. et al. Screening for functional circular RNAs using the CRISPR–Cas13 system.
low-complexity protein hnRNPG. Mol. Cell 76, 70–81 (2019). Nat. Methods 18, 51–59 (2021).
194. Xiao, W. et al. Nuclear m6A reader YTHDC1 regulates mRNA splicing. Mol. Cell 61, 228. Pamudurti, N. R. et al. Translation of CircRNAs. Mol. Cell 66, 9–21 (2017).
507–519 (2016). 229. Legnini, I. et al. Circ-ZNF609 is a circular RNA that can be translated and functions
195. Kasowitz, S. D. et al. Nuclear m6A reader YTHDC1 regulates alternative polyadenylation in myogenesis. Mol. Cell 66, 22–37 (2017).
and splicing during mouse oocyte development. PLoS Genet. 14, 1–28 (2018). 230. Weigelt, C. M. et al. An insulin-sensitive circular RNA that regulates lifespan in Drosophila.
196. Louloupi, A., Ntini, E., Conrad, T. & Ørom, U. A. V. Transient N-6-methyladenosine Mol. Cell 79, 268–279 (2020).
transcriptome sequencing reveals a regulatory role of m6A in splicing efficiency. 231. Hansen, T. B. Signal and noise in circRNA translation. Methods 196, 68–73 (2021).
Cell Rep. 23, 3429–3437 (2018). 232. Yang, Y. et al. Extensive translation of circular RNAs driven by N6-methyladenosine.
197. Ke, S. et al. m6A mRNA modifications are deposited in nascent pre-mRNA and are not Cell Res. 27, 626–641 (2017).
required for splicing but do specify cytoplasmic turnover. Genes. Dev. 31, 990–1006 (2017). 233. Joglekar, A. et al. A spatially resolved brain region- and cell type-specific isoform atlas
198. Martinez, N. M. et al. Pseudouridine synthases modify human pre-mRNA co-transcriptionally of the postnatal mouse brain. Nat. Commun. 12, 1–16 (2021).
and affect pre-mRNA processing. Mol. Cell 82, 645–659 (2022). 234. Dominguez, D. et al. Sequence, structure, and context preferences of human RNA
199. Morais, P., Adachi, H. & Yu, Y.-T. Spliceosomal snRNA epitranscriptomics. Front. Genet. 12, binding proteins. Mol. Cell 70, 854–867 (2018).
652129 (2021). 235. Van Nostrand, E. L. et al. A large-scale binding and functional map of human RNA-binding
200. Goh, Y. T., Koh, C. W. Q., Sim, D. Y., Roca, X. & Goh, W. S. S. METTL4 catalyzes m6Am proteins. Nature 583, 711–719 (2020).
methylation in U2 snRNA to regulate pre-mRNA splicing. Nucleic Acids Res. 48, A massive effort to identify binding sites on RNA and chromatin for 356 RNA binding
9250–9261 (2020). proteins and to determine the functional implications for RNA stability, splicing
201. Lu, Z. & Chang, H. Y. Decoding the RNA structurome. Curr. Opin. Struct. Biol. 36, 142–148 regulation and RNA localization.
(2016). 236. Van Nostrand, E. L. et al. Principles of RNA processing from analysis of enhanced CLIP
202. Sun, L. et al. RNA structure maps across mammalian cellular compartments. Nat. Struct. maps for 150 RNA binding proteins. Genome Biol. 21, 1–26 (2020).
Mol. Biol. 26, 322–330 (2019). 237. Feng, H. et al. Complexity and graded regulation of neuronal cell-type-specific
203. Saldi, T., Riemondy, K., Erickson, B. & Bentley, D. L. Alternative RNA structures formed alternative splicing revealed by single-cell RNA sequencing. Proc. Natl Acad. Sci. USA
during transcription depend on elongation rate and modify RNA processing. Mol. Cell 81, 118, 1–12 (2021).
1789–1801 (2021). 238. Shalek, A. K. et al. Single-cell transcriptomics reveals bimodality in expression and
204. Bartys, N., Kierzek, R. & Lisowiec-Wachnicka, J. The regulation properties of RNA splicing in immune cells. Nature 498, 236–240 (2013).
secondary structure in alternative splicing. Biochim. Biophys. Acta — Gene Regul. Mech. 239. Song, Y. et al. Single-cell alternative splicing analysis with expedition reveals splicing
1862, 194401 (2019). dynamics during neuron differentiation. Mol. Cell 67, 148–161 (2017).

Nature Reviews Genetics

Review article

240. Liu, W. & Zhang, X. Single-cell alternative splicing analysis reveals dominance of single 273. Castelo, R. & Guigó, R. Splice site identification by idlBNs. Bioinformatics 20, 69–76
transcript variant. Genomics 112, 2418–2425 (2020). (2004).
241. Huang, Y. & Sanguinetti, G. BRIE2: computational identification of splicing phenotypes 274. Barash, Y. et al. Deciphering the splicing code. Nature 465, 53–59 (2010).
from single-cell transcriptomic experiments. Genome Biol. 22, 1–15 (2021). 275. Xiong, H. Y. et al. The human splicing code reveals new insights into the genetic
242. Linker, S. M. et al. Combined single-cell profiling of expression and DNA methylation determinants of disease. Science 347, 1254806 (2015).
reveals splicing regulation and heterogeneity. Genome Biol. 20, 1–14 (2019). 276. Jha, A., Gazzara, M. R. & Barash, Y. Integrative deep models for alternative splicing.
243. Kim, H. S., Grimes, S. M., Hooker, A. C., Lau, B. T. & Ji, H. P. Single-cell characterization Bioinformatics 33, i274–i282 (2017).
of CRISPR-modified transcript isoforms with nanopore sequencing. Genome Biol. 22, 277. Jha, A. et al. Enhanced integrated gradients: improving interpretability of deep learning
1–16 (2021). models using splicing codes as a case study. Genome Biol. 21, 1–22 (2020).
244. Thompson, M. et al. Splicing in a single neuron is coordinately controlled by RNA binding 278. Zuallaert, J. et al. Splicerover: interpretable convolutional neural networks for improved
proteins and transcription factors. eLife 8, 1–19 (2019). splice site prediction. Bioinformatics 34, 4180–4188 (2018).
245. Lukačišin, M., Espinosa-Cantú, A. & Bollenbach, T. Intron-mediated induction of phenotypic 279. Bretschneider, H., Gandhi, S., Deshwar, A. G., Zuberi, K. & Frey, B. J. COSSMO: predicting
heterogeneity. Nature 605, 113–118 (2022). competitive alternative splice site selection using deep learning. Bioinformatics 34,
246. Wan, Y. et al. Dynamic imaging of nascent RNA reveals general principles of transcription i429–i437 (2018).
dynamics and stochastic splice site selection. Cell 184, 2878–2895 (2021). 280. Cheng, J., Çelik, M. H., Kundaje, A. & Gagneur, J. MTSplice predicts effects of genetic
This study reports a quasi-genome-scale platform for observing the synthesis and variants on tissue-specific splicing. Genome Biol. 22, 1–19 (2021).
procesing kinetics of single nascent RNA molecules in real time; it reveals large 281. Rentzsch, P., Witten, D., Cooper, G. M., Shendure, J. & Kircher, M. CADD: predicting
kinetic variation of single intron removal in single cells and widespread stochastic the deleteriousness of variants throughout the human genome. Nucleic Acids Res. 47,
recursive splicing within introns. D886–D894 (2019).
247. Martin, R. M., Rino, J., Carvalho, C., Kirchhausen, T. & Carmo-Fonseca, M. Live-cell 282. Rentzsch, P., Schubach, M., Shendure, J. & Kircher, M. CADD-Splice — improving
visualization of pre-mRNA splicing with single-molecule sensitivity. Cell Rep. 4, 1144–1155 genome-wide variant effect prediction using deep learning-derived splice scores.
(2013). Genome Med. 13, 1–12 (2021).
248. Burnette, J. M., Miyamoto-Sato, E., Schaub, M. A., Conklin, J. & Lopez, A. J. Subdivision 283. Zeng, T. & Li, Y. I. Predicting RNA splicing from DNA sequence using Pangolin. Genome
of large introns in Drosophila by recursive splicing at nonexonic elements. Genetics 170, Biol. 23, 103 (2022).
661–674 (2005). 284. Zhang, X. H. F., Arias, M. A., Shengdong, K. E. & Chasin, L. A. Splicing of designer exons
249. Sibley, C. R. et al. Recursive splicing in long vertebrate genes. Nature 521, 371–375 (2015). reveals unexpected complexity in pre-mRNA splicing. RNA 15, 367–376 (2009).
250. Kim, S. W. et al. Widespread intra-dependencies in the removal of introns from human 285. Schirman, D., Yakhini, Z., Pilpel, Y. & Dahan, O. A broad analysis of splicing regulation
transcripts. Nucleic Acids Res. 45, 9503–9513 (2017). in yeast using a large library of synthetic introns. PLOS Genet. 17, e1009805 (2021).
251. Jangi, M. & Sharp, P. A. Building robust transcriptomes with master splicing factors. Cell 286. Adamson, S. I., Zhan, L. & Graveley, B. R. Vex-seq: high-throughput identification of the
159, 487–498 (2014). impact of genetic variation on pre-mRNA splicing efficiency. Genome Biol. 19, 1–12 (2018).
252. Kokic, G., Wagner, F. R., Chernev, A., Urlaub, H. & Cramer, P. Structural basis of human 287. Mount, S. M. et al. Assessing predictions of the impact of variants on splicing in CAGI5.
transcription–DNA repair coupling. Nature 598, 368–372 (2021). Hum. Mutat. 40, 1215–1224 (2019).
253. Zhang, S. et al. Structure of a transcribing RNA polymerase II–U1 snRNP complex. 288. Tubeuf, H. et al. Large-scale comparative evaluation of user-friendly tools for predicting
Science 371, 305–309 (2021). variant-induced alterations of splicing regulatory elements. Hum. Mutat. 41, 1811–1829
254. Haselbach, D. et al. Structure and conformational dynamics of the human spliceosomal (2020).
bact complex. Cell 172, 454–464 (2018). 289. Cortés-López, M. et al. High-throughput mutagenesis identifies mutations and
255. Sebbag-Sznajder, N. et al. Dynamic supraspliceosomes are assembled on different RNA-binding proteins controlling CD19 splicing and CART-19 therapy resistance.
transcripts regardless of their intron number and splicing state. Front. Genet. 11, 1–14 Nat. Commun. 13, 5570 (2022).
(2020). 290. Luco, R. F., Allo, M., Schor, I. E., Kornblihtt, A. R. & Misteli, T. Epigenetics in alternative
256. Papasaikas, P., Tejedor, J. R., Vigevani, L. & Valcárcel, J. Functional splicing network pre-mRNA splicing. Cell 144, 16–26 (2011).
reveals extensive regulatory potential of the core spliceosomal machinery. Mol. Cell 57, 291. Schwartz, S., Meshorer, E. & Ast, G. Chromatin organization marks exon–intron structure.
7–22 (2015). Nat. Struct. Mol. Biol. 16, 990–995 (2009).
257. Statello, L., Guo, C. J., Chen, L. L. & Huarte, M. Gene regulation by long non-coding RNAs 292. Tilgner, H. et al. Nucleosome positioning as a determinant of exon recognition.
and its biological functions. Nat. Rev. Mol. Cell Biol. 22, 96–118 (2021). Nat. Struct. Mol. Biol. 16, 996–1001 (2009).
258. Deveson, I. W. et al. Universal alternative splicing of noncoding exons. Cell Syst. 6, 293. Andersson, R., Enroth, S., Rada-Iglesias, A., Wadelius, C. & Komorowski, J. Nucleosomes
245–255 (2018). are well positioned in exons and carry characteristic histone modifications. Genome Res.
259. Bergen, V., Lange, M., Peidli, S., Wolf, F. A. & Theis, F. J. Generalizing RNA velocity to 19, 1732–1741 (2009).
transient cell states through dynamical modeling. Nat. Biotechnol. 38, 1408–1414 (2020). 294. David, C. J., Boyne, A. R., Millhouse, S. R. & Manley, J. L. The RNA polymerase II C-terminal
260. North, K. et al. Synthetic introns enable splicing factor mutation-dependent targeting domain promotes splicing activation through recruitment of a U2AF65–Prp19 complex.
of cancer cells. Nat. Biotechnol. 40, 1103–1113 (2022). Genes. Dev. 25, 972–982 (2011).
In this study, splicing alterations characteristic of cancer cells are leveraged 295. Morgan, J. T., Fink, G. R. & Bartel, D. P. Excised linear introns regulate growth in yeast.
to engineer synthetic introns that are specifically spliced in and induce the death Nature 565, 606–611 (2019).
of cancer cells. 296. Parenteau, J. et al. Introns are mediators of cell response to starvation. Nature 565,
261. Zhang, X. et al. Structures of the human spliceosomes before and after release of the 612–617 (2019).
ligated exon. Cell Res. 29, 274–285 (2019). 297. Mattick, J. S. & Gagen, M. J. The evolution of controlled multitasked gene networks:
262. Tholen, J., Razew, M., Weis, F. & Galej, W. P. Structural basis of branch site recognition the role of introns and other noncoding RNAs in the development of complex organisms.
by the human spliceosome. Science 375, 50–57 (2022). Mol. Biol. Evol. 18, 1611–1630 (2001).
263. Zhan, X., Yan, C., Zhang, X., Lei, J. & Shi, Y. Structures of the human pre-catalytic 298. Dieci, G., Preti, M. & Montanini, B. Eukaryotic snoRNAs: a paradigm for gene expression
spliceosome and its precursor spliceosome. Cell Res. 28, 1129–1140 (2018). flexibility. Genomics 94, 83–88 (2009).
264. Li, X. et al. A unified mechanism for intron and exon definition and back-splicing. Nature 299. Kelemen, O. et al. Function of alternative splicing. Gene 514, 1–30 (2013).
573, 375–380 (2019). 300. Vuong, C. K., Black, D. L. & Zheng, S. The neurogenetics of alternative splicing.
265. Zhan, X., Yan, C., Zhang, X., Lei, J. & Shi, Y. Structure of a human catalytic step I Nat. Rev. Neurosci. 17, 265–281 (2016).
spliceosome. Science 359, 537–545 (2018). 301. Tress, M. L., Abascal, F. & Valencia, A. Alternative splicing may not be the key to
266. Bertram, K. et al. Cryo-EM structure of a human spliceosome activated for step 2 proteome complexity. Trends Biochem. Sci. 42, 98–110 (2017).
of splicing. Nature 542, 318–323 (2017). 302. Rodriguez, J. M., Pozo, F., Di Domenico, T., Vazquez, J. & Tress, M. L. An analysis of
267. Fica, S. M., Oubridge, C., Wilkinson, M. E., Newman, A. J. & Nagai, K. A human postcatalytic tissue-specific alternative splicing at the protein level. PLoS Comput. Biol. 16, e1008287
spliceosome structure reveals essential roles of metazoan factors for exon ligation. (2020).
Science 363, 710–714 (2019). 303. Weatheritt, R. J., Sterne-Weiler, T. & Blencowe, B. J. The ribosome-engaged landscape
268. Townsend, C. et al. Mechanism of protein-guided folding of the active site U2/U6 RNA of alternative splicing. Nat. Struct. Mol. Biol. 23, 1117–1123 (2016).
during spliceosome activation. Science 370, eabc3753 (2020). 304. Ellis, J. D. et al. Tissue-specific alternative splicing remodels protein-protein interaction
269. Shapiro, M. B. & Senapathy, P. RNA splice junctions of different classes of eukaryotes: networks. Mol. Cell 46, 884–892 (2012).
sequence statistics and functional implications in gene expression. Nucleic Acids Res. High-throughput analyses of the impact of alternative splicing reveal widespread
15, 7155–7174 (1987). effects on the modulation of protein–protein interactions.
270. Reese, M. G., Eeckman, F. H., Kulp, D. & Haussler, D. Improved splice site detection 305. Sinha, A. & Nagarajaram, H. A. Effect of alternative splicing on the degree centrality
in Genie. J. Comput. Biol. 4, 311–323 (1997). of nodes in protein–protein interaction networks of Homo sapiens. J. Proteome Res. 12,
271. Pertea, M., Lin, X. & Salzberg, S. L. GeneSplicer: a new computational method for splice 1980–1988 (2013).
site prediction. Nucleic Acids Res. 29, 1185–1190 (2001). 306. Yang, Y. et al. Determination of a comprehensive alternative splicing regulatory network
272. Yeo, G. & Burge, C. B. Maximum entropy modeling of short sequence motifs with and combinatorial regulation by key factors during the epithelial-to-mesenchymal
applications to RNA splicing signals. J. Comput. Biol. 11, 377–394 (2004). transition. Mol. Cell. Biol. 36, 1704–1719 (2016).

Nature Reviews Genetics

Review article

307. Climente-González, H., Porta-Pardo, E., Godzik, A. & Eyras, E. The functional impact Research Council, European Innovation Council, LaCaixa Health, Worldwide Cancer Research,
of alternative splicing in cancer. Cell Rep. 20, 2215–2226 (2017). AGAUR, Spanish Ministry of Economy and Competitiveness and the Centre of Excellence
308. Louadi, Z. et al. Functional enrichment of alternative splicing events with NEASE reveals Severo Ochoa. The authors acknowledge support of the Spanish Ministry of Science and
insights into tissue identity and diseases. Genome Biol. 22, 1–22 (2021). Innovation to the EMBL partnership and the CERCA Programme/Generalitat de Catalunya.
309. Ezkurdia, I. et al. Comparative proteomics reveals a significant bias toward alternative
protein isoforms with conserved structure and function. Mol. Biol. Evol. 29, 2265–2283 Author contributions
(2012). All authors contributed to all aspects of the article.
310. Buljan, M. et al. Alternative splicing of intrinsically disordered regions and rewiring
of protein interactions. Curr. Opin. Struct. Biol. 23, 443–450 (2013).
Competing interests
311. Buljan, M. et al. Tissue-specific splicing of disordered segments that embed binding
The authors declare no competing interests.
motifs rewires protein interaction networks. Mol. Cell 46, 871–883 (2012).
312. Lewis, B. P., Green, R. E. & Brenner, S. E. Evidence for the widespread coupling of
alternative splicing and nonsense-mediated mRNA decay in humans. Proc. Natl Acad. Additional information
Sci. USA 100, 189–192 (2003). Supplementary information The online version contains supplementary material available at
313. Lindeboom, R. G. H., Supek, F. & Lehner, B. The rules and impact of nonsense-mediated https://doi.org/10.1038/s41576-022-00556-8.
mRNA decay in human cancers. Nat. Genet. 48, 1112–1118 (2016).
314. Supek, F., Lehner, B. & Lindeboom, R. G. H. To NMD or not to NMD: nonsense-mediated Correspondence should be addressed to Juan Valcárcel.
mRNA decay in cancer and other genetic diseases. Trends Genet. 37, 657–668 (2021).
315. Tapial, J. et al. An atlas of alternative splicing profiles and functional associations Peer review information Nature Reviews Genetics thanks R. Luco, and the other, anonymous,
reveals new regulatory programs and genes that simultaneously express multiple major reviewer(s) for their contribution to the peer review of this work.
isoforms. Genome Res. 27, 1759–1768 (2017).
This study provides a comprehensive database of alternative splicing in tissues and Reprints and permissions information is available at www.nature.com/reprints.
organisms and facilitates the identification of tissue-specific regulatory programmes.
316. Schlautmann, L. P. & Gehring, N. H. A day in the life of the exon junction complex. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
Biomolecules 10, 1–17 (2020). published maps and institutional affiliations.
317. Thomas, J. D. et al. RNA isoform screens uncover the essentiality and tumor-suppressor
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this
activity of ultraconserved poison exons. Nat. Genet. 52, 84–94 (2020).
article under a publishing agreement with the author(s) or other rightsholder(s); author self-
This study reports a high-throughput functional analysis of poison exons, revealing
archiving of the accepted manuscript version of this article is solely governed by the terms
their functions in cancer biology.
of such publishing agreement and applicable law.

Acknowledgements
The authors thank M. Irimia, B. Lehner, members of our group and four reviewers for Related links
suggestions on the manuscript. M.E.R. was supported by a MSCA Postdoctoral Fellowship Genome Aggregation Database: https://gnomad.broadinstitute.org/
and C.V. by a FPI-Severo Ochoa PhD Fellowship from the Spanish Ministry of Economy and
Competitiveness. Work in the authors' laboratory has been supported by the European © Springer Nature Limited 2022

Nature Reviews Genetics