concept · intermediate · partly written
Splicing
Splicing cuts introns out of pre-mRNA and joins the exons by two RNA-catalysed reactions, run by a machine of more than 100 proteins rebuilt on every intron.
Also called: pre-mRNA splicing, spliceosome, alternative splicing
The chemistry, measured in vitro more
Two transesterifications. Step 1: the 2'-hydroxyl of an adenosine in the intron (the branch point) attacks the 5' splice site, releasing the 5' exon and leaving a lariat on the 3' exon. Step 2: the freed exon's 3'-hydroxyl attacks the 3' splice site, joining the exons and releasing the lariat. The catalytic centre is RNA: U6 small nuclear RNA folds with U2 and holds two magnesium ions. A direct test in yeast extract swapped single phosphate oxygens of U6 for sulfur at 20 positions; five blocked branching and all five were rescued by cadmium. Limits: no ATP, non-physiological metals, n = 2 to 3. In human C* maps the two metals sit about 4.2 angstroms apart, but map resolution is 3.0 to 4.5, so that distance is a model fit. 123
Reported, not yet replicated
The parts more
Yeast: about 70 spliceosomal proteins. Human: more than 100 (the only human count the sources give), five snRNAs (U1, U2, U4, U5, U6), at least ten assembly states and eight DExD/H helicases (RNA-unwinding ATPases). About 85 core proteins are shared with yeast. The 'PRP' names are yeast 'pre-mRNA processing' mutant-screen labels, a historical accident of that screen; HGNC symbols differ (for example Prp16 is DHX38, DEAH-box helicase 38, and Prp8 is PRPF8, pre-mRNA processing factor 8; HGNC lookups on 2026-11-17). The minor (U12-type) spliceosome handles under 1% of introns: in one K562 set, 182 introns against 139,344 with GU...AG ends. 1345
Reported, not yet replicated
How a site is recognised, in order more
U1 snRNP pairs with the 5' splice site; SF1 and U2AF bind the branch point and the polypyrimidine tract; U2 replaces SF1 with the branch adenosine bulged out (human A-like complex at 8.4 angstroms overall, 3.1 locally). The tri-snRNP joins, U1 is unwound from the 5' site (pre-B to B). Many proteins swap as U4 leaves (human Bact: 52 proteins, about 1.8 megadaltons, 4.9 to 6.5 angstroms). Step 1, then the branch helix is undocked and the 3' site docks for step 2 (human P complex at 3.3 angstroms, stalled by a dead helicase). Method caveat: human complexes were frozen by stalling (a dead helicase, a missing 3' exon, a drug) and ordered from snapshots; the order is inferred from several traps, not filmed in one molecule. 6783
Reported, not yet replicated
Finding a short exon in a long intron more
Median human protein-coding values: exon 133 nt, intron 1,632 nt, 8 exons per transcript (one gene-statistics database). The model of exon definition says U1 on the downstream 5' site and U2 with U2AF on the upstream 3' site pair across the exon first, then convert to span the intron. Evidence runs from in vitro (elongated exons without a downstream 5' site assembled poorly; an exon under about 300 nt was needed in one reporter series) to yeast structure. Against it in living human cells: a review reports that lariat formation can finish with the polymerase only 18 to 26 bases past the 3' site, so exon definition may be less favoured when splicing is co-transcriptional. This atlas found no human structure of a cross-exon complex. Direct in-cell human evidence is missing and some points against. 91
Contested
Fidelity: proofreading by helicases more
Helicases such as Prp16 (DHX38) are proposed as kinetic proofreaders: a poor site that fails to dock in a time window is rejected and resampled. In yeast extract, single-molecule FRET (a distance ruler) showed Prp16 moving U80-PS(SP)-stalled spliceosomes from a high FRET of 0.73 to near zero (0.07), on trajectories of about 1 s. The claim that accuracy and flexibility are one setting is supported by yeast extract only; it is not a measured property of human cells. 10
Reported, not yet replicated
How often the choice is wrong more
No direct per-intron human error rate was found. Indirect numbers: about 0.7% of reads at junctions with conserved sites go to unconserved sites, about 2% of transcripts of a median four-exon gene (75 lymphoblastoid lines, 35 to 46 bp reads, about 1.4 billion reads; conservation stands in for function). In GTEx v8 (14,408 samples, 42 tissues) 3,865,268 novel junctions over 324,956 annotated introns made 0.32 to 1.08% of junction reads; the 3' site is less accurate than the 5' site, and knockdown of 56 splicing genes raised mis-splicing in 89.3% of genes at the 5' site and 91.1% at the 3' site. (That preprint is the version opened.) The numbers are floors, since nonsense-mediated decay removes some products: about 50% of aberrant mRNAs in SF3B1-mutant cells, the only measured decay share, and not wild-type. Median intron splicing time in K562 is 7.2 min (a first-order TT-seq model explaining 19% of donor half-life variance); two nascent-RNA methods disagree on how much splicing precedes polymerase movement (about 40% of reads already spliced in HeLa by POINT-nano; splicing several kb downstream by another). 111213414
Reported, not yet replicated
Alternative splicing: function or noise, both readings more
How common: 92 to 94% of human genes are reported alternatively spliced, a sigmoid extrapolation from short reads, with 86% having a minor isoform at 15% or more. Reading A (mostly noise at protein level): across eight proteomics sets and 277,244 peptides, evidence of any splice event for 246 genes (282 events, 0.62% of annotated), more than one alternative isoform for 20 genes; 99.63% of detected peptides map to one isoform per gene. Reading B (many translated): at least 75% of exon-skipping events in moderate-to-high abundance transcripts appear in ribosome-profiling data; deep proteomics found junction peptides for 4,608 of 13,450 (34.3%) alternative events and for both isoforms in 1,141 of 6,145 events with expression evidence for both (18.6%), with 64% detection for frame-preserving events in abundant transcripts. Agreement: one dominant isoform per gene; real conserved exon switches. Disagreement: the tail. One side's reply concedes 'the short answer is that no-one knows'. The 2017 reply on the other side was not read. See questions/how-much-alternative-splicing-is-functional. 151617181920
Contested
How it is measured, and the limits more
Short-read RNA-seq counts junctions but cannot phase distant exons and hides decayed products. Long reads solve phasing, not accuracy or annotation: a multi-group benchmark of more than 427 million reads found accurate prediction of novel transcripts challenging. A nanopore survey reported just over 70,000 novel transcripts of annotated genes, with protein-level validation for about 10% (abstract). Minigene reporters work out of native context (internal exons of 100 nt or less, about half of exons outside this; 81% concordance with patient tissue, n = 32). Every human error rate here depends on a junction call, on conservation or annotation, and on how much decay removed. See assays/rna-seq and assays/long-read-rna-sequencing. 212223
Reported, not yet replicated
Where it breaks in disease deep
SF3B1 hotspot mutations shift branch-point use so cryptic upstream 3' sites are chosen, and about half of the aberrant mRNAs are removed by decay. The SF3b binder H3B-8800 in a phase 1 of 84 patients gave no responses by the formal criteria; transfusion independence in 9 of 62 transfusion-dependent patients (15%); in SF3B1-mutant MDS patients the abstract says 5 of 15. SpliceAI estimated 9 to 11% of pathogenic rare-disease mutations act through cryptic splicing; a reporter assay found about 10% (513 of 4,964) of disease-causing exonic mutations altered splicing in both assays. SMN2 differs from SMN1 by one silent C-to-T at exon 7 position +6; in HEK293 minigenes SMN2 gave 1 +/- 2% exon 7 inclusion and SMN1 98 +/- 3% (why is disputed: enhancer loss against silencer gain), and the antisense drug nusinersen gave a motor-milestone response of 51% (37 of 73) against 0% (0 of 37) sham in infants (abstract). 132425232627
Reported, not yet replicated
Plain answer
Splicing is one RNA-catalysed reaction done twice: a branch-point adenosine attacks the 5' splice site, then the freed exon attacks the 3' site, joining the exons and releasing a lariat. A machine of five small nuclear RNAs and more than 100 proteins assembles anew on each intron. The chemistry and the yeast machine are well measured; human error rates and the function of rare isoforms are not. 13
Reported, not yet replicated
Why it matters
Still open here
- How does it work, and can two transcripts from different parts of the genome be fused? The how is answered below; the fusion half of his question was not covered by this report.
- No direct measurement of the human per-intron error rate; see questions/what-is-the-human-per-intron-splicing-error-rate.
- Exon definition in living human cells: the primary CoLa-seq paper and the 1992 Hoffman and Grabowski paper were not opened.
- Only 'more than 100' is verified as the human spliceosome protein count; human structures are stalled snapshots at lower resolution than yeast.
- Group II introns and the spliceosome: kinship by structure is documented (see concepts/conservation-of-gene-expression-machinery); common origin is inferred.
Not yet written: Equation, Real numbers.
Linked from
- Long-read RNA sequencing (nanopore, PacBio)
- Alternative start codons and two protein forms
- The central dogma as Crick stated it, against how it is retold
- Conservation of the gene-expression machinery across bacteria, archaea and eukaryotes
- Transcription by RNA polymerase II: initiation, pausing, elongation, termination
- Where the textbook arrow fails or is incomplete
- Nusinersen (Spinraza)
- Antisense oligonucleotide (ASO)
- Christopher B. Burge
- How much of observed alternative splicing makes a protein that does something?
- What is the splicing error rate per intron in a living human cell?
Sources
- Plaschka C, Newman AJ, Nagai K. Structural Basis of Nuclear pre-mRNA Splicing: Lessons from Yeast.. Cold Spring Harb Perspect Biol (2019). paper; opened 2026-11-17 all records citing this
- Fica SM, Tuttle N, Novak T, et al.. RNA catalyses nuclear pre-mRNA splicing.. Nature (2013). paper; opened 2026-11-17 all records citing this
- Kastner B, Will CL, Stark H, et al.. Structural Insights into Nuclear pre-mRNA Splicing in Higher Eukaryotes.. Cold Spring Harb Perspect Biol (2019). paper; opened 2026-11-17 all records citing this
- Wachutka L, Caizzi L, Gagneur J, et al.. Global donor and acceptor splicing site kinetics in human cells.. Elife (2019). paper; opened 2026-11-17 all records citing this
- HGNC REST service (genenames.org): gene symbol lookups. HUGO Gene Nomenclature Committee (2026). database, not peer reviewed; opened 2026-11-17 all records citing this
- Cretu C, Gee P, Liu X, et al.. Structural basis of intron selection by U2 snRNP in the presence of covalent inhibitors.. Nat Commun (2021). paper; opened 2026-11-17 all records citing this
- Zhang X, Yan C, Zhan X, et al.. Structure of the human activated spliceosome in three conformational states.. Cell Res (2018). paper; opened 2026-11-17 all records citing this
- Fica SM, Oubridge C, Wilkinson ME, et al.. A human postcatalytic spliceosome structure reveals essential roles of metazoan factors for exon ligation.. Science (2019). paper; opened 2026-11-17 all records citing this
- Shenasa H, Bentley DL. Pre-mRNA splicing and its cotranscriptional connections.. Trends Genet (2023). paper; opened 2026-11-17 all records citing this
- Semlow DR, Blanco MR, Walter NG, et al.. Spliceosomal DEAH-Box ATPases Remodel Pre-mRNA to Activate Alternative Splice Sites.. Cell (2016). paper; opened 2026-11-17 all records citing this
- Pickrell JK, Pai AA, Gilad Y, et al.. Noisy splicing drives mRNA isoform diversity in human cells.. PLoS Genet (2010). paper; opened 2026-11-17 all records citing this
- García-Ruiz S, Zhang D, Gustavsson EK, et al.. Splicing accuracy varies across human introns, tissues and age.. bioRxiv (2023). preprint, not peer reviewed; opened 2026-11-17 all records citing this
- Darman RB, Seiler M, Agrawal AA, et al.. Cancer-Associated SF3B1 Hotspot Mutations Induce Cryptic 3' Splice Site Selection through Use of a Different Branch Point.. Cell Rep (2015). paper; opened 2026-11-17 all records citing this
- Sousa-Luís R, Dujardin G, Zukher I, et al.. POINT technology illuminates the processing of polymerase-associated intact nascent transcripts.. Mol Cell (2021). paper; opened 2026-11-17 all records citing this
- Wang ET, Sandberg R, Luo S, et al.. Alternative isoform regulation in human tissue transcriptomes.. Nature (2008). paper; opened 2026-11-17 all records citing this
- Abascal F, Ezkurdia I, Rodriguez-Rivas J, et al.. Alternatively Spliced Homologous Exons Have Ancient Origins and Are Highly Expressed at the Protein Level.. PLoS Comput Biol (2015). paper; opened 2026-11-17 all records citing this
- Tress ML, Abascal F, Valencia A. Alternative Splicing May Not Be the Key to Proteome Complexity.. Trends Biochem Sci (2017). paper; opened 2026-11-17 all records citing this
- Weatheritt RJ, Sterne-Weiler T, Blencowe BJ. The ribosome-engaged landscape of alternative splicing.. Nat Struct Mol Biol (2016). paper; opened 2026-11-17 all records citing this
- Sinitcyn P, Richards AL, Weatheritt RJ, et al.. Global detection of human variants and isoforms by deep proteome sequencing.. Nat Biotechnol (2023). paper; opened 2026-11-17 all records citing this
- Tress ML, Abascal F, Valencia A. Most Alternative Isoforms Are Not Functionally Important.. Trends Biochem Sci (2017). paper; opened 2026-11-17 all records citing this
- Pardo-Palacios FJ, Wang D, Reese F, et al.. Systematic assessment of long-read RNA-seq methods for transcript identification and quantification.. Nat Methods (2024). paper; opened 2026-11-17 all records citing this
- Glinos DA, Garborcauskas G, Hoffman P, et al.. Transcriptome variation in human tissues revealed by long-read sequencing.. Nature (2022). paper; opened 2026-11-17 all records citing this
- Soemedi R, Cygan KJ, Rhine CL, et al.. Pathogenic variants that alter protein code often disrupt splicing.. Nat Genet (2017). paper; opened 2026-11-17 all records citing this
- Steensma DP, Wermke M, Klimek VM, et al.. Phase I First-in-Human Dose Escalation Study of the oral SF3B1 modulator H3B-8800 in myeloid neoplasms.. Leukemia (2021). paper; opened 2026-11-17 all records citing this
- Jaganathan K, Kyriazopoulou Panagiotopoulou S, McRae JF, et al.. Predicting Splicing from Primary Sequence with Deep Learning.. Cell (2019). paper; opened 2026-11-17 all records citing this
- Cartegni L, Hastings ML, Calarco JA, et al.. Determinants of exon 7 splicing in the spinal muscular atrophy genes, SMN1 and SMN2.. Am J Hum Genet (2006). paper; opened 2026-11-17 all records citing this
- Finkel, R. S., et al.. Nusinersen versus sham control in infantile-onset spinal muscular atrophy (ENDEAR). New England Journal of Medicine (2017). paper; opened 2026-11-17 all records citing this
- Fong N, Kim H, Zhou Y, et al.. Pre-mRNA splicing is facilitated by an optimal RNA polymerase II elongation rate.. Genes Dev (2014). paper; opened 2026-11-17 all records citing this
Where it sits on the map
History of this record (8)
This record has been changed 8 times, on the dates below. The note kept against each change is written for the people who maintain the atlas, in their own shorthand, so it is not shown here; it is in the record itself.
- Status
- partly written
- Last updated
- 2026-11-19
- Record
- content/concepts/splicing.json