A splice junction is the precise boundary where a coding segment of a gene (an exon) meets a non-coding segment (an intron) in a freshly copied RNA molecule. Cells use these boundaries as cut-and-paste guides: molecular machinery recognizes the junctions, snips out the introns, and stitches the exons together to form a finished messenger RNA ready to be translated into protein. Getting those cuts right, down to a single nucleotide, is so important that errors at splice junctions are now linked to thousands of inherited diseases and many cancers.
How RNA Gets Edited After Copying
When a gene is first copied from DNA into RNA, the result is a rough draft called precursor messenger RNA, or pre-mRNA. This draft contains all the exons (the parts that code for protein) interleaved with introns (stretches that do not end up in the final protein). Before the cell can use the message, it has to remove every intron and join the flanking exons seamlessly. That removal-and-joining process is called splicing, and the spots where it happens are splice junctions.
The heavy lifting is done by the spliceosome, a large complex built from five small nuclear RNAs and dozens of proteins that assembles fresh on each intron.1PubMed Central. Mechanisms and regulation of spliceosome-mediated pre-mRNA splicing in Saccharomyces cerevisiae The spliceosome carries out two chemical reactions in sequence: it first cuts one end of the intron and loops it back on itself to form a lasso-shaped structure called a lariat, then it cuts the other end and joins the two exons together.2PubMed. Molecular choreography of pre-mRNA splicing by the spliceosome The whole operation is fast and remarkably precise, but it depends entirely on the spliceosome finding the right junctions.
Most human introns are processed by the “major” spliceosome, but a small subset, known as U12-type introns, are handled by a separate “minor” spliceosome that shares only one of its five RNA components with the major version.3PubMed. Structural basis of U12-type intron engagement by the fully assembled human minor spliceosome Mutations that cripple the minor spliceosome cause their own set of developmental disorders, which underscores how critical even this less-studied pathway is.
What the Spliceosome Actually Recognizes
Splice junctions are not random. The nucleotides right at the boundary follow a highly conserved pattern. Most introns begin with the two-letter sequence GU and end with AG, a motif so reliable that it was one of the first rules discovered when scientists stumbled upon introns in the late 1970s. Beyond those two letters, the surrounding nucleotides also carry a statistical bias: efficient splicing requires a particular nucleotide composition around the junction, and that preferred composition tends to encode amino acids associated with flexible, disordered regions of proteins rather than rigid, structured ones.4PubMed Central. Splice junctions are constrained by protein disorder In other words, the splicing code and the protein-structure code overlap, placing an evolutionary constraint on both simultaneously.
Short DNA motifs near the junction, called splicing regulatory elements, act as landing pads for helper proteins. Some of these elements recruit proteins that encourage the spliceosome to include a particular exon, while others recruit proteins that tell the spliceosome to skip it. Two major families of these helper proteins often compete at the same junction: one family generally promotes exon inclusion, and the other generally promotes exon skipping.5Signal Transduction and Targeted Therapy. Alternative splicing and related RNA binding proteins in human health and disease The outcome of that tug-of-war determines which version of the RNA the cell ends up with.
Alternative Splicing and Protein Diversity
If every gene were spliced the same way every time, we would have roughly one protein per gene. Instead, the same pre-mRNA can be cut at different combinations of splice junctions to produce multiple distinct messenger RNAs, and therefore multiple distinct proteins, from a single gene. This phenomenon, called alternative splicing, is a major reason the human proteome is far more complex than the roughly 20,000 protein-coding genes in the genome would suggest.6PubMed Central. Alternative Splicing and Isoforms: From Mechanisms to Diseases
Alternative splicing is not random variation. It is tightly regulated by tissue type, developmental stage, and even environmental signals. A heart cell and a brain cell reading the same gene can produce different protein variants (called isoforms) by selecting different exons. A proteomics study that mapped splice-junction-derived protein isoforms across a dozen human tissues identified more than 1,500 non-standard protein forms, including roughly 400 that had never been catalogued in major reference databases.7Cell Reports. A Splice-Junction-Centric Proteomic Approach Maps Alternative Splicing across Human Tissues and Cell Lines Those numbers keep growing as detection methods improve.
A concrete example: a calcium-pump gene in rats produces four different protein variants depending on which combination of two small exons is included or excluded at particular splice sites, and the choice varies sharply between tissues.8PubMed Central. Primary structure of rat plasma membrane Ca(2+)-ATPase isoform 4 and analysis of alternative splicing patterns at splice site A The same logic applies across thousands of human genes. Alternative splicing essentially lets the genome punch above its weight, generating a far richer toolkit of proteins than the gene count alone would allow.
When Splice Junctions Go Wrong
Because the spliceosome relies on precise sequence signals to find its cut sites, even a single-letter change in the DNA at or near a splice junction can derail the process. Such mutations can destroy an existing junction, create a fake one, or scramble the regulatory elements that guide the spliceosome. The downstream consequences include exons being skipped entirely, pieces of introns being mistakenly kept in the final RNA, or the reading frame of the protein being thrown off so badly that the cell produces a truncated or nonfunctional protein.9PubMed Central. Splicing mutations in human genetic disorders: examples, detection, and confirmation
One particularly sneaky category is cryptic splice-site activation, where a mutation does not sit at a known junction but instead creates a new one that the spliceosome treats as real. An analysis of disease-associated variants from ClinVar found that about 28% of known pathogenic variants and 22% of uncertain-significance variants were predicted to activate cryptic splice sites, suggesting this mechanism is more common than previously appreciated.10American Journal of Human Genetics. Cryptic Splice-Site Activation Is a Common Cause of Human Genetic Disease In practical terms, a variant that looks harmless based on its position in a coding region could still be pathogenic if it tricks the spliceosome into cutting in the wrong place.
Cancer adds another layer. Tumors frequently carry mutations either in splicing-regulatory elements within specific genes or in the components of the splicing machinery itself.11PubMed Central. Splicing-factor alterations in cancers A large-scale analysis across 33 cancer types confirmed that mutations in splicing-factor genes are recurrent and that those mutations lead to measurably altered splicing patterns.12PubMed. Somatic Mutational Landscape of Splicing Factor Genes and Their Functional Consequences across 33 Cancer Types In inherited cancer predisposition, mutations in tumor suppressor genes can produce a wide spectrum of splicing defects, from whole-exon skipping to partial intron retention and combinations of multiple errors at once.13PubMed Central. Characterization of splice-altering mutations in inherited predisposition to cancer
The Cell’s Own Quality Control
Cells are not defenseless against bad splicing. A surveillance system called nonsense-mediated mRNA decay, or NMD, monitors freshly spliced messenger RNAs as they are first read by the ribosome. If the ribosome encounters a premature stop signal that sits upstream of where it expects a splice-junction marker, the cell tags that RNA for destruction rather than letting it produce a defective protein.14PubMed. Mechanistic links between nonsense-mediated mRNA decay and pre-mRNA splicing in mammalian cells This cleanup system catches a large fraction of splicing errors before they cause harm.
NMD does not only target mistakes. Roughly a third of naturally occurring alternatively spliced mRNAs are also routed to NMD, meaning the cell deliberately produces certain splice variants only to destroy them.14PubMed. Mechanistic links between nonsense-mediated mRNA decay and pre-mRNA splicing in mammalian cells This sounds wasteful, but it gives the cell an extra dial for controlling how much of a given protein gets made: by shifting the ratio of NMD-targeted versus productive splice variants, the cell can tune protein levels without changing how fast the gene is copied. In disease states like sepsis, the rate of NMD-triggering splicing events shifts measurably compared with healthy controls, hinting that the interplay between splicing and NMD is part of broader stress responses.15Life Science Alliance. Predicting nonsense-mediated mRNA decay from splicing events in sepsis using RNA-sequencing data
Drugs That Fix or Exploit Splice Junctions
Understanding splice junctions has already led to approved medicines. The clearest success story involves spinal muscular atrophy (SMA), a devastating motor-neuron disease caused by loss of the SMN1 gene. Humans carry a backup copy called SMN2, but SMN2 usually skips a critical exon during splicing and produces mostly nonfunctional protein. Two different drugs attack this problem from opposite angles, both by manipulating splice junctions.
Nusinersen is an antisense oligonucleotide, a short synthetic strand of modified RNA delivered by injection into the spinal fluid. It binds to a specific spot in the SMN2 pre-mRNA and displaces proteins that normally repress inclusion of the skipped exon, thereby shifting the balance toward a functional protein.16PubMed Central. Counteracting chromatin effects of a splicing-correcting antisense oligonucleotide improves its therapeutic efficacy in spinal muscular atrophy Risdiplam takes a different route: it is a small-molecule pill that also promotes inclusion of the same exon but works by a distinct mechanism, establishing proof of concept that orally available splicing-modifier drugs are feasible.17PubMed Central. Risdiplam, the First Approved Small Molecule Splicing Modifier Drug as a Blueprint for Future Transformative Medicines
Cancer researchers are also exploring small molecules that target the splicing machinery. One experimental compound, H3B-8800, binds a core component of the spliceosome and preferentially kills tumor cells that already carry mutations in splicing factors.18PubMed. A Small-Molecule Splicing Modulator Targets Spliceosome-Mutant Cells The idea is that tumors relying on a crippled spliceosome are more vulnerable to further disruption than normal cells, creating a therapeutic window.
How Researchers Find and Map Splice Junctions
Much of what we know about splice junctions today comes from RNA sequencing. Researchers extract RNA from cells, convert it into short readable fragments, and then map those fragments back to the genome. When a fragment spans the boundary between two exons that are not adjacent in the DNA, the alignment software must “split” the read across the gap, revealing a splice junction. Specialized tools like STAR and MapSplice were built to handle exactly this problem.
STAR uses a fast search algorithm that finds the longest stretch of a read that maps cleanly to the genome, then searches for where the leftover portion maps, effectively detecting junctions without needing to know their locations in advance.19PubMed Central. Mapping RNA-seq Reads with STAR MapSplice takes a complementary approach, emphasizing both sensitivity and specificity in junction detection and performing well even with non-standard junction sequences that do not follow the typical GU-AG rule.20PubMed Central. MapSplice: accurate mapping of RNA-seq reads for splice junction discovery
A limitation of these short-read methods is that each fragment covers only a small stretch of RNA. When a gene has many alternative exons, it can be difficult to figure out which combination of exons appears together in a single complete transcript just by looking at short overlapping pieces. Long-read sequencing technologies from companies like Oxford Nanopore and PacBio address this by reading entire RNA molecules end to end, which lets researchers see the full exon arrangement of each isoform directly.21PubMed Central. Isoform Age – Splice Isoform Profiling Using Long-Read Technologies Long reads have proven especially valuable for cataloguing complex isoforms and non-standard splicing events that short reads tend to miss or mis-assemble.22PubMed Central. Long-Read Sequencing Reveals RNA Splicing Complexity in Human Diseases The latest frontier combines long-read sequencing with single-cell resolution, making it possible to see which splice variants each individual cell produces.23Nature Communications. Single cell and spatial alternative splicing analysis with Nanopore long read sequencing
Seeing the Spliceosome in Action
For decades, the spliceosome was essentially a black box: biochemists knew what went in and what came out, but the internal mechanics were murky because the complex is enormous, flexible, and constantly rearranging itself. That changed with the cryo-electron microscopy revolution. By flash-freezing spliceosome complexes and imaging them with electron beams, structural biologists obtained near-atomic-resolution snapshots of the machine at different stages of its work cycle. These structures revealed how the spliceosome positions the splice sites, how it activates its catalytic core, and how protein rearrangements driven by molecular motors reshape the active site between the two chemical steps.24PubMed Central. Cryo-electron microscopy snapshots of the spliceosome: structural insights into a dynamic ribonucleoprotein machine These images have been transformative for understanding not just normal splicing but also how disease-causing mutations distort the machinery.
Evolutionary Origins of Splicing
Why do cells bother with introns and splicing at all? The leading hypothesis traces the spliceosome back to ancient self-splicing elements called group II introns, which are found in bacteria and in the organelles of eukaryotic cells. Group II introns use the same two-step chemical mechanism as the spliceosome and fold into a strikingly similar catalytic architecture, supporting the idea that the spliceosome evolved from these self-sufficient genetic parasites.25PubMed. The mechanism of splicing as told by group II introns: Ancestors of the spliceosome According to this model, group II introns invaded an ancestral genome, and over time, the host cell domesticated them: the self-splicing RNA broke into pieces that became the spliceosome’s small nuclear RNAs, while protein cofactors replaced functions the RNA originally handled alone.26PubMed Central. Mobile Group II Introns as Ancestral Eukaryotic Elements
Evolution has also acted on splice-junction signals themselves. A study comparing intron-poor organisms spanning fungi, plants, and protists found that several unrelated lineages independently evolved nearly identical strong consensus sequences at the branch point near the 3′ splice junction, a case of convergent evolution.27PLOS Genetics. Evolutionary Convergence on Highly-Conserved 3′ Intron Structures in Intron-Poor Eukaryotes and Insights into the Ancestral Eukaryotic Genome When organisms shed most of their introns over evolutionary time, the ones that remain face stronger selective pressure to maintain crisp junction signals, pushing distant lineages toward the same solution.
CRISPR Tools That Target Splice Junctions Directly
Gene-editing technologies are now being aimed squarely at splice junctions. One approach uses paired guide RNAs and a modified version of the Cas9 enzyme to delete a specific splice junction in DNA, selectively knocking out one gene without disturbing a nearly identical relative. Researchers demonstrated this by targeting splice junctions to separately disable two closely related human genes, FOXO3A and FOXO3B, and in the process proved that FOXO3B, long dismissed as a non-functional pseudogene, actually encodes a real protein.28PubMed Central. A splice junction-targeted CRISPR approach (spJCRISPR) reveals human FOXO3B to be a protein-coding gene
An even more ambitious strategy works at the RNA level rather than the DNA level. Programmable “splice editors” use a deactivated Cas13 protein fused to splicing components to perform trans-splicing on RNA, essentially swapping out large stretches of a faulty transcript with a corrected version. Early demonstrations achieved multi-kilobase RNA replacement in human cell lines and showed that the full editor could be packaged into adeno-associated virus vectors for delivery to living cells.29PubMed Central. Programmable multi-kilobase RNA editing using CRISPR-mediated trans-splicing Because this approach edits the RNA rather than the DNA, the changes are reversible and do not permanently alter the genome, which could offer a safety advantage for therapeutic applications. These tools are still in early stages, but they illustrate how deeply splice-junction biology has moved from basic science into engineering territory.
Backsplicing and Circular RNA
Splice junctions are not always used in the expected linear order. In a process called backsplicing, the spliceosome joins a downstream splice donor to an upstream splice acceptor, creating a covalently closed circle of RNA instead of the usual linear message. These circular RNAs, or circRNAs, are found across animal and plant tissues and are often surprisingly stable because they lack the free ends that enzymes normally chew on.
Backsplicing depends on features in the flanking introns. In one well-characterized system, repetitive DNA elements in the introns on either side of the circularized exon pair up and bring the distant splice sites close together. Removing either of these elements abolished both circular RNA formation and expression of a reporter gene, confirming that the flanking sequences are essential for the process. Interestingly, inserting even a short extra stretch of sequence into the left intron sharply reduced circRNA production, while the right intron tolerated insertions up to about 1,500 nucleotides with no measurable loss. That asymmetry suggests the geometry of the junction and the surrounding sequence context matter in ways researchers are still working out.
The biological functions of most circRNAs remain uncertain. Some act as sponges that absorb small regulatory RNAs, some interact with proteins, and others may even be translated into small peptides under certain conditions. What is clear is that backsplicing represents yet another way cells exploit splice junctions for purposes well beyond simple intron removal, adding another dimension to the already complex landscape of RNA processing.