Alternative splicing lets a single gene produce multiple different proteins by rearranging which segments of its genetic instructions get included in the final message. This is one of the most powerful ways cells control what proteins they make, where they make them, and how much of each protein is available at any given time. About 95% of human multi-exon genes undergo alternative splicing, and the average human gene produces roughly seven different transcript versions. That means our roughly 20,000 protein-coding genes generate close to 150,000 distinct transcripts, effectively quadrupling the number of protein forms compared to the gene count alone.
From One Gene to Many Proteins
When a gene is first copied into a preliminary RNA message, that message contains both coding segments (exons) and non-coding segments (introns). Before the message can be used to build a protein, the introns need to be cut out and the exons stitched together. In standard splicing, every exon is included in order. Alternative splicing deviates from that default: certain exons can be skipped, extra sequences can be kept in, or the precise cut points can shift. The result is that one gene’s RNA can be assembled in several different configurations, each producing a protein with a slightly (or dramatically) different structure and function.
This matters because proteins do their jobs through their physical shape. Removing or swapping a section of a protein can change which molecules it binds to, whether it acts as an enzyme, where it sits in the cell, or whether it’s active at all. A protein missing one particular domain might promote cell growth while a version of the same protein with that domain included might suppress it.
The Main Types of Alternative Splicing
There are several patterns cells use to rearrange exons, and they show up at different frequencies across the genome.
- Exon skipping: The most common form, accounting for about 30% of alternative splicing events in vertebrates. An entire exon is left out of the final message, shortening the protein or removing a functional domain.
- Alternative splice site selection: Instead of skipping a whole exon, the molecular machinery cuts at a slightly different spot within an exon, either at the beginning (5′ end) or the end (3′ end). This accounts for roughly 25% of events and tends to make subtler changes to the protein.
- Mutually exclusive exons: Two or more exons compete for the same slot. Only one can be included in any given transcript, so the cell essentially chooses between protein variants with different properties.
- Intron retention: A segment that would normally be removed is instead kept in the final message. In humans, retained introns tend to appear in non-coding regions of the transcript and are associated with weaker splice signals and shorter intron lengths.
These patterns can also be combined within a single gene. The fruit fly gene Dscam1 illustrates the extreme end of this: it contains three clusters of variable exons, and selecting one exon from each cluster can generate over 19,000 distinct protein combinations from a single gene. That’s comparable to the total number of genes in the fly’s entire genome.
How Cells Decide Which Version to Make
Splicing decisions aren’t random. They’re controlled by a tug-of-war between two forces: short signal sequences embedded in the RNA itself, and proteins that read those signals.
The signal sequences fall into four categories. Some sit within exons and encourage inclusion of that exon (enhancers), while others sit within exons and promote skipping (silencers). The same logic applies to sequences within introns: some recruit the splicing machinery closer, and others push it away. The two major families of proteins that interpret these signals are SR proteins and hnRNPs. These two protein families generally have opposing effects. SR proteins tend to promote exon inclusion, while hnRNPs tend to promote exon skipping, though their actual behavior depends on exactly where they bind on the RNA.
The balance between these competing signals determines which version of a protein gets made. Because different cell types express different amounts of these regulatory proteins, the same gene can be spliced differently in the brain than in the heart, or differently in an embryo compared to an adult.
Tailoring Proteins to Specific Tissues
One of the most important consequences of alternative splicing is tissue-specific protein production. The tissues with the most distinctive splicing patterns are nervous tissue and cardiac tissue. In muscle (particularly the heart), alternatively spliced genes tend to involve the structural components of muscle fibers, especially proteins that make up the contractile machinery. In the brain, the alternatively spliced genes are more often involved in cell-to-cell connections and the internal scaffolding of neurons.
The gene NEBL offers a clean example. It produces two main protein forms that differ at one end. The longer form, called nebulette, appears exclusively in heart tissue, where it plays a role in muscle fiber structure. The shorter form shows up primarily in nervous and urinary tissues and is absent from muscle. Same gene, different protein, different tissue, different job.
In the nervous system, this tissue-specific splicing helps wire the brain. Neurons use alternative splicing of surface proteins to generate unique molecular “identity tags.” When two branches growing from the same neuron encounter each other, matching tags cause them to repel and spread apart, ensuring proper coverage. This self-avoidance mechanism depends on each neuron expressing a distinct combination of splice variants.
Controlling Gene Output Without Changing Transcription
Alternative splicing doesn’t just change which protein gets made. It can also control whether any functional protein gets made at all. This works through a quality-control system called nonsense-mediated decay. When splicing introduces a premature stop signal into an RNA message, the cell recognizes it as defective and destroys it before it can be translated into protein. The general rule is that if a stop signal appears more than 50 to 55 nucleotides before the last exon-exon junction, the transcript gets flagged for destruction.
Cells exploit this deliberately. The brain protein PSD-95, which is critical for the connections between neurons, is regulated this way. When exon 18 of its gene is included, the protein is produced normally. When exon 18 is skipped, the reading frame shifts, a premature stop signal appears in the next exon, and the transcript is destroyed without ever making protein. This acts as an on-off switch for gene expression that operates entirely independently of whether the gene is being actively copied into RNA. The cell can be transcribing the gene at full speed, but if splicing routes most transcripts toward the degradation-prone version, protein output drops.
When Splicing Goes Wrong
Given how much of human biology depends on correct splicing, it’s not surprising that splicing errors cause disease. Somewhere between 15% and 50% of all disease-causing mutations in humans disrupt splicing signals or the proteins that read them. These mutations can strengthen or weaken splice sites, create new ones where none should exist, or disable the regulatory sequences that guide exon selection.
Spinal Muscular Atrophy
One of the clearest examples is spinal muscular atrophy, a neurodegenerative disease caused by loss of the SMN protein, which motor neurons need to survive. Humans carry two copies of the gene that makes this protein: SMN1 and SMN2. In people with the disease, SMN1 is non-functional. SMN2 is almost identical but has a single-letter change that causes exon 7 to be skipped most of the time, producing a shortened, unstable protein. The gene is there, it’s being transcribed, but alternative splicing renders most of its output useless. The approved drug nusinersen works by binding to a silencer sequence in SMN2’s RNA, blocking the signal that causes exon 7 skipping. This shifts splicing back toward the full-length version and increases functional protein levels.
Cancer
Cancer cells frequently hijack alternative splicing to gain survival advantages. The gene VEGFA, which controls blood vessel growth, produces both pro-angiogenic and anti-angiogenic protein variants depending on which of its final exons is included. The anti-angiogenic version binds to the same receptor but can’t activate the signaling cascade, so it effectively blocks new blood vessel formation. As tumors progress, expression of this anti-angiogenic version typically decreases, tipping the balance toward the pro-growth variants that feed the tumor’s blood supply.
Splicing of the CD44 gene follows a similar pattern. Different splice variants of CD44 are associated with increased cell invasion and the ability of cancer cells to colonize distant organs like the lungs. Other genes involved in cell adhesion and migration, including RAC1, RON, and MENA, also produce invasion-promoting splice variants that contribute to metastasis.
Alternative Splicing and Evolutionary Complexity
One of the more interesting findings about alternative splicing is that its frequency tracks closely with an organism’s complexity rather than its total gene count. Humans don’t have dramatically more genes than a roundworm, but we have far more alternative splicing. Research across multiple species has found a strong positive correlation between the proportion of alternatively spliced genes and organismal complexity. The splicing machinery itself, particularly the proteins that make up the core splicing complex, shows the highest gene expression levels, the greatest evolutionary conservation, and the most structurally flexible protein regions of any pathway involved in this process. This suggests that the expansion of alternative splicing, rather than the addition of new genes, was a major driver of biological complexity during evolution.