Polyadenylation

Why polyadenylation matters

• It is the mark of a "real" mRNA. With the rare exception of replication-dependent histone mRNAs, essentially every protein-coding transcript in a eukaryotic cell carries a poly(A) tail. That single feature is exploited daily in the lab: oligo-dT beads pull polyadenylated mRNA out of a soup dominated by rRNA and tRNA, and it is the entire basis of standard poly(A)-selected RNA-seq and of oligo-dT priming for cDNA synthesis.
• It sets mRNA lifespan. The half-life of a message is largely governed by how fast its tail is removed. Stable mRNAs (globin, ~10+ hours) keep long tails; short-lived regulatory mRNAs (c-fos, cytokines, ~10-30 minutes) carry AU-rich elements that recruit deadenylases. Tuning tail turnover is one of the fastest ways a cell can change its protein output without touching transcription.
• It couples 3'-end formation to translation. Because PABP on the tail contacts eIF4G at the cap, the tail is a positive translational element. In frog and fly eggs, thousands of maternal mRNAs are stored with short tails and translationally silent; cytoplasmic re-polyadenylation at fertilization switches them on, driving the earliest events of development before any new transcription occurs.
• It is an immune-system switch. The choice between two poly(A) sites in the immunoglobulin heavy-chain gene determines whether a B cell makes membrane-bound antibody (a B-cell receptor) or secreted antibody — the single molecular decision that turns a naive B cell into an antibody factory.
• It is dysregulated in cancer and disease. Proliferating tumor cells systematically shift to shorter 3' UTRs via alternative polyadenylation, stripping microRNA control off oncogenes like cyclin D1 and IGF2BP1. A single-base mutation in the AAUAAA of alpha-globin causes alpha-thalassemia, and mutations affecting FIP1L1 and CSTF2 have been tied to blood and other disorders.
• It underpins modern medicine. Every mRNA vaccine and mRNA therapeutic ends in an engineered poly(A) tail — typically ~100-120 encoded A's, often with a segmented design — because tail length directly tunes how long the mRNA survives and how much protein it makes inside the patient's cells.

How polyadenylation works, step by step

3'-end formation is a two-reaction machine — an endonucleolytic cleavage followed by template-independent tail synthesis — carried out by a ~20-subunit assembly that is recruited co-transcriptionally onto RNA polymerase II. The whole thing is orchestrated so that the cut and the tail happen almost simultaneously, leaving no naked 3' end exposed to nucleases.

1. Signal recognition. As Pol II transcribes past the end of the gene, the nascent RNA reveals the polyadenylation signal, a small constellation of elements: the upstream AAUAAA hexamer (~10-30 nt before the cut), a downstream GU-rich / U-rich element, and often auxiliary upstream sequence elements. The cleavage and polyadenylation specificity factor (CPSF) — through its CPSF160, WDR33, and CPSF30 subunits — directly reads the AAUAAA bases, while cleavage stimulation factor (CstF) grips the downstream GU-rich element via CstF-64.

2. Cleavage. With the site defined, the endonuclease subunit CPSF73 (CPSF3), a metallo-β-lactamase-fold nuclease, cuts the phosphodiester backbone, usually just after a CA dinucleotide. This produces an upstream fragment (the future mRNA) and a downstream fragment that is quickly destroyed. Cleavage factors I and II (CFIm, CFIIm) help position and stabilize the complex; CFIm (CPSF5/CPSF6) in particular biases which site is chosen and is a major regulator of alternative polyadenylation.

3. Tail synthesis. The newly generated 3'-hydroxyl is handed straight to poly(A) polymerase (PAP), tethered by CPSF. PAP is a template-independent nucleotidyltransferase: it uses ATP, adds only adenosines, and releases pyrophosphate. Nuclear poly(A)-binding protein PABPN1 coats the growing tail and, together with CPSF, keeps PAP processive. When the tail reaches roughly 200 to 250 nucleotides in mammals, the PABPN1-CPSF-PAP grip is lost and synthesis stops — a "molecular ruler" that measures length by protein occupancy rather than counting bases.

4. Coating, export, and translation. In the cytoplasm the tail is bound by cytoplasmic poly(A)-binding protein PABPC1, roughly one PABP per ~27 adenosines. PABPC1 shields the 3' end from exonucleases and, crucially, binds eIF4G in the cap-binding complex, physically circularizing the mRNA (the "closed-loop" model). This closed loop enhances ribosome recruitment and reinitiation, so a longer, better-coated tail generally means more efficient translation.

5. Deadenylation and decay. The tail is not permanent. Two deadenylase complexes — PAN2-PAN3 (initial trimming) and the CCR4-NOT complex (bulk removal, via its CNOT6/CCR4 and CNOT7/CAF1 nucleases) — progressively shorten it. As adenosines vanish, PABP molecules dissociate; below ~10-12 A's the message is decapped by DCP1-DCP2 and degraded 5'→3' by Xrn1, or attacked 3'→5' by the exosome. Deadenylation is the committed, usually rate-limiting first step of mRNA turnover.

The 3'-end processing machinery

Factor / complex	Key subunits	Role in polyadenylation
CPSF	CPSF160, WDR33, CPSF30, CPSF100, CPSF73, FIP1	Reads AAUAAA; CPSF73 is the endonuclease; recruits and tethers PAP
CstF	CstF-64, CstF-77, CstF-50	Binds the downstream GU/U-rich element; helps set cleavage position
CFIm	CPSF5 (NUDT21), CPSF6	Recognizes UGUA upstream element; major regulator of poly(A)-site choice (APA)
CFIIm	CLP1, PCF11	Assists cleavage; PCF11 links to transcription termination
Poly(A) polymerase	PAP (PAPOLA)	Template-independent addition of adenosines using ATP
PABPN1	nuclear PABP	Coats growing tail; with CPSF sets ~200-250 nt length limit
PABPC1	cytoplasmic PABP	Protects tail; bridges to eIF4G for closed-loop translation
CCR4-NOT / PAN2-PAN3	CNOT6/7, PAN2	Deadenylases that shorten the tail and initiate decay

Nuclear vs cytoplasmic polyadenylation

Property	Nuclear (constitutive) polyadenylation	Cytoplasmic polyadenylation
Where	Nucleus, co-transcriptional	Cytoplasm, on mature stored mRNA
Coupled to cleavage?	Yes — cleavage then tail addition	No — re-extends an existing short tail
Signal	AAUAAA + downstream GU-rich element	CPE (cytoplasmic polyadenylation element) + AAUAAA, bound by CPEB
Polymerase	Canonical PAP (PAPOLA)	GLD-2-type PAP (e.g. TENT2/GLD2)
Length made	~200-250 A's, then capped by the ruler	Variable; extends silenced tails to activate
Purpose	Make a stable, exportable, translatable mRNA	Switch stored maternal / synaptic mRNAs ON
Classic setting	Every new protein-coding transcript	Oocyte maturation, early embryo, synaptic plasticity

Common misconceptions

• The poly(A) tail is encoded in the gene. It is not. There is no run of ~200 T's in genomic DNA at the end of a gene; poly(A) polymerase synthesizes the tail template-independently after the transcript is cleaved. The genome encodes only the AAUAAA signal and the downstream element that mark where to cut.
• All mRNAs have a poly(A) tail. Replication-dependent histone mRNAs in metazoans are the famous exception — they end in a conserved stem-loop bound by SLBP, processed by U7 snRNP and (again) the CPSF73 endonuclease, with no tail at all. Their abundance is instead tuned to S phase.
• The tail simply "protects the end," full stop. Protection is only one of three jobs. The tail also licenses export and, through PABP-eIF4G bridging, actively stimulates translation. A message can be stabilized yet translationally silent (short-tailed maternal mRNAs) or long-tailed and highly translated — length and function are linked.
• Polyadenylation happens after transcription finishes. It is co-transcriptional. CPSF and CstF ride on the Ser2-phosphorylated CTD of Pol II and are delivered to the RNA the instant the signal emerges. In fact cleavage causes termination: the Xrn2 "torpedo" degrades the downstream RNA and dislodges Pol II.
• A longer tail always means more protein. In steady-state somatic cells the correlation between absolute tail length and translation is surprisingly weak once tails are above a threshold; below it, and in eggs/embryos, tail length becomes a strong translational switch. Context matters — the closed loop needs PABP occupancy, not maximal length.
• Prokaryotes don't polyadenylate. They do — but with the opposite meaning. In E. coli, poly(A) polymerase I (PcnB) adds short A-rich tails that recruit the degradosome and accelerate RNA decay. Bacterial polyadenylation is a destruction tag, not a stabilizing one.

Alternative polyadenylation and its consequences

Most human genes contain more than one usable poly(A) site, so a single gene routinely produces mRNAs with different 3' ends. This alternative polyadenylation (APA) comes in two flavors with very different outcomes. When the competing sites both lie in the terminal exon, only the length of the 3' untranslated region (3' UTR) changes — the protein is identical, but a shorter UTR can shed microRNA-binding sites and AU-rich destabilizing elements, so the same protein is made at a higher level. When an alternative site lies in an intron or an internal exon, APA truncates the coding sequence and changes the protein itself.

The textbook example of the second type is the immunoglobulin μ heavy chain. Resting B cells use a downstream poly(A) site to make the membrane-anchored form (the B-cell receptor); upon activation, a shift toward an upstream site — driven by rising CstF-64 — produces the secreted antibody. APA is globally regulated too: rapidly dividing cells and many tumors favor shorter 3' UTRs, an oncogenic escape from microRNA control, whereas differentiated neurons express some of the longest 3' UTRs in the body, packed with localization and regulatory elements. The CFIm factor (CPSF5/NUDT21) is a master dial: lowering it globally shortens 3' UTRs and can transform cells.

Key experiments and history

• Discovery of poly(A) (late 1960s-70s). Mary Edmonds and Richard Abrams, and independently George Brawerman and others, detected long runs of adenylate at the 3' ends of mRNA and isolated a poly(A) polymerase activity from nuclei. This established that the tail is a real, enzymatically added feature rather than an artifact.
• The AAUAAA signal (1976). Nick Proudfoot and George Brownlee compared the 3' ends of mRNAs and found the conserved hexanucleotide AAUAAA a short distance before the poly(A), publishing the sequence in Nature. Point-mutation studies over the following decade proved that changing AAUAAA (e.g. to AAGAAA) abolishes both cleavage and tail addition.
• Cleavage precedes tailing (1980s). In vitro processing extracts from Moore, Sharp, Manley, Keller and others separated the reaction into an endonucleolytic cut followed by poly(A) addition, and purified the factors — CPSF, CstF, cleavage factors, and PAP — that carry out each step, defining the machinery still studied today.
• The alpha-thalassemia mutation. A human family with alpha-thalassemia was found to carry a single A→G change in the AATAAA of an alpha-globin gene (AATAAA→AATAAG). The mutation cripples 3'-end processing, letting Pol II read through, and drastically reduces functional globin mRNA — clinical proof that the six-base signal is essential in humans.
• Genome-wide APA maps (2000s-2010s). 3'-end sequencing methods (3'-seq, PolyA-seq, PAS-seq) revealed that the great majority of human genes use multiple poly(A) sites and that site choice shifts systematically with proliferation, differentiation, and cancer — turning APA from a curiosity into a recognized layer of gene regulation.