# Chapter 29. 3' End Formation, Polyadenylation, Tailing Enzymes, and Alternative Polyadenylation

## Scope Note

Most eukaryotic messenger RNAs do not end where RNA polymerase II happens to stop. The mature 3′ end is usually created by an endonucleolytic cleavage reaction in the nascent pre-mRNA, followed by synthesis of a poly(A) tail, a stretch of adenosine residues added without a DNA template. This chapter explains how cleavage and polyadenylation factors recognize RNA sequence elements, how canonical and noncanonical tailing enzymes extend RNA 3′ ends, how uridine and mixed nucleotide tails change RNA fate, and how alternative polyadenylation reshapes transcriptomes by changing coding regions, untranslated regions, localization signals, regulatory motifs, and mRNA stability.

The chapter centers on nuclear RNA polymerase II transcripts, especially metazoan protein-coding pre-mRNAs, but it also compares nuclear mRNAs with noncoding RNAs, pervasive transcription products, organellar and prokaryotic tailing logic, replication-dependent histone mRNAs, and therapeutic RNA design. Replication-dependent histone transcripts are a mechanistically important counterexample: in most metazoan somatic contexts, their mature 3′ ends are made by U7 small nuclear ribonucleoprotein (U7 snRNP)-directed cleavage and do not receive a long poly(A) tail. [Chapter 25](chapter1024.md) covers co-transcriptional processing as a broader nascent-RNA problem; [Chapter 26](chapter1025.md) covers the 5′ cap; Chapters [27](chapter1026.md) and [28](chapter1027.md) cover splicing and alternative splicing; Chapters [30](chapter1029.md) and [31](chapter1030.md) cover mRNP export and nuclear surveillance; Chapters [32](chapter1031.md)-[35](chapter1033.md) cover RNA decay; [Chapter 72](chapter1067.md) covers integrated mRNA architecture; and Chapters [153](chapter1137.md) and [159](chapter1142.md) cover therapeutic mRNA design and manufacturing.

## Executive Summary

The 3′ end of a typical metazoan mRNA is produced by a coupled cleavage and polyadenylation reaction. A pre-mRNA contains an upstream polyadenylation signal, commonly AAUAAA or a close variant, a cleavage site, and downstream sequence elements that recruit a multi-protein processing machinery. Cleavage and polyadenylation specificity factor, or CPSF, recognizes the upstream signal and provides the CPSF73 endonuclease that cuts the RNA. Cleavage stimulation factor, or CstF, and other cleavage factors help define efficient sites, especially through downstream GU-rich or U-rich elements.

Cleavage is followed by nontemplated adenosine addition. Canonical nuclear poly(A) polymerases such as PAPOLA and PAPOLG add adenosines to the new 3′ hydroxyl. Nuclear poly(A)-binding protein PABPN1 stimulates processive tail extension and helps define tail length. The product is not merely a passive handle. A poly(A) tail bound by poly(A)-binding proteins promotes mRNA stability, export competence, and translation, while progressive deadenylation is a common entry point into mRNA decay.

Polyadenylation is not limited to one canonical enzyme or one biological meaning. Noncanonical poly(A) polymerases and terminal nucleotidyltransferases act on small RNAs, damaged or misprocessed RNAs, noncoding RNAs, organellar RNAs, and cytoplasmic mRNAs. In some contexts adenylation stabilizes or activates an RNA; in other contexts it marks the RNA for surveillance or decay. The same chemical addition, an adenosine at a 3′ end, can therefore have different consequences depending on enzyme, substrate, compartment, RNP context, and downstream reader proteins.

Uridylation is the nontemplated addition of uridine residues to an RNA 3′ end. Uridylation can help trigger decay of short-tailed mRNAs, regulate small RNAs, or mark transcripts produced by pervasive transcription termination. Some RNAs carry mixed tails that contain both adenosines and uridines or other terminal nucleotide patterns. Tail-code models propose that tail length, composition, and associated proteins together influence RNA fate, but this idea should not be overstated as a fully decoded universal grammar. The evidence is strongest for particular tailing enzymes and substrates, and weaker for a single general code that applies to every RNA class.

Alternative polyadenylation, or APA, means that one gene or transcription unit can use more than one cleavage and polyadenylation site. APA can produce mRNA isoforms with different 3′ untranslated region lengths, different terminal exons, or different coding sequences. A proximal poly(A) site often shortens the 3′ untranslated region and can remove microRNA sites, RNA-binding protein sites, localization elements, or decay-promoting motifs. A distal site often retains more regulatory sequence. These effects are context-dependent; a shorter 3′ untranslated region does not automatically mean higher protein production.

3′ end formation is coupled to RNA polymerase II elongation, transcription termination, splicing, capping, and export. Cleavage creates an entry point for termination models in which exonucleases degrade the downstream transcript and help terminate polymerase, while processing factors bound near the cap, splice sites, and poly(A) site help determine whether a transcript matures, remains nuclear, or is eliminated. U1 snRNP can suppress premature cleavage and polyadenylation, export factors such as THO/TREX can influence transcription termination in plants, and cap-binding complex-associated ARS2 has been implicated in 3′ end maturation of several RNA families.

The biological consequences of 3′ end decisions are broad. APA changes the regulatory information carried by an mRNA. Tail length and tail composition affect translation and decay, especially through poly(A)-binding proteins, deadenylases, terminal uridyltransferases, and decay machineries. Therapeutic mRNAs exploit this biology: the encoded protein, untranslated regions, cap, nucleotide modifications, codon design, purification, and poly(A) tail all jointly shape translation output, innate immune activation, and persistence. The local [Chapter 29](chapter1028.md) reference scaffold contains broad RNA polyadenylation and APA sources, but direct therapeutic mRNA poly(A)-tail engineering references should be added before final clinical or manufacturing claims are considered complete.

Replication-dependent histone mRNAs show why “mRNA 3′ end” must not be equated with “poly(A) tail.” Most canonical histone mRNAs made during S phase in metazoan somatic cells are intron-poor or intronless and end shortly after a conserved stem-loop rather than in a long poly(A) tail. SLBP binds the stem-loop; the 5′ end of U7 snRNA base-pairs with a histone downstream element; the U7-specific LSM10-LSM11 ring, FLASH, and a CPSF73-containing cleavage complex assemble a catalytically competent machine. CPSF73 cleaves the pre-mRNA, but no canonical poly(A) polymerase is then engaged. This pathway links histone-protein supply to DNA replication and rapid end-of-S-phase turnover.

Histone 3′ processing is specialized without being isolated from general mRNA metabolism. The pathway reuses CPSF73, CPSF100, symplekin, export adaptors, translation factors, and decay enzymes, while U7 snRNP, FLASH, SLBP, and histone-locus organization provide substrate specificity and cell-cycle regulation. Failure of U7-dependent processing can expose downstream canonical poly(A) signals and generate readthrough or polyadenylated histone transcripts. Conversely, replication-independent histone variants, histone genes outside the canonical metazoan program, and a recently characterized class of Drosophila maternal histone mRNAs can be normally polyadenylated. Polyadenylated histone reads therefore require gene-, organism-, developmental-stage-, and perturbation-specific interpretation.

## Concept Inventory

- **3′ end formation:** the set of processing reactions that create the mature downstream end of an RNA molecule. For most metazoan mRNAs, the key steps are endonucleolytic cleavage and poly(A) tail synthesis.
- **Polyadenylation signal:** an RNA sequence element, often AAUAAA or a variant in metazoan pre-mRNAs, that helps recruit cleavage and polyadenylation factors. The signal contributes to site choice but does not by itself define a mature 3′ end.
- **Cleavage site:** the phosphodiester bond cut during 3′ end formation. The cut creates the upstream RNA end that receives the poly(A) tail and a downstream RNA fragment that is usually degraded.
- **Cleavage and polyadenylation specificity factor:** a multi-protein complex that recognizes upstream polyadenylation signals and includes CPSF73, the endonuclease responsible for pre-mRNA cleavage in canonical metazoan mRNA 3′ processing.
- **Cleavage stimulation factor:** a polyadenylation factor complex that helps recognize downstream sequence elements and promotes efficient cleavage-site use.
- **U1-dependent premature-polyadenylation suppression:** a cotranscriptional function of U1 snRNP, also called telescripting, in which U1 binding suppresses use of premature cleavage and polyadenylation sites and thereby favors productive elongation. This function overlaps with, but is mechanistically distinguishable from, U1's role in splice-site recognition.
- **Poly(A) tail:** a nontemplated adenosine-rich extension added to an RNA 3′ end. In nuclear mRNAs it usually promotes maturation, stability, export, and translation, but adenylation can have different meanings in other RNA classes and organisms.
- **Poly(A) polymerase:** an enzyme that adds adenosine residues to an RNA 3′ end without using a nucleic-acid template. Canonical nuclear poly(A) polymerases act during mRNA maturation, whereas noncanonical enzymes act on selected RNA classes and regulatory contexts.
- **Terminal nucleotidyltransferase:** an enzyme that adds nucleotides to RNA 3′ ends without a template. The term includes enzymes that add adenosines, uridines, or mixed nucleotide tails.
- **Uridylation:** nontemplated addition of one or more uridine residues to an RNA 3′ end. It can promote RNA decay, regulate small RNAs, or mark cryptic and pervasive transcripts.
- **Mixed tail:** a 3′ terminal extension containing more than one nucleotide type, such as adenosines interrupted or extended by uridines. Mixed tails are measured by tail-sensitive sequencing methods and interpreted in relation to RNA decay or regulation.
- **Tail-code model:** proposes that tail length, nucleotide composition, and binding proteins act together as regulatory information at RNA 3′ ends. It is a useful hypothesis but not a complete universal code.
- **Alternative polyadenylation:** the use of alternative cleavage and polyadenylation sites within a gene or transcription unit, producing RNA isoforms with different 3′ ends.
- **Proximal poly(A) site:** a poly(A) site that lies closer to the promoter or coding region than another site in the same transcription unit. Use of a proximal site often shortens the 3′ untranslated region or changes the terminal exon.
- **Distal poly(A) site:** a poly(A) site that lies farther downstream than an alternative site in the same transcription unit. Use of a distal site often preserves more downstream untranslated-region sequence.
- **3′ untranslated region:** the portion of a mature mRNA downstream of the stop codon and upstream of the poly(A) tail. It often contains binding sites for microRNAs, RNA-binding proteins, localization machinery, and decay regulators.
- **Replication-dependent histone mRNA:** a transcript encoding a canonical histone whose abundance is coupled to DNA synthesis. In most studied metazoan somatic cells, the mature transcript terminates after a conserved stem-loop rather than a long poly(A) tail.
- **Histone pre-mRNA stem-loop:** a conserved RNA hairpin upstream of the cleavage site. SLBP binds this structure and supports processing, export, translation, and stability.
- **Histone downstream element:** a purine-rich element downstream of the cleavage site that base-pairs with the 5′ end of U7 snRNA. Its sequence and spacing contribute to U7 recruitment and cleavage-site positioning.
- **Stem-loop binding protein:** the principal protein reader of the histone mRNA 3′ stem-loop, commonly abbreviated SLBP. Its functions are regulated across the cell cycle and differ between processing, export, translation, and decay contexts.
- **U7 small nuclear ribonucleoprotein:** a low-abundance snRNP containing U7 snRNA and an unusual Sm-like ring in which LSM10 and LSM11 replace SmD1 and SmD2. U7 snRNP recognizes the histone downstream element and helps recruit the cleavage machinery.
- **FLASH:** a large protein whose N-terminal region binds the extended N terminus of LSM11 and helps recruit and activate the CPSF73-containing histone cleavage complex.
- **Histone cleavage complex:** the CPSF73-CPSF100-symplekin-centered catalytic module used in replication-dependent histone pre-mRNA cleavage. Its sharing of components with canonical cleavage and polyadenylation does not imply that the histone RNA is polyadenylated.
- **Histone locus body:** a nuclear condensate-like compartment assembled at clustered histone genes, enriched in transcription and 3′-processing machinery. In animals it helps coordinate high histone-output during S phase.

## What to Know Before Reading This Chapter

RNA has polarity. The 5′ end and 3′ end are not interchangeable labels; they describe the orientation of the ribose-phosphate backbone. A newly synthesized RNA polymerase II transcript grows by adding nucleotides to its 3′ end, but the final 3′ end of a mature mRNA is often produced later by cleavage. [Chapter 2](chapter1002.md) explains backbone polarity and phosphodiester chemistry. For this chapter, the important point is that an RNA molecule can be cut internally, and the upstream fragment can then receive a new nontemplated tail.

A pre-mRNA is a newly made RNA that has not completed all processing steps. During maturation, the same pre-mRNA can be capped near its 5′ end, spliced, cleaved at its 3′ end, polyadenylated, assembled with proteins, exported, translated, or degraded. These events are not always strictly sequential. Many occur while RNA polymerase II is still transcribing the gene. [Chapter 25](chapter1024.md) introduces co-transcriptional processing; this chapter focuses on the 3′ end.

The running example is a human protein-coding pre-mRNA with two possible 3′ end sites. The upstream or proximal site lies closer to the coding sequence. The downstream or distal site lies farther into the 3′ untranslated region. If the proximal site is used, the mature mRNA has a shorter 3′ untranslated region. If the distal site is used, the mature mRNA carries additional regulatory sequence. Both mRNAs may encode the same protein, but their localization, translation, and stability can differ because the untranslated region contains regulatory information.

The word "polyadenylation" can refer to a pathway or to a chemical product. In the canonical nuclear mRNA pathway, cleavage and polyadenylation are coupled steps that create a mature mRNA end. In other contexts, polyadenylation can mean tailing of a noncoding RNA, repair or activation of a cytoplasmic mRNA, or addition of adenosines that help decay enzymes degrade an RNA. Replication-dependent histone mRNAs add another boundary: they undergo endonucleolytic 3′ cleavage without subsequent long-tail synthesis. Readers should therefore ask four questions whenever polyadenylation is mentioned: which enzyme acts, which RNA class is the substrate, whether cleavage is followed by tail synthesis, and what proteins or decay pathways read the resulting end.

Finally, 3′ end data are assay-dependent. A short-read RNA-seq dataset may show that one terminal exon is more abundant, but it may not precisely map the cleavage site. A 3′ end sequencing method can map poly(A)-proximal reads, but internal priming at genomic A-rich tracts can create artifacts. A tail-length assay can measure poly(A) length, but it may not capture all terminal uridines or distinguish nuclear precursor tails from cytoplasmic mature mRNA tails. Direct RNA sequencing can preserve native terminal information, but coverage, base-calling, and sample preparation still require controls. [Chapter 127](chapter1158.md) owns the experimental and computational measurement framework for these end, tail, and cleavage assays; [Chapter 5](chapter1005.md) provides general evidence standards, while this chapter retains the underlying 3′-end biology.

## 29.1. Cleavage and polyadenylation signals and factors

Cleavage and polyadenylation signals are the RNA features that guide where a pre-mRNA is cut and where a poly(A) tail is added. In many metazoan mRNAs, the best-known upstream signal is AAUAAA, located roughly 10 to 30 nucleotides before the cleavage site. Related variants such as AUUAAA can function with different efficiencies. Downstream of the cleavage site, GU-rich or U-rich sequence elements often help recruit cleavage stimulation factor. The cleavage site itself is not defined by one universally invariant nucleotide, although local sequence and spacing influence efficiency.

![Figure 29.1. Canonical Cleavage and Polyadenylation Site Architecture](../assets/figures/chapter1028_figure1.png)

**Figure 29.1. Canonical Cleavage and Polyadenylation Site Architecture.** A metazoan pre-mRNA poly(A) site is defined by distributed RNA sequence elements rather than a single motif. The upstream polyadenylation signal, most often AAUAAA or a close variant such as AUUAAA, lies roughly 10 to 30 nucleotides before the cleavage site and recruits the CPSF complex; downstream of the cut, GU-rich or U-rich elements recruit cleavage stimulation factor CstF. After CPSF73-mediated endonucleolytic cleavage, the upstream RNA fragment is polyadenylated and the downstream fragment is degraded, while motif scans, factor CLIP, precise 3′ end mapping, and reporter mutagenesis together provide converging evidence for site identity.

> **Box 29.1. Do Not Treat AAUAAA as a Complete Poly(A) Site**
>
> - AAUAAA is the most common upstream polyadenylation signal in metazoan pre-mRNAs but is a necessary contributor, not a sufficient site definition.
> - Correct spacing of roughly 10 to 30 nucleotides between the signal and the cleavage region is required for efficient processing.
> - Downstream GU-rich or U-rich elements and the local RNA-protein environment jointly determine whether the site is used.
> - Close variant hexamers such as AUUAAA can function, sometimes with lower efficiency.
> - A strong AAUAAA hexamer can remain unused when surrounding sequence context, spacing, or factor availability is unfavorable.
> - Weak or variant signals can become active when local factor concentrations, transcription kinetics, or RNA-binding protein context change, as often seen during alternative polyadenylation.

The main processing machine is a set of protein complexes rather than one single enzyme. CPSF recognizes the upstream signal. In current mechanistic language, CPSF includes subunits that bind the polyadenylation signal, scaffold the complex, and perform cleavage. CPSF73 is the endonuclease subunit that cuts the RNA. CstF recognizes downstream sequence information and helps stimulate cleavage. Additional factors, including CFIm, CFIIm, symplekin, Fip1, and poly(A) polymerase, contribute to site choice, complex assembly, cleavage, and tail synthesis. The names are historical and sometimes confusing: a "specificity factor" or "stimulation factor" is not a single activity but a multi-component module.

A useful causal sequence is as follows. First, RNA polymerase II transcribes through a candidate poly(A) site, exposing the upstream polyadenylation signal, the cleavage region, and downstream elements in the nascent RNA. Second, CPSF and associated factors assemble on the upstream signal and nearby RNA. Third, CstF and cleavage factors engage downstream elements and stabilize a productive processing complex. Fourth, CPSF73 cuts the phosphodiester backbone at the cleavage site. Fifth, poly(A) polymerase begins adding adenosines to the upstream cleavage product, while the downstream RNA fragment is handed to degradation and termination pathways. This stepwise description is simplified because the factors can be recruited co-transcriptionally and can influence one another before the full site has emerged.

**Table 29.1. Cleavage and Polyadenylation Signals and Factors.** Core RNA sequence elements and protein factors that define a metazoan mRNA poly(A) site, their molecular roles, evidence basis, and interpretive caveats.

| Feature or Factor | Molecular Role | RNA Location or Binding Target | Main Evidence Type | Effect of Mutation or Depletion | Caveat |
| --- | --- | --- | --- | --- | --- |
| **AAUAAA** | Upstream polyadenylation signal | ~10–30 nt upstream of cleavage site | Biochemical fractionation, reporter mutation | Loss of efficient cleavage and polyadenylation | Hexamer alone does not define a site; correct spacing and downstream context required |
| **Variant PAS (AUUAAA, others)** | Weaker or tissue-specific upstream signal | Same upstream region as AAUAAA | Motif analysis, reporter and CLIP studies | Reduced but not absent processing efficiency | Activity varies; cell state and factor levels can compensate for weak signal |
| **Cleavage site** | Phosphodiester bond cleaved by CPSF73 | Between upstream signal and downstream element | Precise 3′ end mapping, reporter assay | Shifted or absent mature mRNA end | Not defined by one invariant nucleotide; local sequence and spacing contribute |
| **Downstream GU-rich/U-rich element** | Recruits CstF; stimulates cleavage efficiency | Downstream of cleavage site | CLIP, mutagenesis, biochemical binding | Reduced cleavage at adjacent site | Sequence-degenerate; multiple elements can cooperate |
| **CPSF (complex)** | Recognizes upstream signal; scaffolds cleavage machinery | Upstream polyadenylation signal and flanking RNA | Factor depletion, cross-linking, in vitro reconstitution | Failure of signal recognition and pre-mRNA cleavage | Multi-subunit; individual subunit contributions must be distinguished |
| **CPSF73 (CPSF3)** | Endonuclease; cuts pre-mRNA at cleavage site | Cleavage site region | Active-site mutants, reconstituted cleavage assay | No cleavage; transcriptional readthrough | Catalytically active only within assembled CPSF context |
| **CstF (complex)** | Recognizes downstream elements; stimulates cleavage | GU-rich/U-rich downstream sequence | CLIP, depletion, biochemical reconstitution | Reduced or absent cleavage at many sites | Three-subunit complex; CstF64 directly contacts RNA |
| **CFIm** | Promotes use of UGUA-containing sites; modulates site choice | UGUA motifs upstream of poly(A) site | Depletion, eCLIP, 3′ end sequencing | Shift toward proximal sites upon depletion | Can favor or repress sites depending on local sequence context |
| **CFIIm** | Cleavage factor; stimulates the cleavage complex | Associates with CPSF and CstF | In vitro reconstitution, depletion | Reduced cleavage efficiency | Least well characterized of the major cleavage factors |
| **Symplekin** | Scaffold subunit connecting CPSF and CstF modules | Bridges CPSF and CstF subcomplexes | Co-immunoprecipitation, structural studies | Impaired complex assembly | Not an enzyme; provides structural and regulatory scaffold |
| **Poly(A) polymerase (PAPOLA/PAPOLG)** | Adds adenosine residues to cleaved upstream RNA 3′ end | Upstream cleavage product 3′ hydroxyl | In vitro tailing assay, factor depletion | Absent or very short poly(A) tails | Requires assembled processing complex context for efficient processive addition |
| **PABPN1** | Stimulates polymerase processivity; helps define mature tail length | Emerging poly(A) tail in the nucleus | Depletion, biochemical tail extension assays | Shorter and more heterogeneous poly(A) tails | Nuclear function is distinct from cytoplasmic poly(A)-binding protein roles |

The classic evidence for poly(A)-signal recognition came from biochemical fractionation, RNA binding, and reconstituted processing reactions showing that protein factors specifically interact with AAUAAA-containing substrates and that CstF recognizes downstream elements. Modern evidence adds structures, transcriptome-wide binding maps, perturbation studies, and 3′ end sequencing. A productive site is supported most strongly when motif analysis, factor binding, factor perturbation, precise end mapping, and RNA-product changes point to the same processing event.

![Figure 29.2. From Nascent RNA to Mature mRNP and Termination](../assets/figures/chapter1028_figure2.png)

**Figure 29.2. From Nascent RNA to Mature mRNP and Termination.** Cleavage and polyadenylation are tightly coupled to transcription termination and mRNP maturation rather than being independent post-transcriptional events. As RNA polymerase II transcribes through a poly(A) site, CPSF and CstF assemble on nascent RNA elements; CPSF73 then cleaves the pre-mRNA, separating the future mRNA from the downstream transcript. Poly(A) polymerase and PABPN1 extend the upstream product while downstream RNA degradation and processing-factor rearrangements promote polymerase release. The resulting polyadenylated mRNP is handed to export factors, or, if maturation is defective, routed to nuclear surveillance.

Cleavage and polyadenylation signals are probabilistic. AAUAAA is common and strong, but an AAUAAA hexamer in genomic DNA does not automatically become a mature mRNA end. The motif must be in a usable transcript context, with appropriate spacing, downstream elements, and accessible RNA-protein architecture. Conversely, weak or variant signals can be used efficiently when nearby elements and factor concentrations support them. This point is especially important in APA, where a weak proximal site may become active during proliferation, stress, differentiation, or factor imbalance.

The machinery also acts on more than protein-coding mRNAs. Many noncoding RNA classes have distinct 3′ processing pathways, and some promoter-proximal, enhancer, or pervasive transcripts are rapidly terminated and degraded rather than exported as stable mRNAs. Cap-associated ARS2/SRRT can favor maturation of several RNA families, illustrating that 5′ and 3′ processing decisions communicate across the transcript. The boundary with [Chapter 31](chapter1030.md) is important: not every cleaved or tailed RNA is a successful mRNA. Some processed products are surveillance substrates.

Plant and metazoan systems share the broad logic of signal recognition and cleavage, but factor composition and sequence grammar are not identical. Plant CPSF100 has been linked to poly(A)-site anchoring and transcription termination, and plant stress studies show transcriptome-wide APA remodeling. These examples caution against treating human AAUAAA-centered rules as a universal polyadenylation grammar across eukaryotes.

## 29.2. Replication-dependent histone mRNA 3-prime-end formation, U7 snRNP, and cell-cycle control

Replication-dependent histone mRNAs are the major protein-coding exception to the usual metazoan cleavage-and-polyadenylation pathway. Canonical histones package newly replicated DNA into nucleosomes, so cells must produce large amounts of histone protein during S phase and rapidly reduce that output when DNA synthesis ends or stalls. In most studied metazoan somatic cells, the corresponding replication-dependent histone genes are clustered, lack introns, and produce mature mRNAs without a long poly(A) tail. Instead, each mature transcript ends a few nucleotides after a conserved stem-loop. This architecture is not a minor curiosity: it couples a specialized 3′-end processing reaction to histone-locus transcription, nuclear export, translation, and unusually rapid cell-cycle-regulated decay.

The RNA substrate contains two principal cis-acting elements. The upstream element is a highly conserved stem-loop near the future mature 3′ end. Stem-loop binding protein (SLBP) recognizes the hairpin with high sequence-and-structure specificity. The second element, called the histone downstream element (HDE), lies beyond the cleavage site and is usually purine-rich. The 5′ end of U7 small nuclear RNA (U7 snRNA) base-pairs with the HDE. The distance between the stem-loop, cleavage region, and HDE helps define where cleavage occurs, but the system is not governed by one invariant nucleotide count: HDE complementarity, local sequence, RNA structure, and protein contacts all influence efficiency.

U7 snRNA is short and scarce compared with spliceosomal small nuclear RNAs. It is packaged in an unusual Sm-like protein ring containing SmB/B′, SmD3, SmE, SmF, SmG, LSM10, and LSM11. LSM10 and LSM11 replace the SmD1 and SmD2 proteins found in the canonical spliceosomal Sm core. A specialized survival motor neuron (SMN) assembly pathway helps build this particle. LSM11 also has an extended N-terminal domain that provides a binding platform for FLICE-associated huge protein (FLASH). These differences make U7 snRNP a distinct molecular machine, not a miniature spliceosome.

FLASH bridges substrate recognition to catalysis. The N-terminal region of FLASH binds the LSM11 N-terminal extension, and FLASH helps recruit the histone cleavage complex. The catalytic module includes CPSF73, CPSF100, and symplekin; CstF64 is also present in active human reconstitutions. CPSF73 is therefore used in two pathways: it cleaves canonical pre-mRNAs before poly(A)-tail synthesis, and it cleaves replication-dependent histone pre-mRNAs without subsequent long-tail synthesis. Sharing an endonuclease and scaffold does not make the two outputs equivalent. U7 snRNP, FLASH, SLBP, the HDE, and specialized complex assembly determine the histone pathway's substrate and product.

The reaction can be followed as a causal sequence. First, RNA polymerase II transcribes the histone coding region, stem-loop, cleavage region, and HDE. Second, SLBP binds the stem-loop. Third, the exposed 5′ end of U7 snRNA base-pairs with the HDE, while interactions involving SLBP and FLASH stabilize productive U7 recruitment. Fourth, LSM11-bound FLASH recruits the CPSF73-containing histone cleavage complex. Fifth, formation of the HDE-U7 duplex and protein-protein contacts position the RNA and drive rearrangements that make the CPSF73 active site competent for cleavage. Sixth, CPSF73 cuts the pre-mRNA between the stem-loop and HDE through a metallo-β-lactamase/β-CASP nuclease mechanism. The upstream product becomes the mature nonpolyadenylated histone mRNA; the downstream product is degraded, with CPSF73 itself capable of 5′-to-3′ exonucleolytic processing in reconstituted systems.

Biochemical reconstitution and structural work explain why earlier descriptions of a static “molecular ruler” were useful but incomplete. Moving the HDE downstream shifts cleavage by a corresponding distance, while RNA-contact experiments identified SmB, SmD3, and LSM10 in the U7-specific ring as proteins that rigidify the segment between the cleavage site and U7-binding region. This ruler function should not be assigned to LSM11 merely because LSM11 recruits FLASH: the two roles occupy different parts of the same particle. Reconstitution with 13 proteins and two RNAs established a minimal active human machinery. Cryogenic electron microscopy then captured an active assembly in which the pre-mRNA enters the CPSF73 active site and the U7-HDE duplex organizes the catalytic architecture. These experiments establish direct cleavage competence more strongly than colocalization or RNA-binding data alone. They do not show that every endogenous histone locus uses identical geometry, however: reconstituted substrates are engineered, structures trap selected states, and HDE sequences vary across genes and species.

Histone 3′ processing takes place within a larger gene-expression program. Replication-dependent histone genes are concentrated in histone locus bodies, nuclear compartments that enrich transcription factors, U7 machinery, and processing components near histone-gene clusters. In mammals, nuclear protein ataxia-telangiectasia locus (NPAT) is activated near the G1/S transition downstream of cyclin E-cyclin-dependent kinase 2 signaling and promotes histone-gene transcription. Drosophila uses the NPAT functional counterpart multi sex combs (Mxc) as a central histone-locus-body scaffold. These bodies increase local efficiency and coordination, but microscopy showing two factors in the same body does not prove a direct catalytic interaction.

The mature mRNP retains SLBP at its 3′ stem-loop. SLBP can couple processing to subsequent steps, but its contribution is not identical in every experimental system. In mammalian cells, the export adaptor Aly/REF export factor (ALYREF) and the transcription-export complex (TREX) connect histone 3′ processing with nuclear export, including contacts involving SLBP and LSM11. Earlier Xenopus oocyte experiments found that export of short intronless histone RNAs could depend strongly on RNA length and the TAP/NXF1 pathway without requiring SLBP. These results are not necessarily contradictory: transcript length, organism, developmental state, injected versus endogenous RNA, and mRNP assembly route can change which feature is limiting. The defensible conclusion is that histone mRNAs use general mRNA-export machinery, while SLBP- and U7-linked coupling is important in defined metazoan cellular contexts rather than an unconditional export rule.

In the cytoplasm, SLBP helps promote efficient histone mRNA translation. SLBP-interacting protein 1 (SLIP1) links the stem-loop-bound mRNP to translation-initiation machinery, allowing a nonpolyadenylated transcript to recruit productive ribosomes without a conventional poly(A)-binding-protein bridge. Translation and decay are also coupled: experiments in mammalian cells show that efficient destruction after DNA-replication inhibition depends on translation termination near the stem-loop. The 3′ stem-loop is therefore simultaneously an end-processing signal, an mRNP assembly site, a translation-control element, and a decay-control element.

Cell-cycle control operates at several levels rather than through one binary switch. At the G1/S transition, histone-locus transcription increases, the processing machinery becomes productive, SLBP accumulates, and histone mRNA abundance rises to support chromatin assembly behind replication forks. Near the end of S phase, transcription and processing decline, histone mRNAs are destabilized, and SLBP is degraded through a pathway involving cyclin A/CDK1-dependent phosphorylation. Acute inhibition of DNA replication triggers histone mRNA decay even before a normal S phase has ended. The outcome is a rapid reduction of histone-protein synthesis when new DNA is no longer available, limiting toxic excess of free histones.

Histone mRNA decay involves 3′-end remodeling, decapping, and degradation from both directions. A classic model placed terminal oligouridylation upstream of LSM1-associated decapping, 5′-to-3′ exonucleolysis, and exosome-linked 3′-to-5′ decay. TUT7, also called TENT3B or ZCCHC6, and the 3′ exonuclease 3′hExo/ERI1 remodel the stem-loop end. Genetic knockout experiments indicate cooperation and partial redundancy: cells lacking either factor can retain substantial histone mRNA regulation, whereas combined loss reveals stronger defects. Thus, oligouridylation is an important route and biochemical marker, but it should not be taught as one obligatory, irreversible trigger operating identically on every histone transcript.

The nonpolyadenylated architecture has biologically important boundaries. Replication-independent histone variants, such as many H3.3 transcripts, generally use conventional splicing and/or polyadenylation rather than the canonical replication-dependent end. Across eukaryotes, histone-gene architectures also vary. Even within a replication-dependent locus, a cryptic or downstream canonical poly(A) signal can become visible when U7, SLBP, FLASH, or cleavage activity is impaired. Human histone pre-mRNAs can contain overlapping sequence information capable of assembling either histone-specific or canonical mRNA 3′-processing complexes, creating direct competition between end pathways.

Development adds a still sharper exception. Drosophila maternal histone mRNAs deposited into the egg can be polyadenylated through an SLBP-dependent but U7-independent pathway, and their stem-loop region is developmentally remodeled. This finding means that “canonical histone mRNA” does not specify one immutable end architecture across every life stage. A polyadenylated histone transcript may represent a normal histone variant, an organism-specific or developmental program, or defective replication-dependent processing. Gene identity, tissue, developmental stage, end sequence, HDE use, and perturbation must be resolved before assigning mechanism or pathology.

Genetic and disease studies reveal pathway requirements but require restraint in causal interpretation. Drosophila U7 snRNA mutations block normal processing, produce readthrough and polyadenylated histone RNAs, and disrupt oogenesis. LSM10 or LSM11 mutants produce an indistinguishable histone-RNA processing defect, yet U7-null animals are viable whereas LSM10- or LSM11-null animals do not survive to adulthood. This phenotypic separation is evidence that LSM10 and LSM11 perform at least one essential function outside U7-dependent histone processing; it argues against assigning their organismal lethality to histone misprocessing alone. SMN deficiency reduces U7 snRNP biogenesis and alters histone mRNA 3′ formation in spinal muscular atrophy models and patient material. Those findings establish a molecular defect associated with SMN loss; they do not establish that histone misprocessing is the principal cause of motor-neuron disease.

CPSF73 inhibition illustrates a second interpretive problem. The prodrug JTE-607 and related perturbations can disrupt core histone mRNA processing, cause downstream readthrough, reduce histone expression, arrest cells in S phase, and inhibit pancreatic-cancer cell proliferation. This provides a plausible histone-processing vulnerability. CPSF73/CPSF3 is also the endonuclease for canonical cleavage and polyadenylation, however, so global transcriptional and mRNA-processing effects accompany histone defects. A cancer phenotype after CPSF3 inhibition cannot be assigned solely to histone mRNAs without substrate-selective rescue or other separation-of-function evidence.

Different methods answer different questions about this pathway. S1 nuclease mapping, RNase-protection assays, northern blotting, and carefully designed 3′ rapid amplification of cDNA ends can distinguish correctly cleaved, readthrough, and polyadenylated products. U7 antisense oligonucleotides, cis-element mutations, factor depletion, rescue, immunodepletion, and defined biochemical reconstitution test causal requirements. Cryogenic electron microscopy localizes substrates and factors in selected reaction states. Nascent-RNA assays and histone-locus chromatin measurements connect processing with transcription, whereas ribosome profiling and half-life measurements test translation and decay. No single readout covers the complete path from transcription to cleavage to protein production.

Sequencing introduces especially consequential artifacts. Poly(A)-selected libraries systematically undercount correctly processed replication-dependent histone mRNAs while enriching the rare or induced polyadenylated products; the resulting ratio can make misprocessing look like increased total histone expression. Internal oligo(dT) priming can create false ends, and the short, clustered, highly similar histone genes produce multimapping reads that are difficult to assign to individual loci. Long-read or direct-RNA sequencing can preserve more end context but still inherits selection, coverage, and alignment biases. Synchronization drugs add another confounder because thymidine, hydroxyurea, or replication inhibitors alter the very DNA-replication stress pathway that controls histone mRNA abundance. Strong studies combine total-RNA and poly(A)-selected measurements, locus-aware end mapping, spike-ins or recovery controls, orthogonal northern or nuclease mapping, and minimally perturbing cell-cycle measurements such as time courses or single-cell phase assignment.

This chapter owns the histone-specific 3′-end mechanism and its boundary with polyadenylation. [Chapter 30](chapter1029.md) develops general mRNP export; [Chapter 35](chapter1033.md) develops eukaryotic mRNA decay; [Chapter 102](chapter1097.md) develops maternal-RNA and developmental timing programs; and [Chapter 127](chapter1158.md) compares RNA-end profiling technologies in depth. Those chapters should use, rather than restate, the U7-SLBP-FLASH-CPSF73 mechanism established here.

The core lesson is a boundary rule: endonucleolytic cleavage is not synonymous with polyadenylation, and an mRNA need not carry a long poly(A) tail to be exported, translated, and regulated. Replication-dependent histone mRNAs use a specialized cis-element and RNP system while sharing selected catalytic and mRNP components with conventional mRNAs. The pathway therefore teaches both specificity and reuse. Any general statement that “eukaryotic mRNAs end in poly(A)” should be qualified by RNA class, organism, and developmental context.

## 29.3. Canonical and noncanonical poly(A) polymerases

A poly(A) polymerase is a template-independent enzyme that adds adenosine residues to an RNA 3′ hydroxyl. Template-independent means that the enzyme does not read a DNA or RNA template to decide the added nucleotide sequence. In canonical nuclear mRNA maturation, poly(A) polymerase adds a long adenosine-rich tail after CPSF73-mediated cleavage. The tail is then bound by nuclear poly(A)-binding protein, which promotes further extension and helps create a mature tail-length distribution.

The canonical pathway should be understood as a coordinated reaction rather than a naked enzyme adding adenosines indefinitely. After cleavage, poly(A) polymerase begins extension on the newly generated upstream RNA end. PABPN1 binds the emerging tail and stimulates processivity, meaning that the enzyme adds many adenosines during a productive processing event. Tail length is constrained by the processing complex, poly(A)-binding proteins, and later cytoplasmic deadenylation. The mature tail length measured in the cytoplasm is therefore not simply the number of nucleotides first added in the nucleus.

Canonical nuclear poly(A) polymerases include PAPOLA and PAPOLG in mammals. These enzymes act in the context of cleavage and polyadenylation complexes and are best understood as part of mRNA maturation. Their product, the poly(A) tail, supports nuclear export and translation competence partly by recruiting poly(A)-binding proteins. The tail also protects the mRNA from immediate 3′-to-5′ decay. This protective effect is not absolute; deadenylases progressively shorten tails, and short-tailed mRNAs can be decapped, degraded by exonucleases, or uridylated.

> **Box 29.2. Poly(A) Can Stabilize or Destabilize Depending on Context**
>
> - In canonical nuclear mRNA maturation, poly(A) tails added by PAPOLA or PAPOLG recruit PABPN1 and export factors, supporting mRNA stability and translation.
> - In developmental contexts, cytoplasmic polyadenylation by noncanonical enzymes activates stored, short-tailed mRNAs and promotes their translation.
> - In bacteria and organelles, poly(A) tails create single-stranded platforms that help exonucleases engage and degrade structured RNAs, making adenylation a decay-promoting modification.
> - Aberrant, misprocessed, or pervasive nuclear transcripts can be adenylated or uridylated as part of nuclear RNA quality-control pathways before degradation.
> - The consequence of any tail addition should be interpreted only after specifying the enzyme, RNA substrate class, cellular compartment, and downstream reader proteins.

Noncanonical poly(A) polymerases broaden the picture. The term refers to enzymes that add adenosines outside the main nuclear pre-mRNA cleavage-and-polyadenylation reaction. Some act in the cytoplasm to extend selected mRNA tails during developmental regulation or neuronal activity. Others modify small RNAs, noncoding RNAs, aberrant RNAs, or organellar RNAs. In many organisms, including bacteria and organelles, polyadenylation can promote degradation by creating a single-stranded platform for exonucleases rather than stabilizing the RNA. The chemistry is similar, but the biological meaning is different.

This chapter retains the enzymology boundary and the contrast with nuclear maturation, but it does not own the regulatory mechanism by which cytoplasmic poly(A) tails switch selected mRNAs between repression and translation. [Chapter 72](chapter1067.md) develops the CPEB-cis-element, deadenylase, noncanonical polymerase, poly(A)-binding-protein, and initiation-factor sequence of events; [Chapter 102](chapter1097.md) applies that mechanism to maternal RNAs and developmental timing. Nuclear poly(A)-site choice and cytoplasmic tail-length remodeling are therefore related but distinct decisions.

The family-level terminology is a source of confusion. Many noncanonical tailing enzymes are now discussed as terminal nucleotidyltransferases, often abbreviated TENT enzymes in metazoan literature. Some members preferentially add adenosines, some add uridines, and some can add different nucleotides depending on substrate and cofactor context. Thus, "poly(A) polymerase-like" does not always mean that the enzyme makes a long stabilizing poly(A) tail on an mRNA. A conceptually safer phrase is "3′ terminal nucleotidyltransferase" until the enzyme, substrate, and product are specified.

**Table 29.2. Canonical and Noncanonical Tailing Enzymes.** Representative tailing enzymes and enzyme classes, their preferred nucleotide additions, substrates, compartments, and biological consequences.

| Enzyme or Enzyme Class | Preferred Nucleotide Addition | Typical Substrate Class | Compartment | Common Biological Consequence | Evidence Needed for Substrate-Specific Claims |
| --- | --- | --- | --- | --- | --- |
| **PAPOLA / PAPOLG** | Adenosine | Cleaved pre-mRNA upstream fragment | Nucleus | Poly(A) tail promoting mRNA maturation, export, and translation | Enzyme depletion, tail-length measurement, and mRNA output assay |
| **Cytoplasmic poly(A) polymerases** | Adenosine | Stored or developmentally regulated mRNAs | Cytoplasm | Tail extension activating translation during oogenesis or neuronal stimulation | Enzyme identity, substrate identification, tail-length measurement, translation reporter |
| **TENT/TUTase family (e.g., TUT4, TUT7)** | Uridine (primary); adenosine for some members | Short-tailed mRNAs, miRNA precursors, surveillance RNAs | Cytoplasm / nucleus | Uridylation triggering decay or regulating small-RNA loading | Substrate specificity assay, tail-composition sequencing, factor knockout or knockdown |
| **Organellar poly(A) polymerases** | Adenosine | Mitochondrial or chloroplast mRNAs and rRNAs | Mitochondria / chloroplast | Single-stranded tail extension facilitating exonucleolytic decay | Organelle-specific depletion, tail measurement in isolated organelles |
| **Bacterial polyadenylation enzymes (PAP I)** | Adenosine | Bacterial mRNAs and structured RNA decay intermediates | Bacterial cytoplasm | Decay-promoting tail exposing RNA ends to exonucleases | In vitro addition assay, exonuclease sensitivity test, mutant stability measurement |
| **Surveillance-associated tailing enzymes** | Adenosine or uridine (context-dependent) | Cryptic, pervasive, or misprocessed nuclear RNAs | Nucleus | Marking aberrant transcripts for elimination by nuclear decay machineries | Substrate identification, coupling to nuclear surveillance complexes (e.g., NEXT, TRAMP) |

Evidence for canonical polymerase function often combines biochemical reconstitution, factor depletion, tail-length measurement, and mRNA-output assays. Evidence for noncanonical polymerases is more substrate-specific. A strong claim that a noncanonical enzyme stabilizes a transcript should show enzyme-dependent tail addition, altered tail length or composition, and a functional effect on RNA abundance, localization, translation, or decay. A tailing enzyme bound near an RNA is not by itself enough to prove tailing or functional consequence.

Poly(A)-tail profiling methods have become central to this field. Reviews of tail-measurement methods emphasize that different assays measure different features: average tail length, molecule-level tail length, terminal nucleotide composition, or full-length isoform context. These distinctions matter because a change in average tail length can reflect a mixture of isoforms, a shift in RNA abundance, selective loss of short-tailed RNAs during library construction, or true regulation of a specific mRNA pool.

The major boundary case is that polyadenylation can be both maturation and quality control. Nuclear mRNA polyadenylation usually supports productive gene expression. By contrast, polyadenylation of some defective, processed, organellar, bacterial, or surveillance-targeted RNAs can enhance decay. A reader should not memorize "poly(A) equals stable." The better rule is that a poly(A) tail creates a binding and processing platform whose interpretation depends on the RNP environment.

## 29.4. Uridylation, adenylation, mixed tails, and tail-code models

Uridylation is the addition of one or more uridine residues to an RNA 3′ end without a template. Enzymes that catalyze uridylation are often called terminal uridylyltransferases or TUTases, although nomenclature differs across organisms and enzyme families. Uridylation is common in small-RNA biology, mRNA decay, histone mRNA turnover, and surveillance of defective or pervasive transcripts. It is not the opposite of polyadenylation in a simple stabilizing-versus-destabilizing binary, but many well-studied mRNA contexts connect uridylation to decay.

One causal model begins with deadenylation. A cytoplasmic mRNA gradually loses adenosines from its poly(A) tail through deadenylase activity. Once the tail becomes short enough, uridyltransferases can add terminal uridines. These uridines can recruit or stimulate decay machinery, including exonucleases or decapping-linked pathways. In this model, uridylation marks a late stage in mRNA lifetime. The model is useful but incomplete because uridylation also regulates RNAs that do not follow the same deadenylation-first path.

Adenylation and uridylation also regulate small RNAs. In the small-RNA chapters, miRNAs and other guide RNAs are discussed as substrates for tailing and trimming. Tail addition can destabilize a small RNA, change Argonaute association, or reflect target-directed turnover. That material belongs mainly in Chapters [84](chapter1079.md) and [85](chapter1080.md), but it is important here because tailing enzymes do not respect the textbook boundary between "mRNA enzymes" and "small-RNA enzymes." The same enzyme family can participate in multiple RNA pathways.

Mixed tails contain more than one nucleotide type. A short-tailed mRNA might have a poly(A) segment followed by uridines, or a tail with interspersed non-A residues detected by high-resolution tail sequencing. Mixed-tail observations motivate "tail-code" models. A tail-code model proposes that tail length, nucleotide composition, and tail-binding proteins together encode regulatory information about translation, storage, localization, or decay. The analogy to a code is attractive because it reminds readers that a tail is more than length; composition and readers matter.

![Figure 29.3. Tailing Enzyme Outcomes Across RNA Classes](../assets/figures/chapter1028_figure3.png)

**Figure 29.3. Tailing Enzyme Outcomes Across RNA Classes.** The biological meaning of 3′ terminal nucleotide addition depends on the enzyme, substrate, and cellular compartment, not on the nucleotide identity alone. Canonical nuclear poly(A) polymerases such as PAPOLA add stabilizing adenosine-rich tails to cleaved pre-mRNAs, whereas cytoplasmic or noncanonical adenylation can activate stored mRNAs in developmental contexts; bacterial and organellar polyadenylation instead creates single-stranded extensions that promote exonucleolytic decay. Uridylation of short-tailed or surveillance-targeted RNAs by TUTase enzymes can trigger decapping or degradation, and mixed tails containing both adenosines and uridines—detected by specialized tail-sequencing methods—reflect intermediate or composite regulatory states.

The evidence for tail-code models should be framed carefully. It is established that particular tailing enzymes add particular terminal nucleotides to particular RNA substrates, and that these modifications can change decay or translation. It is also established that tail-length distributions correlate with translation and stability in many systems. What remains less settled is whether there is a universal combinatorial code that can be read across all RNA classes, organisms, and cell states. In this chapter, "tail code" is treated as a useful organizing model, not a fully solved grammar.

Recent work on products of pervasive transcription termination highlights the surveillance side of tailing. Pervasive transcription produces many short, unstable, or cryptic RNAs. Tailing at their 3′ ends can help cells distinguish or eliminate unwanted products of termination, protecting transcriptome integrity. The principle is that tailing is part of RNA quality control, not only part of mRNA maturation.

Assay limitations are especially important for terminal nucleotide claims. Standard RNA-seq often loses or masks exact RNA ends. Oligo(dT)-primed libraries enrich polyadenylated molecules but can miss nonadenylated or uridylated substrates. Internal priming can create false 3′ ends at genomic A-rich sequences. Some tail-length assays infer tail length from mobility or signal rather than reading every nucleotide. Direct RNA sequencing and specialized tail-sequencing methods improve resolution, but they still require spike-ins, enzyme controls, and orthogonal validation.

A common misconception is that uridylation always means immediate degradation. Uridylation can promote decay in many settings, but the consequence depends on RNA class, enzyme, tail length, cellular compartment, and reader proteins. Another misconception is that adenylation always means productive mRNA maturation. In bacteria and organelles, or on aberrant nuclear RNAs, adenylation can help exonucleases engage the RNA. Tail identity is meaningful only in context.

## 29.5. Alternative polyadenylation and transcriptome remodeling

Alternative polyadenylation, or APA, is the use of more than one cleavage and polyadenylation site by a gene or transcription unit. APA changes the 3′ end of the RNA. The most familiar form, 3′ UTR APA, changes untranslated-region length while leaving the protein-coding sequence unchanged. Other forms change the terminal exon or coding sequence, producing proteins with different C-terminal regions or transcripts with different noncoding fates. APA is therefore an isoform-generation mechanism, not merely a tail-length change.

![Figure 29.4. Alternative Polyadenylation Mechanisms and Consequences](../assets/figures/chapter1028_figure4.png)

**Figure 29.4. Alternative Polyadenylation Mechanisms and Consequences.** Alternative polyadenylation (APA) allows one gene to produce RNA isoforms with different 3′ ends, with distinct regulatory and sometimes coding consequences. Use of a proximal poly(A) site shortens the 3′ UTR and can remove microRNA target sites, RNA-binding protein motifs, localization elements, or AU-rich decay signals; use of the distal site preserves these regulatory sequences. APA also operates through alternative terminal exon selection coupled to splicing, potentially changing the protein C-terminus, and the functional outcome—altered protein output, changed mRNA localization, or no detectable effect—depends on which regulatory elements are gained or lost in the specific cell context.

The simplest APA example uses two sites. A proximal poly(A) site is closer to the coding region. A distal poly(A) site lies farther downstream. If the proximal site is used, the mature mRNA has a shorter 3′ UTR. If the distal site is used, the mature mRNA includes additional downstream sequence. That downstream sequence may contain microRNA target sites, AU-rich elements, RNA-binding protein motifs, localization signals, secondary-structure elements, or sites that influence translation. The encoded protein can be identical, while post-transcriptional regulation differs.

**Table 29.3. Alternative Polyadenylation Classes.** Types of alternative polyadenylation (APA), the RNA structural changes each produces, protein-coding consequences, regulatory elements affected, detection methods, and key interpretive caveats.

| APA Class | RNA Structure Changed | Protein Product Changed? | Regulatory Elements Affected | Common Assay | Major Interpretation Caveat |
| --- | --- | --- | --- | --- | --- |
| **Tandem 3′ UTR APA** | 3′ UTR length only | No | miRNA sites, RBP motifs, AU-rich elements, localization signals | 3′ end sequencing, UTR reporter assay | Internal priming at A-rich genomic sequences can create false proximal sites |
| **Alternative terminal exon APA** | Terminal exon identity and 3′ UTR sequence | Sometimes (different C-terminus encoded) | Terminal-exon-specific regulatory and sometimes coding sequences | 3′ end sequencing combined with RNA-seq splice-junction reads | Requires joint analysis of splicing and polyadenylation; confounded by isoform-abundance changes |
| **Intronic polyadenylation** | Transcript truncated within an intron | Yes (loss of downstream exons and protein domains) | All downstream coding and regulatory sequences | 3′ end sequencing, RT-PCR for truncated isoform | Truncated isoforms may be rare or unstable; must distinguish from degradation intermediates |
| **Coding-region-changing APA** | Poly(A) site within or upstream of a terminal coding exon | Yes (altered or truncated protein C-terminus) | Downstream coding exons and 3′ regulatory regions | Long-read sequencing, targeted proteomics | Functional protein isoform requires validation by mass spectrometry or biochemical assay |
| **Promoter-proximal or inter-promoter premature termination** | Short transcript ending near a transcription start site or between alternative promoters | No productive full-length upstream transcript; downstream-promoter output can change in defined gene architectures | Downstream gene body and, in multi-promoter genes, chromatin traversed before a downstream promoter | 3′ end sequencing plus TT-seq/PRO-seq, chromatin accessibility, and U1 or site-specific perturbation | U1 inhibition also affects splicing; a downstream-promoter effect requires a mapped intervening premature site and rescue or site-specific suppression |
| **Readthrough-associated distal end shifts** | Extended 3′ end beyond annotated gene termination | No change in main ORF | Sequences downstream of normal poly(A) site; potential antisense overlap with neighboring gene | Long-read RNA-seq, nascent RNA assay | Termination defect versus altered isoform stability must be distinguished |

Mechanistically, APA reflects competition among poly(A) sites and regulatory inputs. A strong proximal site may be used when cleavage and polyadenylation factors are abundant or when RNA polymerase II elongation kinetics favor early recognition. A weak proximal site may be ignored when distal-site recognition is favored by local sequence context, RNA-binding proteins, chromatin context, or coupling to splicing. CFIm and other processing factors can promote or repress particular site classes. Tissue-specific RNA-binding proteins can also bias site use by binding near candidate sites or by affecting linked splicing decisions.

APA and alternative splicing overlap. A terminal exon can be chosen by splicing to a downstream exon that contains a poly(A) site, or by cleavage at an intronic poly(A) site that prevents downstream exon inclusion. A change in cleavage factor abundance can therefore look like a splicing change, and a splicing regulator can change apparent poly(A)-site use. [Chapter 28](chapter1027.md) treats alternative splicing in detail; this chapter emphasizes that terminal exon choice is often a joint splicing-polyadenylation decision. CPSF1 regulation of IL7R exon 6 alternative splicing is one example of processing crosstalk.

APA remodels transcriptomes during proliferation, differentiation, stress, and disease. Reviews synthesize a broad pattern in which proliferating cells and some cancers often show 3′ UTR shortening, whereas differentiated cells such as neurons frequently express long 3′ UTR isoforms with rich regulatory information. A whole-transcriptome termini survey in *Xenopus tropicalis* associated widespread APA changes with the maternal-to-zygotic transition, sex, and growth from embryo to adult, providing a developmental atlas rather than direct evidence that each site shift causes the associated phenotype. Stress responses can shift APA in transcript- and pathway-specific ways rather than through one universal direction. Plant abiotic stress studies show that APA-mediated remodeling is not restricted to animals.

Neurons provide a concrete example of why APA matters. Neuronal mRNAs often have long 3′ UTRs that carry localization and translational control information. A distal poly(A) site can allow an mRNA to include elements needed for transport into dendrites or axons, whereas proximal-site use may remove those elements. APA therefore contributes to spatial RNA regulation, local translation, and cell-type identity. The relationship is not deterministic; a long 3′ UTR can carry stabilizing and destabilizing elements, and the same site can behave differently in different neuronal cell types.

Disease studies increasingly use single-cell or single-nucleus approaches to connect APA with cell states. A 2024 multiplexed single-cell study characterized APA regulators across many perturbations, showing that APA factor effects can be measured at cellular resolution. ALS-related single-nucleus atlas work links APA modeling to disease-associated brain cell states, although mechanistic interpretation requires caution because single-nucleus transcriptomes measure accumulated RNA products and computationally inferred site usage rather than direct cleavage reactions. Kidney disease and lymphoma examples in the local reference scaffold suggest additional disease contexts, but broad clinical generalization requires targeted review.

APA evidence must distinguish site use from abundance. If a short 3′ UTR isoform increases in RNA-seq, the cause might be increased proximal cleavage, increased stability of the short isoform, decreased stability of the long isoform, altered splicing, or cell-composition change. Strong APA inference uses methods that map 3′ ends directly, controls internal priming, quantifies isoform abundance, and tests factor dependence. Functional claims require perturbing the site or regulatory element and measuring translation, localization, decay, or phenotype.

A common overgeneralization is that 3′ UTR shortening always causes oncogene activation by removing microRNA sites. This mechanism can occur, but APA effects are more diverse. Shortening can remove destabilizing motifs, stabilizing motifs, localization elements, or translational repressors. It can also have little effect if the relevant regulators are absent in that cell type. Distal-site use can add regulatory complexity but does not automatically reduce protein output. APA is a transcript- and context-specific remodeling mechanism.

**Table 29.4. Methods for 3′ End and Tail Analysis.** Experimental approaches for mapping poly(A) sites and characterizing RNA tail properties, with their direct outputs, strongest inferences, common artifacts, validation strategies, and suitability for single-cell experiments.

| Method | Direct Output | Strongest Inference | Common Artifact | Best Orthogonal Validation | Suitable for Single-Cell Use? |
| --- | --- | --- | --- | --- | --- |
| **3′ end sequencing (e.g., 3′ READS, PolyA-seq)** | Positions of poly(A)-proximal read pile-ups | Poly(A)-site identity and relative site usage | Internal priming at genomic A-rich tracts | Reporter mutation or CRISPR deletion of candidate site | Limited; specialized protocols exist but with low coverage |
| **RNA-seq (standard)** | Transcript-level and exon-level read counts | Terminal exon abundance differences | 3′ bias in many protocols; does not resolve exact cleavage position | 3′ end sequencing for site-position confirmation | Yes, widely used in single-cell workflows |
| **Long-read cDNA sequencing (PacBio, ONT cDNA)** | Full-length isoform sequences including 3′ end context | Isoform-level structure linking splice junctions to poly(A) sites | Variable 3′ end truncation during library preparation | 3′ end sequencing for cleavage-site accuracy | Emerging single-cell long-read protocols available |
| **Direct RNA sequencing (Oxford Nanopore)** | Native RNA sequence with poly(A) tail and terminal nucleotides | Native tail length and composition without PCR or reverse-transcription bias | Coverage gaps; base-calling errors at homopolymer runs | Spike-in controls; comparison with biochemical tail assay | Not yet routine at single-cell scale |
| **Poly(A)-tail profiling (PAL-seq, TAIL-seq, Nanopore tail estimation)** | Molecule-level poly(A) tail length distributions | Distribution of tail lengths per transcript | Short tails under-sampled; degradation fragments inflate short-tail fraction | Internal length standards; cross-method comparison | Not standard; requires relatively high RNA input |
| **Terminal nucleotide profiling (TAIL-seq-style)** | Terminal nucleotide composition and mixed-tail patterns | Uridylation frequency and mixed-tail prevalence per transcript class | Ligation bias favoring certain 3′ terminal sequences | Enzyme controls (TUTase knockout); spike-in standards | Not yet in standard single-cell workflows |
| **CLIP (crosslinking immunoprecipitation)** | RNA sequences bound by a specific protein at or near its binding site | Binding-site map for one factor in the context of 3′ end elements | Background from abundant RNAs; antibody cross-reactivity | Multiple antibodies; eCLIP input controls; functional mutagenesis | Adapted single-cell CLIP protocols remain experimental |
| **Reporter assays** | Translation, stability, or localization of a defined model transcript | Cis-element contribution of a specific 3′ UTR or poly(A)-site sequence | May miss endogenous chromatin and isoform context | Endogenous locus perturbation (CRISPR); multiple construct variants | No |
| **Factor depletion (siRNA, auxin degron, CRISPRi)** | RNA-level and isoform-level changes after loss of a target protein | Factor necessity for a specific poly(A)-site or isoform choice | Indirect effects through altered cell growth or RNA stability | Rescue with wild-type or catalytic-dead factor | Yes, applicable in single-cell perturbation screens |
| **Nascent RNA assays (GRO-seq, PRO-seq, TT-seq)** | Sites of active transcription, readthrough extent, and termination position | Transcription rate and termination site in relation to poly(A) signals | Contamination from processed or stable RNA; protocol-specific background | Comparison with RNA polymerase ChIP; run-on controls | Single-cell nascent RNA methods exist but have limited sensitivity |
| **Ribosome profiling** | Ribosome-protected mRNA fragment positions per codon | Relative translational efficiency across transcripts or 3′ UTR isoforms | Does not measure protein output directly; stalled ribosomes complicate interpretation | Quantitative proteomics; translation reporter assays | Single-cell ribosome profiling remains very low throughput |

## 29.6. Coupling to transcription termination and export

3′ end formation is physically linked to transcription termination. RNA polymerase II often transcribes beyond the mature mRNA 3′ end. After cleavage, the upstream RNA becomes the mRNA precursor that receives the poly(A) tail, while the downstream RNA remains associated with the elongating polymerase or is released and degraded. Termination models include exonuclease-mediated degradation of the downstream RNA, sometimes called a torpedo-like mechanism, and allosteric changes in the elongation complex after processing-factor engagement. These models are not mutually exclusive.

The causal sequence can be described in cellular terms. Processing factors accumulate near the poly(A) site as the nascent RNA emerges. Cleavage separates the future mRNA from downstream transcript. The downstream RNA exposes a 5′ end that can be degraded by exonucleases, and degradation or processing-factor rearrangements can promote polymerase release. At the same time, the upstream RNA is polyadenylated and packaged into an mRNP competent for nuclear quality control and export. Thus, one cleavage event helps define both the end of the RNA product and the end of the transcription cycle.

The U1 relay model emphasizes coordination among early transcript processing, poly(A)-site suppression or licensing, and mRNA maturation. U1 snRNP is classically a splicing factor, but U1-dependent suppression of premature cleavage and polyadenylation, often called telescripting, is also a form of transcriptome boundary control. In human genes with a premature poly(A) site between tandem promoters, antisense inhibition of U1 activated premature processing and reduced downstream-promoter output; restoring U1 activity or directly inhibiting the premature site rescued that output. Reduced chromatin accessibility, RNA polymerase II Ser5 phosphorylation, and promoter escape accompanied the downstream-promoter effect, supporting a model in which productive upstream elongation can influence a downstream promoter. This result establishes a defined multi-promoter mechanism, not a rule that every U1-sensitive poly(A) site controls a neighboring promoter. The broader pedagogical point is that splicing factors, cap-associated factors, and polyadenylation factors communicate; a transcript is not processed by isolated stations that ignore one another.

Export is also coupled to 3′ end processing. Mature mRNPs carry proteins that signal completion of capping, splicing or intron status, 3′ end formation, and surveillance passage. Defective 3′ end formation can retain transcripts in the nucleus or route them to degradation. Conversely, export and transcription complexes can feed back on termination. In plants, THO/TREX contributes to transcription termination, showing that export-associated complexes can influence where transcription ends.

Cap-binding complex and ARS2 connect the 5′ end to 3′ maturation for multiple RNA families. This cross-end coordination is important for short transcripts, promoter-proximal RNAs, and noncoding RNA families where the physical distance between the cap and processing site is small. It also illustrates a general mRNP principle developed in [Chapter 30](chapter1029.md): RNA processing creates a protein-decorated particle whose identity depends on the history of processing events.

Boundary cases include transcripts that terminate but are not exported, transcripts that are cleaved but rapidly degraded, and readthrough transcripts that continue beyond annotated gene ends. Pervasive transcription produces many such RNAs, and tailing pathways can help safeguard cells from inappropriate products of termination. Long-read and nascent RNA methods are especially useful for detecting readthrough and termination defects, but interpretation must separate transcriptional readthrough from downstream RNA stability changes.

APA affects termination distance. A proximal poly(A) site can shorten not only the mature mRNA but also the transcribed region needed before termination. A distal site can extend transcription and expose additional chromatin or regulatory regions to polymerase passage. In compact genomes or gene-dense regions, failed or shifted termination can influence downstream genes or antisense transcription. These genome-architecture effects are developed further in Chapters [15](chapter1014.md), [24](chapter1023.md), and [93](chapter1088.md).

The main misconception is that polyadenylation happens after transcription is over. In most nuclear mRNA cases, recognition of the poly(A) site, cleavage, tailing, and termination are co-transcriptional or tightly linked to elongation. The mature mRNA end is created before the polymerase necessarily stops. The end of the RNA product and the end of transcription are related but not identical events.

## 29.7. Effects on translation, decay, and therapeutic RNA design

The mature mRNA 3′ end influences translation and decay because it recruits proteins and defines regulatory sequence content. The poly(A) tail binds poly(A)-binding proteins. In many eukaryotic systems, interactions between cap-binding factors at the 5′ end and poly(A)-binding proteins at the 3′ end promote translation initiation and protect the mRNA from decay. This is often drawn as a closed-loop model, but the drawing should not be interpreted as one rigid circular structure present on every translating mRNA. It is a functional interaction network between ends.

**Table 29.5. Therapeutic RNA Design Variables Connected to 3′ Ends.** Key design parameters of synthetic mRNAs that engage 3′ end biology, their mechanistic effects on translation and stability, manufacturing considerations, and measurement approaches.

| Design Variable | Mechanistic Effect | Manufacturing Concern | Measurement Method |
| --- | --- | --- | --- |
| **Encoded poly(A) tail (DNA template-encoded)** | Defined length at transcription; supports PABP binding and translation initiation | Homogeneity limited by run-off transcription; long A-tracts can cause premature termination | Gel electrophoresis, Nanopore sequencing, analytical HPLC |
| **Enzymatically added poly(A) tail** | Length tunable post-transcriptionally; can achieve longer or more uniform tails | Batch-to-batch length variation; additional enzymatic step increases cost and process complexity | Gel electrophoresis, PAL-seq or Nanopore tail-length estimation |
| **Tail length distribution** | Longer tails increase PABP occupancy and initial translation; very long tails may be trimmed in cells | Homogeneity and reproducibility across production lots must be controlled | Nanopore direct RNA sequencing; analytical HPLC size profiling |
| **UTR selection** | 3′ UTR sequence determines mRNA stability, translation efficiency, miRNA susceptibility, and localization | Must avoid cryptic splice sites, internal poly(A) signals, or destabilizing elements in production host cell | Reporter assay in target cell type; RNA half-life measurement |
| **Cap chemistry (cap analog, ARCA, cotranscriptional cap)** | Affects translation initiation efficiency and innate immune recognition | Cap incorporation efficiency and identity must be confirmed; some analogs reduce transcription yield | Mass spectrometry; cap-specific gel or HPLC analysis |
| **Nucleotide modification (e.g., N1-methylpseudouridine)** | Reduces TLR-mediated innate immune sensing; can improve translation and RNA stability | Modified NTPs must be incorporated efficiently and uniformly; residual unmodified bases may trigger immune response | Enzymatic digestion plus mass spectrometry; HPLC nucleoside analysis |
| **Purification of dsRNA contaminants** | Removal of immunostimulatory double-stranded RNA byproducts reduces innate immune activation in recipients | Chromatographic purification steps add cost and must be monitored in-process | ELISA for dsRNA; HPLC purity profile |
| **Storage stability** | RNA integrity and tail length influence protein expression and immunogenicity on the shelf | Cold-chain requirements; formulation must protect RNA from hydrolysis and nuclease degradation during storage | Capillary electrophoresis or integrity gel; tail-length monitoring at defined storage time points |

Deadenylation is a common first step in mRNA decay. Deadenylase complexes shorten the poly(A) tail, weakening poly(A)-binding protein protection and making the RNA more susceptible to decapping, 5′-to-3′ decay, 3′-to-5′ decay, or terminal uridylation. [Chapter 35](chapter1033.md) covers eukaryotic mRNA decay in depth. The connection for this chapter is that the tail produced during mRNA maturation is later remodeled by decay machinery. Tail length is dynamic, not a fixed birthmark.

![Figure 29.5. Dynamic Tail Life Cycle: Translation, Deadenylation, Uridylation, Decay](../assets/figures/chapter1028_figure5.png)

**Figure 29.5. Dynamic Tail Life Cycle: Translation, Deadenylation, Uridylation, Decay.** A poly(A) tail is a dynamic feature that is remodeled throughout mRNA life rather than a fixed maturation mark. The newly exported mRNA carries a long poly(A) tail bound by poly(A)-binding proteins, which promote cap-poly(A) end interactions that stimulate translation initiation and protect from decay; deadenylase complexes then progressively shorten the tail, weakening this protection. Once the tail becomes short, terminal uridyltransferases can add uridines to form mixed tails that recruit decay machinery and trigger decapping or exonucleolytic degradation. Therapeutic mRNA design exploits these dynamics, with tail length, manufacturing route, UTR sequence, cap analog, and modified nucleotides all interacting to tune translation output and RNA persistence in target cells.

APA influences translation and decay mainly by changing regulatory content. If a proximal APA isoform removes microRNA target sites or destabilizing RNA-binding protein motifs, protein output can rise even if transcription is unchanged. If shortening removes stabilizing motifs or localization elements, protein output can fall or spatial expression can change. If an alternative terminal exon changes the coding sequence, the protein product itself can change. These possibilities explain why APA must be interpreted isoform by isoform rather than by a universal "shorter equals more translated" rule.

Stress illustrates context dependence. Cells exposed to heat, oxidative stress, nutrient limitation, immune activation, or other stressors often remodel RNA processing and decay. APA can shift in response to stress, but different stresses and cell types can favor different site classes. Stress also changes translation initiation, mRNA storage, stress granules, and decay pathways. Therefore, observing APA during stress does not by itself prove a direct adaptive mechanism. Strong claims require perturbing the poly(A) site or processing factor and measuring stress survival, protein output, or pathway function.

Therapeutic RNA design uses the same principles in an engineered setting. A synthetic mRNA intended to express a protein in cells must include a cap or cap analog, untranslated regions, a coding sequence, a poly(A) tail, and chemical or purification features that limit unwanted innate immune activation. The poly(A) tail contributes to translation and stability, but it is only one design parameter. Tail length, tail purity, UTR sequence, codon choice, modified nucleotides, RNA structure, delivery system, and cell type all interact. [Chapter 153](chapter1137.md) covers mRNA therapeutics, and [Chapter 159](chapter1142.md) covers manufacturing and analytical release testing.

For therapeutic mRNAs, the poly(A) tail can be encoded in the DNA template for in vitro transcription, added enzymatically after transcription, or generated by other manufacturing designs. Each route has tradeoffs in length control, homogeneity, cost, regulatory characterization, and compatibility with production scale. The local reference scaffold for [Chapter 29](chapter1028.md) includes general polyadenylation and APA reviews but lacks direct therapeutic mRNA manufacturing sources. Final claims about optimal tail length, specific vaccine platforms, or regulatory release criteria should therefore be treated as final reference item areas rather than as settled by this chapter's current bibliography.

Tail biology also matters for antisense and RNA interference therapeutics indirectly. An antisense oligonucleotide or siRNA can change abundance of an APA regulator, alter a terminal exon, or interact with UTR sequences that differ between APA isoforms. Conversely, an mRNA isoform with a shortened 3′ UTR may lose a therapeutic target site. These issues connect APA to target selection and biomarker interpretation in therapeutic chapters.

The evidence basis for translation and decay effects includes tail-length measurement, ribosome profiling, reporter assays, RNA half-life measurements, protein quantification, and perturbation of specific poly(A) sites. Each method has limits. Ribosome profiling measures protected fragments, not complete protein output. Reporter assays can isolate UTR effects but may miss endogenous chromatin and isoform context. RNA half-life assays can perturb the cell state. Protein changes can reflect translation, decay, localization, or feedback. Strong interpretation combines several assays.

The practical conclusion is that 3′ end formation defines an mRNA regulatory module. The module includes the cleavage site, poly(A) tail, 3′ UTR, terminal exon identity, tail-binding proteins, and decay or translation machinery. Biology and engineering both depend on that module. Changing a poly(A) site or tail can be as consequential as changing a promoter, splice site, or coding sequence.

## Experimental Foundations and Evidence Standards

Evidence for 3′ end formation falls into several categories. Motif evidence identifies candidate polyadenylation signals, downstream elements, and site strength. End-mapping evidence identifies where RNA molecules end. Factor evidence identifies proteins bound to nearby RNA or chromatin. Perturbation evidence tests whether changing a factor or motif changes site use. Functional evidence tests whether an altered 3′ end changes stability, localization, translation, or phenotype. A complete mechanistic claim usually needs more than one category.

3′ end sequencing methods are powerful but artifact-prone. Oligo(dT)-primed methods can internally prime at A-rich genomic regions, creating false poly(A) sites. Size selection can bias against short or degraded RNAs. PCR and read-depth limitations can distort isoform ratios. Direct RNA sequencing can reduce some library-conversion artifacts and measure native tails, but it has its own coverage and base-calling limits. Strong APA studies explicitly filter internal priming, validate selected sites, and distinguish gene-expression changes from isoform-ratio changes.

Tail-length and tail-composition assays require careful interpretation. Tail length varies across an mRNA population, across cellular compartments, and across time after transcription. A measured average can hide multiple isoforms or decay intermediates. Terminal uridines can be missed if a method assumes a pure poly(A) tail. Conversely, degradation products can be overrepresented if the method captures short-tailed RNAs efficiently. The review by Brouze and colleagues is a useful local source for method comparisons.

Single-cell APA measurements add another level of complexity. They reveal cell-to-cell and cell-type variation, but many single-cell protocols are 3′ biased by design and may not capture full isoform structures. A single-cell perturbation screen can identify regulators and candidate pathways, yet direct biochemical mechanisms usually need follow-up in controlled systems. Single-nucleus disease atlases can nominate disease-associated APA programs, but nucleus-specific RNA composition and computational inference should be kept separate from direct measurements of cleavage.

> **Box 29.3. APA Interpretation Checklist**
>
> - Is the poly(A) site mapped directly by 3′ end sequencing, or inferred from standard RNA-seq coverage?
> - Has internal priming at genomic A-rich sequences been explicitly filtered or controlled?
> - Is the observed isoform-ratio change due to altered cleavage-site selection or to differential stability of existing isoforms?
> - Could a linked change in alternative splicing be altering terminal exon choice rather than poly(A) site use?
> - Which specific regulatory elements—microRNA sites, RNA-binding protein motifs, localization signals, or decay elements—are gained or lost between isoforms?
> - Does the change in RNA isoform produce a measurable change in protein level, localization, or cellular phenotype?
> - Could a shift in cell-type composition within the sample explain the apparent APA change?

Replication-dependent histone mRNAs require an assay branch that does not assume a poly(A) tail. Total-RNA end mapping, northern or nuclease-protection analysis, U7 or factor perturbation, and reconstituted cleavage directly test histone-end formation. Poly(A)-selected RNA-seq is useful for detecting abnormal or exceptional polyadenylated histone products only when the depletion of normal nonpolyadenylated products is made explicit. The mechanistic assembly and the corresponding evidence ladder are summarized in the following original visuals.

![Figure 29.6. U7-Directed Assembly and Cleavage of Replication-Dependent Histone Pre-mRNA](../assets/figures/chapter1028_figure6.png)

**Figure 29.6. U7-Directed Assembly and Cleavage of Replication-Dependent Histone Pre-mRNA.** A left-to-right mechanistic sequence contrasts the histone pre-mRNA substrate with its mature product. The substrate contains a coding region, conserved stem-loop, cleavage site, and downstream HDE. SLBP first binds the stem-loop, U7 snRNA base-pairs with the HDE, and the U7-specific Sm-like ring places LSM11-bound FLASH beside the RNA. Label two separable ring functions: LSM11 binds FLASH to recruit a CPSF73-CPSF100-symplekin-centered histone cleavage complex, whereas SmB, SmD3, and LSM10 contact and rigidify the RNA segment used for cleavage-site positioning. Productive U7-HDE pairing and protein contacts position the RNA in the CPSF73 active site. Cleavage releases a mature upstream histone mRNA that retains the SLBP-bound stem-loop but receives no long poly(A) tail; the downstream product is degraded. A small comparison inset shows canonical CPSF73 cleavage followed by poly(A) polymerase, emphasizing that a shared nuclease does not imply a shared final tail.

![Figure 29.7. Replication-Dependent Histone mRNA Across S Phase and Replication Arrest](../assets/figures/chapter1028_figure7.png)

**Figure 29.7. Replication-Dependent Histone mRNA Across S Phase and Replication Arrest.** A circular cell-cycle and mRNP-life-cycle diagram connects nuclear and cytoplasmic steps. At G1/S, cyclin E-CDK2-linked NPAT activation and histone locus bodies promote histone transcription and U7-dependent processing. The SLBP-bound mature mRNP is exported through general mRNA-export machinery and translated with SLIP1-associated initiation factors to supply histones behind replication forks. At normal S-phase completion or after DNA-replication arrest, histone transcription and processing decline, SLBP is destabilized, and the mRNA end is remodeled by 3′hExo/ERI1 and TUT7/TENT3B before decapping and bidirectional degradation. A boundary inset separates normal nonpolyadenylated replication-dependent messages from normally polyadenylated variants or maternal messages and from polyadenylated products caused by processing failure.

**Table 29.6. Replication-Dependent Histone 3′-Processing Evidence and Artifact Controls.** Methods that interrogate histone mRNA 3′ ends differ in what they observe directly. The table separates cleavage-site evidence from accumulated-RNA, localization, structural, and phenotype evidence and makes the major selection and cell-cycle artifacts explicit.

| Method or Perturbation | Direct Observation | Strongest Supported Inference | Major Artifact or Overinterpretation | Essential Control |
| --- | --- | --- | --- | --- |
| **S1 nuclease or RNase-protection mapping** | Protected fragment ending at the expected stem-loop-proximal site | Correct cleavage versus readthrough at a defined locus | Probe cross-hybridization among similar histone genes; incomplete nuclease digestion | Sequence-specific probe, digestion series, size marker, and an end-mutant control |
| **Northern blotting of total RNA** | Size distribution of normal and extended histone transcripts | Product abundance and gross readthrough without poly(A) selection | Limited single-nucleotide resolution; multimember gene-family signal | Multiple probes, loading spike-in, and orthogonal 3′-end assay |
| **3′ RACE or end-focused total-RNA sequencing** | Molecule-level terminal sequence and downstream junction | Cleavage position, downstream readthrough, or use of a cryptic poly(A) site | Internal priming and ligation bias; degradation intermediates can resemble ends | Anchored primers, genomic A-rich filtering, independent northern or nuclease mapping |
| **Poly(A)-selected RNA-seq** | Poly(A)-positive histone reads and downstream sequence | Presence of polyadenylated products | Depletes normal nonpolyadenylated histone mRNA and inflates the apparent poly(A)-positive fraction | Matched total-RNA library and spike-in recovery control; report selection chemistry |
| **U7 antisense inhibition or cis-element mutation** | Loss or shift of correct cleavage and accumulation of readthrough products | Requirement for U7-HDE pairing or stem-loop architecture | Off-target antisense effects or altered RNA structure beyond the intended contact | Compensatory U7-HDE rescue, multiple antisense reagents, and expression control |
| **U7-null versus LSM10/LSM11-null genetic comparison** | Similar histone readthrough and polyadenylation with different viability outcomes | Shared requirement for processing plus an essential LSM10/LSM11 function outside U7-dependent histone processing | Assigning developmental lethality to histone misprocessing because the RNA defect is present | Compare matched molecular defects across genotypes and use separation-of-function rescue |
| **Factor depletion with rescue** | Product changes after loss and restoration of SLBP, LSM11, FLASH, or CPSF73 | Factor necessity and domain or catalytic requirements | Secondary cell-cycle arrest or broad RNA-processing effects | Acute depletion, wild-type and separation-of-function rescue, cell-cycle profiling |
| **Defined biochemical reconstitution** | Cleavage of a purified substrate by known components | Minimal sufficiency and ordered assembly requirements | Engineered substrates, nonphysiological concentrations, missing chromatin context | Omission series, catalytic-dead CPSF73, cis-element mutants, and kinetic measurements |
| **Cryogenic electron microscopy** | Selected conformations and direct spatial relationships | Structural plausibility of U7-HDE positioning and CPSF73 engagement | Static or trapped state; occupancy does not equal pathway flux | Biochemical activity of the imaged preparation and mutation of predicted contacts |
| **Histone locus body microscopy or chromatin mapping** | Factor localization and locus association | Spatial coordination of transcription and processing | Colocalization interpreted as direct interaction or catalysis | Interaction or functional perturbation plus cell-cycle-resolved imaging |
| **Synchronization or replication inhibition** | Histone RNA changes after cell-cycle or replication perturbation | Response to S-phase transition or stalled DNA synthesis | Thymidine, hydroxyurea, or inhibitors directly activate replication stress | Minimally perturbing time course, FUCCI or single-cell phase assignment, multiple perturbations |
| **Long-read or direct-RNA sequencing** | Longer end context, isoform linkage, and sometimes native tail information | Gene-level end architecture and poly(A)/nonpoly(A) boundary | Coverage, pore-entry, selection, homopolymer, and alignment bias in clustered paralogs | Synthetic end standards, locus-aware mapping, matched short-read/end-mapping validation |

## Biological Contexts Across Organisms and RNA Classes

Metazoan mRNAs use cleavage and polyadenylation as a dominant maturation route. The same cells also produce many noncoding RNAs whose 3′ ends are made by other pathways or by specialized versions of related machinery. Replication-dependent histone mRNAs are the major protein-coding boundary case: U7 snRNP directs cleavage after a stem-loop and the mature RNA usually lacks a long poly(A) tail in metazoan somatic S phase. Replication-independent variants, non-metazoan histone architectures, defective processing products, and Drosophila maternal histone mRNAs demonstrate that neither the nonpolyadenylated nor polyadenylated state can be universalized across every histone gene and context.

Plants use cleavage and polyadenylation and show extensive APA, but plant sequence elements, factor paralogs, and physiological contexts differ from animal systems. Plant CPSF100 and THO/TREX references in the local scaffold support the idea that 3′ processing is tied to termination and stress responses in plants.

Bacteria and organelles remind readers that polyadenylation can promote decay rather than stability. Bacterial polyadenylation often creates single-stranded extensions that help exonucleases degrade structured RNAs. Cyanobacterial RNA decay studies belong mainly in bacterial RNA turnover chapters, but they provide a useful contrast to the metazoan mRNA maturation model. Mitochondrial and chloroplast polyadenylation are treated more fully in [Chapter 17](chapter1016.md) and [Chapter 37](chapter1035.md).

Viral RNAs and therapeutic RNAs use or mimic host 3′ end principles in diverse ways. Some viruses encode or template poly(A) tails; others rely on host processing or alternative 3′ structures. Therapeutic mRNAs use engineered poly(A) tails to obtain expression, but delivery, innate immunity, and manufacturing constraints shape the final design. This chapter supplies the mechanistic vocabulary; platform-specific details are deferred to Chapters [153](chapter1137.md), [156](chapter1139.md), and [159](chapter1142.md).

> **Box 29.4. Tail-Code Models: Useful but Not Universal**
>
> - Poly(A) tail length correlates with translation efficiency and mRNA stability in many systems, supporting the view that tail length carries regulatory information.
> - Terminal nucleotide composition—particularly the presence of uridines or mixed adenosine-uridine tails—can influence decay in specific, well-characterized pathways.
> - Tail-bound proteins, including poly(A)-binding proteins, deadenylases, and terminal uridyltransferases, are part of the functional readout; nucleotide sequence alone does not determine fate.
> - Different tail-measurement assays capture different features: average tail length across a population, molecule-level length distributions, or terminal nucleotide identity.
> - A single universal combinatorial code that predicts RNA fate from tail composition across all RNA classes, organisms, and cell states has not been established.
> - Tail-code thinking is most reliable when applied to specific enzymes, defined substrate classes, and controlled cellular conditions rather than as a predictive grammar for all tails.

> **Box 29.5. Therapeutic mRNA Tail Design Requires Platform-Specific Evidence**
>
> - Poly(A) tail length and homogeneity are important quality attributes of therapeutic mRNAs because they contribute to translation efficiency and RNA persistence in target cells.
> - The manufacturing route for tail addition—encoded in the DNA template or added enzymatically after transcription—affects length distribution, lot-to-lot consistency, and regulatory characterization requirements.
> - Tail effects on translation and stability interact with UTR sequence, cap chemistry, nucleotide modifications, codon design, purification of immunostimulatory byproducts, and delivery vehicle; no single parameter can be optimized in isolation.
> - The mechanistic polyadenylation biology in [Chapter 29](chapter1028.md) provides foundational context but does not supply direct clinical or manufacturing evidence for specific therapeutic RNA platforms.
> - Final recommendations for tail length, tail manufacturing strategy, and analytical release criteria require dedicated therapeutic mRNA sources; see Chapters [153](chapter1137.md) and [159](chapter1142.md).

> **Box 29.6. Do Not Universalize Poly(A) Across Histone mRNAs**
>
> - Most replication-dependent histone mRNAs in metazoan somatic S phase end after a conserved stem-loop and do not carry a long poly(A) tail.
> - CPSF73 cleaves both canonical polyadenylated pre-mRNAs and replication-dependent histone pre-mRNAs; enzyme sharing does not determine whether poly(A) polymerase acts next.
> - Replication-independent histone variants commonly use conventional splicing or polyadenylation, and other eukaryotes can use different histone-gene architectures.
> - U7, SLBP, FLASH, or cleavage defects can expose downstream canonical poly(A) signals and produce readthrough or polyadenylated replication-dependent histone RNAs.
> - Drosophila maternal histone mRNAs provide a normal developmental polyadenylation program that is SLBP-dependent but U7-independent.
> - Poly(A)-selected libraries deplete correctly processed replication-dependent histone mRNAs and enrich exceptional polyadenylated products.
> - Classify a polyadenylated histone read only after resolving the gene, organism, developmental stage, 3′-end sequence, library chemistry, and perturbation.

## Recent Consensus

The current consensus is that canonical metazoan mRNA 3′ end formation is a coupled cleavage-and-polyadenylation pathway directed by distributed RNA sequence elements and multi-protein processing complexes. AAUAAA is central but not sufficient. CPSF, CstF, cleavage factors, poly(A) polymerase, and poly(A)-binding proteins act together, and their activity is coupled to transcription, splicing, export, and surveillance.

APA is now considered a major transcriptome-remodeling mechanism rather than a rare annotation nuisance. It changes regulatory sequence content, terminal exon identity, and sometimes coding potential. APA programs are prominent in development, neurons, stress, immune and proliferative states, plants, and disease contexts. The functional consequence of APA is context-dependent and must be validated rather than inferred from UTR length alone.

Tail composition is increasingly treated as regulatory information. Poly(A) length, terminal uridylation, mixed tails, and tail-binding proteins can influence RNA decay and translation. However, the field has not reduced all tail effects to a single universal code. The most reliable claims remain enzyme-, substrate-, and context-specific.

Replication-dependent histone mRNA processing is now understood as a specialized, structurally defined cleavage pathway. SLBP recognizes the stem-loop, U7 snRNA pairs with the HDE, the LSM10-LSM11-containing U7 ring and FLASH recruit a CPSF73-CPSF100-symplekin-centered catalytic module, and CPSF73 cleaves without obligate long-tail synthesis. Histone-locus organization, S-phase signaling, translation-linked decay, and redundant 3′-end remodeling pathways connect the cleavage reaction to DNA synthesis. The consensus includes important boundaries: histone variants and developmental programs can use polyadenylation, and polyadenylated histone reads are not interpretable without gene and biological context.

## Open Questions, Controversies, Deprecated Models, and Common Misconceptions

Open questions:

- How completely can poly(A)-site choice be predicted from local sequence? Current models improve prediction, but chromatin, transcription speed, splicing, RNA-binding proteins, cell state, and RNA stability remain important.
- Which APA events are functional rather than tolerated isoform variation? Functional claims require perturbation of the site or regulator and measurement of RNA, protein, and phenotype.
- How general is the tail-code model? Tail composition clearly matters in selected pathways, but a universal grammar for all RNA tails remains speculative.
- How do polyadenylation factors choose between productive mRNA maturation and surveillance of cryptic transcripts? Recent work on tailing of pervasive transcription products sharpens this question.
- What are the best direct assays for therapeutic mRNA tail quality and in-cell tail remodeling? The current chapter needs direct manufacturing and clinical-platform references before making final design recommendations.
- How do histone-locus bodies control catalytic flux rather than merely concentrate factors, and which contacts are conserved between mammalian NPAT- and Drosophila Mxc-organized loci?
- How often do endogenous replication-dependent histone genes switch between U7-dependent and canonical poly(A)-site processing in normal development, stress, aging, and disease?
- Which combinations of TUT7/TENT3B, 3′hExo/ERI1, decapping, and exonuclease activities dominate histone mRNA decay in particular cell types and perturbations?

Common misconceptions:

- "AAUAAA alone defines a poly(A) site." Site use depends on spacing, downstream elements, factor context, transcription, and competition.
- "Poly(A) always stabilizes RNA." In canonical nuclear mRNAs it usually supports maturation and stability, but in bacteria, organelles, and surveillance contexts adenylation can promote decay.
- "Shorter 3′ UTRs always produce more protein." The effect depends on which regulatory elements are lost and which factors are active in the cell.
- "Cleavage, polyadenylation, and termination happen after transcription has finished." They are co-transcriptional or tightly coupled to elongating RNA polymerase II.
- "All mature eukaryotic mRNAs carry a long poly(A) tail." Most replication-dependent histone mRNAs in metazoan somatic cells mature by U7-directed cleavage and retain a stem-loop-bound SLBP instead.
- "CPSF73 cleavage proves that polyadenylation follows." CPSF73 is reused in histone pre-mRNA cleavage, where the normal mature product does not receive a long poly(A) tail.
- "A polyadenylated histone read always indicates defective processing." It can mark U7-pathway failure, but it can also derive from a replication-independent variant or a normal organism-, tissue-, or developmental-stage-specific program.
- "Poly(A)-selected RNA-seq quantitatively measures normal replication-dependent histone mRNA." The selection step depletes correctly processed nonpolyadenylated molecules and can preferentially reveal exceptional or misprocessed polyadenylated products.