# Chapter 15. Transcription-Unit Architecture, Promoters, Enhancers, Terminators, and Boundaries

## Scope Note

This chapter explains how genomes are organized to produce RNA. [Chapter 14](chapter1013.md) describes the repertoire of RNA-producing genes and genetic elements. This chapter owns the architecture that decides where transcription begins, how far RNA polymerase travels, which neighboring genes are copied together, how enhancer DNA changes promoter output, where antisense transcription initiates, and where a primary transcription unit ends. The central objects are promoters, enhancers as DNA elements, initiation sites, terminators, operons, polycistrons, readthrough intervals, antisense initiation sites, and unit boundaries. These objects overlap with gene annotation, but they are not identical to genes.

The chapter uses "RNA production" broadly. It includes bacterial and archaeal transcription, eukaryotic nuclear transcription by RNA polymerases I, II, and III, organellar transcription, viral transcription, mobile-element transcription, and regulatory transcription that may produce unstable RNAs. Surveillance, containment, and disposal of unstable or cryptic products belong to [Chapter 31](chapter1030.md). Claims that enhancer, promoter-associated, antisense, or readthrough RNA products act as regulatory molecules belong to [Chapter 93](chapter1088.md). The enzymology of polymerases is developed in Chapters [20](chapter1019.md) through [24](chapter1023.md), and transcript-model versioning is developed in [Chapter 18](chapter1017.md). Here the focus is the genomic logic that places RNA synthesis in space, time, strand, and cellular context.
## Executive Summary

A promoter is a DNA region that enables a transcription system to initiate RNA synthesis at one or more transcription start sites. The definition is functional, not merely sequence-based. A bacterial sigma-factor promoter, an archaeal TATA-box-containing promoter, a mammalian RNA polymerase II CpG-island promoter, an RNA polymerase III internal promoter, a mitochondrial promoter, and a viral promoter can all be promoters, but each is read by a different set of proteins and structural constraints. A promoter motif is therefore a hypothesis about initiation until evidence links the region to transcription initiation in the relevant cell type, organism, polymerase system, and condition.

Transcription initiation is not complete when the first phosphodiester bond forms. RNA polymerase must select a start site, synthesize an initial RNA, escape or clear the promoter, and enter productive elongation. In eukaryotic RNA polymerase II transcription, early elongation includes promoter-proximal pausing, release from initiation factors, kinase-dependent transitions, re-initiation, and transcriptional bursting. Current work on RNA polymerase II re-initiation and bursting, precise polymerase mapping, and CDK7-dependent promoter escape supports the view that promoter output is a kinetic process rather than a simple on/off switch.

An operon is a set of genes transcribed from a shared promoter into a polycistronic primary RNA. Bacterial operons are the classical model because they coordinate genes that often act in the same pathway. Polycistronic transcription also occurs in eukaryotic and organellar systems, but the logic can differ. In trypanosomes, long polycistronic gene arrays are resolved largely by trans-splicing, polyadenylation, chromatin organization, and post-transcriptional control. In mitochondria and plastids, long precursors can be cut, edited, stabilized, or degraded to produce mature RNAs. Shared transcription initiation coordinates RNA production, but mature RNA and protein output can be separated by processing, translation, and decay.

An enhancer is a regulatory DNA element that increases transcription from one or more promoters, often at a distance and often in a cell-type-specific manner. Enhancers act through transcription factor binding, coactivators, chromatin accessibility, nucleosome remodeling, promoter-proximal pausing control, and spatial proximity to promoters. Mediator is a major coactivator that connects transcription factors with RNA polymerase II machinery and participates in several steps of transcription regulation. Enhancer-promoter communication is not explained by one universal geometry. Looping, hubs, transient contacts, transcriptional condensate-like assemblies, and local chromatin context can all contribute, and their relative importance differs by locus and developmental system.

Enhancer and promoter-proximal transcription are major reasons why transcription units cannot be inferred only from stable mature RNA. Active enhancers can initiate short, often unstable transcripts. Promoters can initiate divergent, upstream, paused, or rapidly degraded RNAs. These products reveal initiation architecture even when they do not accumulate at steady state. Their stability and disposal hand off to [Chapter 31](chapter1030.md), while claims that the RNA molecules themselves regulate a target hand off to [Chapter 93](chapter1088.md). The architectural evidence must distinguish enhancer DNA activity, initiation at enhancer DNA, detection of the resulting RNA, and downstream promoter output.

Terminators and transcription boundaries define where RNA synthesis stops, but stopping is mechanistically diverse. Bacteria use intrinsic terminators, factor-dependent termination, regulated antitermination, and context-dependent readthrough. Phage lambda N antitermination shows that protein factors can alter RNA polymerase processivity and suppress termination. Eukaryotic RNA polymerase II termination is coupled to cleavage and polyadenylation and can be shaped by Mediator, elongation factors, chromatin state, and 3′ end processing factors. Readthrough and antisense transcription can be regulated biology, stress response, disease-associated transcription, or technical artifact.

Genome architecture is itself an RNA-production system. Promoters, enhancers, insulators, terminators, operons, repeats, mobile elements, nucleosome positions, chromatin domains, nucleoid organization, replication context, and nuclear compartmentalization all influence which RNAs are made. Repeats and mobile elements can donate promoters, splice sites, polyadenylation sites, antisense sequence, enhancer-like elements, and RNA-processing hazards. Alu-rich human genome regions provide a concrete example: DHX9 suppresses RNA processing defects that arise from Alu expansion. [Chapter 16](chapter1015.md) treats repeat-derived RNAs in detail; [Chapter 15](chapter1014.md) introduces the architectural logic needed to understand why repeats affect transcription units.

## Concept Inventory

- **Promoter:** a DNA region that enables transcription machinery to initiate RNA synthesis at one or more transcription start sites. The region may include a core promoter, nearby regulatory motifs, nucleosome positioning signals, and binding sites for transcription factors or polymerase-specific initiation factors. A promoter motif, such as a bacterial -10 element or a eukaryotic TATA box, is not a promoter by itself unless the surrounding region supports initiation in the relevant biological context.
- **Core promoter:** the promoter-proximal sequence region that positions or helps position basal transcription machinery. In eukaryotic RNA polymerase II transcription, the core promoter can include combinations of TATA box, initiator, downstream promoter element, CpG-rich sequence, and other features. In bacteria, promoter-proximal recognition is mediated largely by sigma-factor contacts with DNA elements. Core promoter architecture differs strongly across polymerase systems and organisms.
- **Enhancer:** a regulatory DNA element that increases transcription from one or more promoters. Enhancers often act at a distance, can work in either orientation in reporter assays, and are frequently cell-type-specific. The word enhancer refers to the DNA element and its regulatory activity. It should not be confused with enhancer RNA function, which is a separate claim.
- **Terminator:** a sequence, RNA structure, protein-dependent process, or processing-coupled event that ends transcription. Bacterial intrinsic terminators, Rho-dependent termination, eukaryotic polyadenylation-coupled termination, RNA polymerase III termination, organellar termination, and viral termination are related in outcome but not identical in mechanism.
- **Operon:** a cluster of genes transcribed from a shared promoter into a polycistronic RNA under coordinated regulation. Operons are common in bacteria and archaea, occur in organellar and some eukaryotic systems, and are also used in engineered expression cassettes. An operon coordinates transcriptional input, but post-transcriptional steps can still produce unequal outputs.
- **Polycistronic transcript:** a single RNA molecule that contains multiple coding regions or mature RNA products. Polycistronic RNAs can encode several proteins, several noncoding RNAs, or a mixture of products after processing.
- **Transcription unit:** the genomic interval copied into one primary RNA under a particular initiation and termination context. A transcription unit can contain one gene, several genes, an enhancer transcript, an antisense transcript, a readthrough interval, a viral transcription cassette, or a processing-dependent precursor. Transcription units can overlap and can change with cell type, stress, development, or infection.
- **Enhancer RNA:** RNA produced from an enhancer region; also called eRNA. Its production maps enhancer-associated initiation, but its stability, disposal, and possible RNA-mediated function are separate questions owned by [Chapter 31](chapter1030.md) and [Chapter 93](chapter1088.md).
- **Promoter-proximal RNA:** a short RNA produced near a promoter during divergent transcription, early elongation, pausing, or abortive initiation. Production is an architectural observation even when surveillance prevents steady-state accumulation.
- **Promoter escape:** the transition in which RNA polymerase leaves the initiation complex and enters productive elongation. It is regulated separately from start-site selection.
- **Transcriptional pausing:** a temporary halt or slowdown of RNA polymerase. Pausing can regulate expression, coordinate processing, help synchronize developmental responses, or reflect obstacles such as DNA-bound proteins, nucleosomes, RNA structures, DNA damage, or torsional stress.
- **Antitermination:** a mechanism that allows RNA polymerase to bypass termination signals. Phage lambda N-dependent antitermination is a classical example, but antitermination also appears in other bacterial, phage, and regulatory contexts.
- **Readthrough:** Transcriptional readthrough is continued transcription beyond an expected terminator or annotated gene boundary. Readthrough can be regulated, stress-associated, disease-associated, or artifactual. It should be interpreted with transcript-end evidence and controls.
- **Antisense transcription:** transcription from the DNA strand opposite a reference gene or transcript. Strand-aware initiation evidence establishes production; regulatory-function claims require the distinct causal standards developed in [Chapter 93](chapter1088.md).
- **Transcription boundary:** a start, stop, processing, chromatin, or topological boundary that constrains the extent of a transcription unit. Annotation boundaries and molecular boundaries can differ.
- **Mediator:** a conserved coactivator complex that connects transcription factors with RNA polymerase II machinery and regulates transcription at several steps, including initiation, early elongation, enhancer responsiveness, and in some contexts termination-linked events.
- **Nascent RNA:** RNA that is being synthesized or has recently emerged from RNA polymerase. Nascent RNA measurements are often required to observe promoter-proximal transcripts, enhancer transcription, pausing, and immediate readthrough.

## What to Know Before Reading This Chapter

The first prerequisite is strand and polarity. DNA has two antiparallel strands, and an RNA polymerase uses one strand as template to synthesize RNA in the 5′ to 3′ direction. The transcribed RNA has the same sequence polarity as the non-template strand except that RNA contains uracil instead of thymine. A promoter must be interpreted with strand information because initiation on opposite strands produces different RNAs.

The second prerequisite is the difference between a DNA feature, a primary transcript, and a mature RNA. A DNA feature is a genomic interval. A primary transcript is the first RNA copied from that interval. A mature RNA is the processed molecule that remains after capping, cleavage, splicing, trans-splicing, trimming, editing, polyadenylation, modification, ribonucleoprotein assembly, export, or decay. A single primary transcript can become many mature products; a single mature RNA class can arise from different primary transcript architectures.

The third prerequisite is the distinction between steady-state abundance and production rate. Steady-state RNA-seq measures molecules that survived extraction and library preparation in a sample. It reflects synthesis, processing, export, localization, translation, storage, and decay. Nascent RNA methods, transcription start site mapping, and polymerase occupancy assays are closer to RNA production, but each has its own biases. Therefore promoter activity, enhancer transcription, readthrough, and antisense transcription are best interpreted with methods that directly address initiation, elongation, or transcript ends.

The fourth prerequisite is that regulatory DNA is interpreted by molecular systems. Bacteria use sigma factors, activators, repressors, nucleoid-associated proteins, and RNA structures. Archaea use a transcription apparatus related to eukaryotic RNA polymerase II initiation machinery. Eukaryotic nuclei use chromatin, general transcription factors, sequence-specific transcription factors, coactivators, polymerase-specific promoter classes, enhancer-promoter communication, and RNA processing. Organelles and viruses use compact and sometimes unusual combinations. A term such as "promoter" or "terminator" is most useful when paired with the system in which it operates.

The running examples in this chapter are bacterial operons, RNA polymerase II promoters, enhancer RNAs, trypanosome polycistronic expression, phage lambda antitermination, and repeat-rich human loci. These examples recur because they show different ways that initiation, elongation, termination, processing, and genome architecture combine to produce RNA.

## 15.1. Promoter classes and initiation logic across organisms

![Figure 15.1. Promoter Logic Across Polymerase Systems](../assets/figures/chapter1014_figure1.png)

**Figure 15.1. Promoter Logic Across Polymerase Systems.** A promoter is defined by its ability to direct transcription initiation in a specific biological system, not by sequence alone. Bacterial sigma-factor promoters rely on -10 and -35 elements read by sigma-factor subunits, archaeal promoters use a TATA box recognized by TBP-related factors, and eukaryotic RNA polymerase I, II, and III promoters each require distinct general transcription factors and cofactors. Organellar, viral, and mobile-element promoters reflect their own evolutionary and replication histories. This comparison illustrates why promoter function must be validated in context, not inferred from motif similarity alone.

### Promoters Are Functional Initiation Regions

A promoter is best defined by what it does: it enables transcription machinery to initiate RNA synthesis. A DNA motif upstream of a gene can suggest a promoter, but promoter function requires the correct polymerase system, regulatory proteins, DNA accessibility, strand orientation, start-site selection, and cellular state. This functional definition matters because different organisms and polymerases use different initiation grammars. A promoter in a bacterial chromosome, an archaeal genome, a mammalian nucleus, a mitochondrion, a chloroplast, a phage genome, or a retrotransposon is not simply a conserved short word in DNA.

The causal steps are similar at a high level. First, a polymerase or polymerase-associated factor recognizes a DNA region. Second, the DNA strands are opened or positioned so the template strand can enter the active site. Third, the first nucleotides are joined. Fourth, the polymerase either aborts, pauses near the promoter, or escapes into productive elongation. Different systems assign these steps to different factors. In bacteria, sigma factors help RNA polymerase find promoter elements and open DNA. In eukaryotic RNA polymerase II transcription, general transcription factors assemble a preinitiation complex, transcription factors and cofactors regulate that complex, and kinases help drive promoter escape and early elongation.

Promoter output is a kinetic result rather than a static label. Two promoters with similar average RNA output can differ in burst frequency, burst size, pause duration, start-site distribution, re-initiation probability, and response to enhancer input. Recent RNA polymerase II-focused work emphasizes this point by treating re-initiation and bursting as regulated components of transcription output. Precise RNA polymerase maps have shown that promoter architecture can direct initiation and pausing patterns, and CDK7 kinase activity can promote promoter escape by facilitating release of initiation factors.

### Bacterial Promoters

In bacteria, the core transcription enzyme associates with sigma factors that confer promoter specificity. A sigma factor is a protein subunit that helps RNA polymerase recognize promoter elements and initiate transcription at appropriate genes. Many bacterial promoters contain recognizable -10 and -35 elements, named for their approximate distance upstream of the transcription start site. These elements are not universal, and their exact sequence, spacing, and factor dependence vary by sigma factor and species. Activators, repressors, DNA bending proteins, nucleoid-associated proteins, DNA supercoiling, and local sequence context can strongly affect promoter function.

A concrete bacterial example is an inducible metabolic operon. A promoter upstream of a sugar-utilization operon can be repressed when the substrate is absent and activated when the substrate is present. Once initiated, the same primary RNA may include multiple coding regions. That design allows transcription initiation to coordinate a pathway, but translation and mRNA stability can still differ across the cistrons.

**Table 15.1. Promoter Classes and Evidence.** Comparison of major promoter classes across biological systems, emphasizing that each is defined by the polymerase and initiation factors it recruits rather than by a universal sequence motif.

| Promoter class | Organism/system | Polymerase | Key sequence features | Evidence methods | Main caveat |
| --- | --- | --- | --- | --- | --- |
| **Sigma-factor promoter** | Bacteria | Multi-subunit bacterial RNAP | -10 and -35 elements; UP element; spacing | TSS mapping; reporter assay; sigma-factor perturbation; in vitro transcription | Motif sequences and spacing vary by sigma type; activators and repressors modify output |
| **TATA-box promoter** | Archaea | Archaeal RNAP | TATA box; BRE element | TSS mapping; TBP and TFB binding; in vitro transcription | Genome organization is operon-like despite eukaryote-related basal factors |
| **RNA Pol I promoter** | Eukaryotes (nucleolus) | RNA Pol I | Species-specific upstream and core promoter elements | rRNA precursor TSS mapping; Pol I ChIP; reporter | Highly species-specific; specialized for high-output rDNA repeat transcription |
| **RNA Pol II promoter** | Eukaryotes (nucleus) | RNA Pol II | TATA box (subset); Inr; DPE; CpG island; focused or broad TSS | CAGE; GRO-seq; nascent RNA-seq; ChIP; perturbation | Highly heterogeneous; many promoters lack canonical motifs; chromatin-dependent |
| **RNA Pol III promoter** | Eukaryotes (nucleus) | RNA Pol III | Internal A-box and B-box (tRNA); upstream elements (5S rRNA, snRNA) | Pol III ChIP; run-on; in vitro transcription | Can be internal to the transcribed region; architecture varies by gene class |
| **Organellar promoter** | Mitochondria; plastids | Single-subunit phage-like polymerase or organelle-specific factors | Variable; phage-T7-like elements in some lineages | TSS mapping; in organello transcription; factor perturbation | Shaped by endosymbiotic history; long precursors require extensive downstream processing |
| **Viral promoter** | DNA viruses; retroviruses | Host or viral RNAP | Immediate-early, early, late motifs; LTR elements | Viral reporter; TSS mapping; RNAP occupancy | Requires viral factors; LTR logic differs before and after chromosomal integration |
| **Mobile-element promoter** | Retrotransposons; SINEs; LINEs | Host RNAP (typically) | LTR elements; internal Pol III promoter in SINEs | Reporter; TSS mapping in native locus | Can donate promoters or antisense promoters to flanking host genes; condition-specific activation |

Evidence for bacterial promoter function can come from transcription start site mapping, reporter assays, mutational analysis of promoter elements, sigma-factor perturbation, RNA polymerase occupancy, in vitro transcription, and genetic tests of gene expression. A motif scan alone is weak evidence because random DNA can contain motif-like sequences. Bacterial transcription-initiation sources, including a biophysical review of mycobacterial initiation and a primary study of alternative bacterial RNA polymerase transcription cycles, support the distinction between motif recognition, open-complex formation, abortive initiation, and productive elongation.

### Archaeal Promoters

Archaeal transcription uses a single multi-subunit RNA polymerase related to eukaryotic RNA polymerases and basal transcription factors related to eukaryotic TATA-binding protein and transcription factor B. Many archaeal promoters include a TATA box and a nearby recognition element, but archaeal promoter grammar is not simply eukaryotic RNA polymerase II grammar placed in a compact genome. Archaeal genomes often have bacterial-like operons and compact organization, while their transcription machinery has eukaryote-like ancestry.

The important boundary case is that archaeal promoters can sit in operon-like genome arrangements. A reader should not assume that eukaryote-like basal factors imply eukaryote-like gene architecture. Archaeal initiation demonstrates that polymerase evolution and genome organization can be decoupled: archaeal RNA polymerase and initiation-factor studies show eukaryote-related transcription machinery, while factor-dependent archaeal termination and compact operon-like arrangements keep the genome-architecture logic distinct from typical metazoan Pol II genes.

### Eukaryotic Nuclear Promoters

Eukaryotic nuclei contain at least three major nuclear RNA polymerase systems. RNA polymerase I transcribes large ribosomal RNA precursors in the nucleolus. RNA polymerase II transcribes most protein-coding genes and many long noncoding, enhancer-associated, promoter-associated, and small nuclear RNA precursors. RNA polymerase III transcribes tRNAs, 5S rRNA, and several other small structured RNAs. Each polymerase has characteristic promoter architectures and factor requirements.

RNA polymerase II promoters are the most heterogeneous. Some are focused promoters with a narrow transcription start site distribution. Others are broad promoters that initiate over a region. Some contain TATA boxes; many do not. CpG-rich mammalian promoters, developmental promoters, tissue-specific promoters, bidirectional promoters, and promoter regions embedded in repeat-rich sequence all differ in chromatin context and regulatory logic. Transcription factors bind regulatory motifs, coactivators such as Mediator help communicate with the polymerase machinery, and chromatin state affects access. Promoter-proximal pausing can make an initiated polymerase wait near the start site until signals release it into elongation.

RNA polymerase I and III promoters are equally important for RNA biology because they make abundant structural and translational RNAs. RNA polymerase I promoter architecture supports high-output rRNA synthesis, and Pol I-specific factors specialize rDNA transcription relative to other nuclear polymerases. RNA polymerase III promoters can be upstream of the gene or internal to the transcribed region, as in many tRNA genes, and Pol III transcription is tightly linked to tRNA production and processing. Internal promoters remind the reader that "promoter upstream of a gene" is not a universal rule.

### Organellar, Viral, and Mobile-Element Promoters

Mitochondria and plastids use specialized transcription systems shaped by endosymbiotic history, genome reduction, and nuclear-encoded factors. Some organellar promoters resemble bacterial promoter logic, whereas others depend on phage-like single-subunit polymerases or organelle-specific factors. Organellar transcription often produces long primary transcripts that require extensive RNA processing, as illustrated by plant organelle genetics, chloroplast phage-type RNA polymerase work, and mitochondrial RNA-processing reviews. [Chapter 17](chapter1016.md) treats organellar RNA genes and processing in detail.

Viruses and mobile elements carry promoter logic that reflects their replication strategy. DNA viruses can use host polymerases, virally encoded polymerases, or both. Retroviruses and long terminal repeat retrotransposons use long terminal repeats as promoter and regulatory regions after integration, while positive-strand RNA viruses and retroviral life cycles show why viral RNA-production units must be interpreted with life-cycle context rather than only host-gene annotation rules. Non-long-terminal-repeat retrotransposons and other mobile elements can influence host transcription by donating promoter-like elements, antisense promoters, or polyadenylation signals; LINE/SINE reviews and Pol III-associated retrotransposon work provide boundary anchors for those claims. These features are covered in more detail in Chapters [16](chapter1015.md), [101](chapter1096.md), [115](chapter1109.md), and [120](chapter1114.md).

**Table 15.2. Transcription Unit Types.** Summary of the major categories of transcription unit, with representative organisms, boundary signals, and common annotation hazards.

| Unit type | Primary RNA | Mature products | Organisms | Boundary signals | Annotation hazard |
| --- | --- | --- | --- | --- | --- |
| **Monocistronic gene** | Single-gene precursor | One mRNA or ncRNA | All domains | Upstream promoter; downstream terminator or poly(A) signal | Alternative TSS or alternative poly(A) sites can extend or split boundaries |
| **Operon** | Polycistronic mRNA | Multiple mRNAs or protein products | Bacteria; archaea; organelles | Shared promoter; operon-distal terminator | Adjacent co-oriented genes are not automatically an operon; continuous transcript evidence required |
| **Eukaryotic polycistronic array** | Long polycistronic precursor | Individual mRNAs via trans-splicing and polyadenylation | Trypanosomes and related kinetoplastids | Chromatin boundaries; trans-splice acceptors; poly(A) signals | Regulation is largely post-transcriptional; per-gene promoters are absent |
| **Organellar polycistronic precursor** | Long multi-gene primary transcript | rRNA, tRNA, and mRNA products via processing | Mitochondria; plastids | Processing sites; organelle-specific endonucleases; no per-mature-RNA terminator | Mature RNA ends do not define individual transcription-unit boundaries |
| **Readthrough transcript** | Extended RNA past annotated end | May overlap downstream gene or regulatory region | All; stress-associated in eukaryotes | Defective or bypassed terminator; altered poly(A) signals | Can resemble a long gene or alternative isoform; transcript-end evidence required |
| **Antisense transcript** | Opposite-strand RNA | Variable; often unstable | All | Independent promoter; bidirectional promoter activity; readthrough from neighbor | Strand-non-specific libraries produce false antisense signal; function hands off to [Chapter 93](chapter1088.md) |
| **Enhancer transcript** | Short eRNA; often bidirectional | Unstable ncRNA; not typically processed to a stable product | Eukaryotes | Enhancer-associated initiation; no canonical terminator | eRNA abundance does not prove RNA-molecule function; active enhancer DNA must be distinguished |
| **Viral transcription unit** | Nested or cascaded viral RNA | Immediate-early, early, late, or subgenomic products | DNA viruses; retroviruses | Viral promoter and terminator elements; LTR | Viral TUs overlap, cascade, and co-opt host and viral factors; stage-specific logic required |

> **Box 15.1. Promoter Motif Versus Promoter Function**
>
> - A promoter motif is a sequence pattern associated with initiation in a particular system; a functional promoter is a genomic region that actually supports initiation in a biological context.
> - Strong evidence for promoter function includes transcription start site mapping, polymerase occupancy, factor binding, nascent RNA production, mutational loss of initiation, and restoration by sequence rescue.
> - A motif without initiation evidence should be annotated as a candidate promoter, not as a proven promoter.

A promoter motif is a sequence pattern associated with initiation in a particular system. A functional promoter is a genomic region that actually supports initiation in a biological context. Strong evidence for promoter function can include transcription start site mapping, polymerase occupancy, factor binding, nascent RNA production, mutational loss of initiation, and restoration by sequence rescue. A motif without initiation evidence should be annotated as a candidate promoter, not as a proven promoter.

## 15.2. Operons, polycistronic transcripts, and processing-dependent gene expression

### Classical Operon Logic

An operon is a set of genes transcribed from a shared promoter into one polycistronic primary RNA. The classical bacterial operon model explains how cells coordinate genes whose products act together, such as enzymes in a metabolic pathway or proteins in a transport system. One regulatory decision at a promoter can change the transcription of several downstream coding regions. This design saves regulatory complexity and makes pathway-level expression possible.

The mechanism has several steps. A regulatory signal changes promoter activity. RNA polymerase initiates at the operon promoter. The polymerase copies multiple coding regions into one primary RNA. Ribosomes can initiate translation at multiple ribosome-binding sites within that RNA. RNA structures, intercistronic regions, translation coupling, cleavage, and decay determine the final amount of each protein. Therefore transcriptional co-regulation does not guarantee equal protein output.

The evidence for an operon includes shared transcription start site evidence, continuous RNA across adjacent genes, co-regulation across conditions, loss of downstream transcription when the upstream promoter is disrupted, and transcript-end evidence beyond the last gene. However, adjacent co-oriented genes are not automatically an operon. Short intergenic distance, functional relatedness, or expression correlation can suggest an operon, but transcript evidence is needed. Reviews of co-regulated genes and gene clusters support using shared transcription and regulation as evidence standards rather than treating proximity alone as an operon definition.

### Polycistronic Transcripts Outside Classical Bacteria

Polycistronic transcription is common in bacteria and archaea, but it is not restricted to them. Some eukaryotes use polycistronic gene arrays. Trypanosomes provide a major example. In trypanosomes, long arrays are transcribed polycistronically, and mature messenger RNAs are generated by trans-splicing and polyadenylation. This arrangement shifts much gene regulation away from a simple one-gene-one-promoter model and toward chromatin, RNA processing, RNA stability, and translation. Precision editing of histone tails in trypanosomes disrupting polycistronic gene expression controls demonstrates that chromatin can influence polycistronic RNA production.

Some animals also use operon-like arrangements for subsets of genes, and some eukaryotic expression systems use engineered polycistronic cassettes. In nuclear eukaryotes, however, most protein-coding genes are transcribed as separate precursor mRNAs. When a long eukaryotic RNA overlaps several genes, the reader should ask whether it is a functional polycistronic precursor, a readthrough transcript, an unprocessed transcript, a lncRNA, an annotation artifact, or a rare regulated isoform.

### Organellar and Viral Polycistronic RNAs

Mitochondria and plastids often rely on polycistronic or multigenic transcription followed by processing. A long organellar precursor can contain rRNAs, tRNAs, and mRNAs. Processing enzymes then cut the precursor, mature ends, add or trim tails, edit bases in some lineages, and stabilize selected products. The mature output depends on both transcription and processing. [Chapter 17](chapter1016.md) develops these systems, but [Chapter 15](chapter1014.md) introduces the general rule: in organelles, transcription-unit boundaries and mature RNA boundaries often differ.

Viruses also use compact transcription strategies. Some viral genomes encode nested transcription units, subgenomic RNAs, overlapping reading frames, or polycistronic messages. Viral transcription architecture is shaped by genome size, host polymerase access, replication strategy, and translation strategy. A viral transcription unit can therefore be both an RNA-production unit and part of a replication cycle, as shown by retroviral genomic RNA packaging logic and alphavirus genomic/subgenomic RNA life-cycle organization.

### Processing-Dependent Gene Expression

Processing-dependent expression means that mature RNA output is determined after or during transcription by cleavage, splicing, trans-splicing, polyadenylation, trimming, RNA editing, RNA modification, localization, translation, and decay. Polycistronic transcription makes processing especially visible because one primary RNA can yield products with different fates. A tRNA embedded in an organellar precursor may be cut out and matured. A trypanosome coding region may require trans-splicing to receive a spliced leader and polyadenylation to become a mature mRNA. A bacterial polycistronic mRNA may be cleaved so that one part is stabilized and another part is degraded.

![Figure 15.2. Polycistronic RNA and Processing-Dependent Expression](../assets/figures/chapter1014_figure2.png)

**Figure 15.2. Polycistronic RNA and Processing-Dependent Expression.** A single polycistronic primary transcript can contain multiple coding regions or mature-RNA precursors. In bacteria, operons allow one initiation decision to coordinate several downstream coding regions, while in trypanosomes long polycistronic arrays are resolved into individual products. The visual emphasizes the shared production unit and uses unequal downstream outputs only as a handoff to processing and [Chapter 31](chapter1030.md).

Polycistronic transcription coordinates initiation across multiple products, but downstream steps can decouple abundance. Internal RNA structures can block or expose translation starts. Cleavage can stabilize one segment and destabilize another. Ribosomes can protect coding regions from decay. Processing factors can favor specific products. Therefore equal transcriptional origin is not equal molecular output.

## 15.3. Enhancers, promoter-proximal initiation, and regulatory transcription architecture

### Enhancers as Regulatory DNA Elements

An enhancer is a regulatory DNA element that increases transcription from one or more promoters. Enhancers are often defined operationally by their ability to stimulate promoter activity in reporter assays, by occupancy of transcription factors and coactivators, by chromatin marks associated with active regulatory DNA, and by their effect on nearby or sometimes distant genes. Many enhancers are cell-type-specific because the necessary transcription factors are present only in certain developmental states, tissues, immune states, or environmental conditions.

Enhancers should be introduced as DNA elements before discussing enhancer-associated RNA products. The DNA element can bind proteins, alter chromatin, and change promoter initiation regardless of whether the RNA product has an independent role. This chapter therefore maps enhancer DNA, enhancer-associated initiation, directionality, and target-promoter output; any claim that the RNA molecule itself acts mechanistically is handed to [Chapter 93](chapter1088.md).

![Figure 15.3. Enhancer-Promoter Communication and eRNA Interpretation](../assets/figures/chapter1014_figure3.png)

**Figure 15.3. Enhancer-Promoter Communication and eRNA Interpretation.** Enhancers stimulate promoter initiation through sequence-specific transcription factors, cofactors including Mediator, chromatin remodeling, topology, and regulation of promoter-proximal pausing. The figure separates enhancer DNA activity, initiation at enhancer DNA, RNA detection, and RNA-product function. This chapter owns the first two production-architecture claims; persistence and disposal hand off to [Chapter 31](chapter1030.md), and product function hands off to [Chapter 93](chapter1088.md).

Enhancer-promoter communication can be described in causal steps. Sequence-specific transcription factors bind enhancer DNA. These factors recruit cofactors, chromatin remodelers, histone-modifying enzymes, and Mediator. Chromatin accessibility and local topology bring the enhancer into a regulatory relationship with a promoter. Promoter-proximal polymerase initiation, pausing, release, and re-initiation change. Nascent RNA output from the target promoter increases, decreases, or changes in burst kinetics. This simplified pathway is useful, but it does not imply one universal physical model.

Current models include stable loops, transient contacts, regulatory hubs, local concentration effects, phase-separation-like assemblies, and chromatin-domain constraints. Drosophila developmental systems have provided important examples of enhancer-promoter specificity and competition. Broader reviews emphasize that enhancer-promoter communication can be interpreted as hubs or loops depending on the assay, time scale, and locus. Mediator is a central coactivator in many of these models, but its roles differ by locus and transcription step.

### Promoter-Proximal RNAs and Pausing

Promoter-proximal RNA is RNA made near a promoter during early transcription. Some promoter-proximal RNAs are abortive initiation products. Some are short divergent transcripts made from bidirectional promoters. Some are upstream antisense RNAs. Some are paused nascent RNAs that mark polymerase waiting near a promoter. These RNAs are often unstable and can be degraded rapidly by nuclear surveillance pathways.

Promoter-proximal pausing allows cells to regulate transcription after initiation but before productive elongation. A paused polymerase can keep a promoter region accessible, prepare a gene for rapid activation, coordinate mRNA capping and early processing, or integrate enhancer signals. Pausing is not merely polymerase failure; it can be an active regulatory state. Precise maps of RNA polymerase helped reveal how promoter architecture directs initiation and pausing. The details of pause factors and elongation regulation are developed in [Chapter 24](chapter1023.md).

### Enhancer RNAs

Enhancer RNAs are RNAs produced from active enhancer regions. Many enhancer RNAs are short, bidirectional, non-polyadenylated or weakly polyadenylated, and unstable, although the details vary. Their detection often requires nascent RNA sequencing, capped RNA methods, or exosome-sensitive conditions rather than standard steady-state poly(A) RNA-seq. Enhancer RNA abundance can correlate with enhancer activity, but correlation does not prove that the RNA molecule drives transcription.

**Table 15.3. Enhancer-Transcription Evidence Boundaries.** Each evidence type supports a different production-architecture claim. RNA-product persistence and function are named only to identify the handoff.

| Evidence type | Enhancer DNA activity | Initiation at enhancer | RNA detection | Interpretation boundary |
| --- | --- | --- | --- | --- |
| **Chromatin accessibility or active marks** | Candidate support | No | No | Regulatory correlation is not initiation evidence |
| **Transcription-factor or Mediator occupancy** | Candidate support | No | No | Occupancy does not define a transcribed unit or target promoter |
| **Reporter assay** | Activity in reporter context | Sometimes, if starts are mapped | Sometimes | Reporter geometry may not reproduce the endogenous locus |
| **Nascent RNA and start-site mapping** | Supports an active regulatory region | Yes | Yes | Establishes production, not steady-state persistence or RNA function |
| **Polymerase occupancy plus directional run-on signal** | Supports active enhancer architecture | Yes | Yes | Paused or short transcription may not yield a stable product |
| **RNA steady-state detection** | Indirect | No | Yes | Persistence and disposal hand off to [Chapter 31](chapter1030.md) |
| **RNA-product perturbation or rescue** | Not an architecture test by itself | No | Product is manipulated | Causal RNA-function interpretation belongs to [Chapter 93](chapter1088.md) |

This chapter uses enhancer-associated RNA only as an output of regulatory transcription architecture. Nascent initiation, directionality, polymerase occupancy, and transcript boundaries can establish that an enhancer region is transcribed. They do not establish that the RNA molecule causes target-promoter regulation. Product stability and surveillance hand off to [Chapter 31](chapter1030.md), and the perturbation and rescue logic needed to test RNA-mediated function hands off to [Chapter 93](chapter1088.md). High-sensitivity nascent transcript sequencing illustrates how assay sensitivity changes the detectable enhancer-transcription landscape without settling the function of each product.

> **Box 15.2. Enhancer Transcription Is Not Automatically eRNA Function**
>
> Four claims must be separated:
> - Enhancer DNA can regulate a promoter.
> - RNA polymerase can transcribe through the enhancer region.
> - The resulting RNA molecule can persist long enough to be detected.
> - That RNA molecule can have a direct regulatory function.
>
> Evidence for one claim does not automatically prove the others. Persistence and disposal belong to [Chapter 31](chapter1030.md); causal RNA-product function belongs to [Chapter 93](chapter1088.md).

Four claims must be separated. First, enhancer DNA can regulate a promoter. Second, RNA polymerase can initiate at the enhancer. Third, the resulting RNA can persist long enough to be detected. Fourth, the RNA molecule can have a direct regulatory function. This chapter owns the first two architectural claims and records the third as an observed product; [Chapter 31](chapter1030.md) and [Chapter 93](chapter1088.md) own disposal and causal RNA-function interpretation.

> **Box 15.3. Polycistronic Does Not Mean Equal Output**
>
> - Polycistronic transcription coordinates initiation across multiple products, but downstream steps can decouple abundance.
> - Internal RNA structures can block or expose translation starts.
> - Cleavage can stabilize one segment and destabilize another.
> - Ribosomes can protect coding regions from decay.
> - Processing factors can favor specific products.
> - Equal transcriptional origin is therefore not equal mature RNA or protein output; maturation and disposal mechanisms are treated in the processing chapters and [Chapter 31](chapter1030.md).

## 15.4. Termination sites, readthrough production, antisense initiation, and unit boundaries

### What Terminators Do

A terminator is any sequence feature, RNA structure, protein-dependent mechanism, or processing-coupled event that causes RNA polymerase to stop transcription and release the RNA or template. Termination prevents uncontrolled transcription into downstream genes, reduces collisions between transcription units, defines RNA ends, and helps coordinate processing. Terminators are not always perfect barriers. Polymerases can read through weak terminators, bypass terminators under antitermination control, or continue beyond an expected boundary in stress and disease contexts.

The first boundary distinction is between the molecular end of transcription and the mature RNA end. In bacteria, the termination event often closely defines the primary RNA end. In eukaryotic RNA polymerase II transcription, cleavage and polyadenylation produce the mature pre-mRNA 3′ end before the polymerase fully terminates downstream. Therefore the mature poly(A) site and the polymerase release point are related but not identical.

### Bacterial Termination and Antitermination

Bacteria use intrinsic termination and factor-dependent termination. Intrinsic terminators often depend on an RNA hairpin followed by a U-rich tract that destabilizes the transcription complex. Factor-dependent termination can involve proteins that engage RNA and polymerase to promote release. Termination efficiency depends on RNA sequence, RNA structure, polymerase state, translation, accessory proteins, and cellular conditions.

![Figure 15.4. Termination, Antitermination, and Readthrough Outcomes](../assets/figures/chapter1014_figure4.png)

**Figure 15.4. Termination, Antitermination, and Readthrough Outcomes.** RNA polymerase can stop at a canonical termination signal, be rendered termination-resistant by antitermination factors such as phage lambda N protein together with Nus factors, or continue through a bypassed terminator to produce readthrough RNA that may overlap downstream genes or antisense regions. In eukaryotic RNA polymerase II transcription, the mature poly(A) site produced by cleavage and the downstream polymerase release zone are spatially separated, so the annotated gene end, the cleavage site, and the transcription termination zone occupy different positions. The figure illustrates the range of boundary outcomes and the evidence needed to distinguish regulated biology from technical artifact.

Antitermination occurs when factors modify the transcription complex or RNA context so that polymerase ignores termination signals. Phage lambda N antitermination is a classical example. The lambda N antiterminator and host Nus factors strengthen a termination-resistant state of RNA polymerase, allowing transcription to continue through terminators that would otherwise stop it. This example teaches a general principle: termination is not merely a passive sequence feature; it can be actively regulated by proteins and RNA elements.

Bacterial termination should be treated as a mechanism family rather than a single motif rule. Structural work on intrinsic termination, a Bacillus subtilis terminator atlas, and processive antitermination reviews provide direct anchors for intrinsic terminators, genome-scale terminator annotation, and regulated bypass of termination signals. [Chapter 78](chapter1073.md) treats RNA leader-based attenuation and riboswitch-related termination in more detail.

### Eukaryotic RNA Polymerase II Termination

RNA polymerase II termination is coupled to pre-mRNA 3′ end formation. A protein-coding pre-mRNA typically contains a polyadenylation signal and downstream elements that recruit cleavage and polyadenylation factors. The RNA is cleaved, poly(A) polymerase adds a poly(A) tail to the upstream product, and the downstream RNA remaining attached to polymerase helps promote termination through mechanisms involving RNA degradation, changes in polymerase state, and termination factors. [Chapter 29](chapter1028.md) treats cleavage and polyadenylation in detail.

The important point for [Chapter 15](chapter1014.md) is architectural: a Pol II gene's annotated 3′ end, the cleavage site, the polyadenylated mature RNA end, downstream transcription, and the polymerase termination zone can occupy different positions. A transcription unit therefore extends beyond the mature mRNA in many cases. Mediator can regulate transcriptional termination through crosstalk with pre-mRNA 3′ end processing factors, connecting enhancer/coactivator logic with transcript-end formation.

**Table 15.4. Termination Mechanisms.** Overview of major RNA polymerase termination mechanisms, key molecular features, and their consequences for RNA product ends.

| System | Terminator feature | Proteins/factors | RNA product consequence | Example | Caveat |
| --- | --- | --- | --- | --- | --- |
| **Bacterial intrinsic** | GC-rich RNA hairpin followed by U-rich tract | None required; RNA structure-driven | RNA released; defined 3ʹ end near hairpin | Trp operon attenuator | Efficiency depends on RNA folding rate relative to elongation speed |
| **Bacterial factor-dependent (Rho)** | C-rich rut site in RNA | Rho helicase; NusG | RNA released by Rho-driven translocation along RNA | Rho termination at many E. coli genes | Translating ribosomes can protect mRNA and block Rho access |
| **Phage antitermination** | nut site RNA element; modified elongation complex | N protein; NusA; NusB; NusE; NusG | Polymerase continues past intrinsic terminators | Phage λ N antitermination | Requires nut site and full Nus factor assembly; context-specific |
| **RNA Pol II polyadenylation-coupled** | Poly(A) signal (AAUAAA); downstream sequence elements | CPSF; CstF; cleavage factors; Mediator | RNA cleaved at poly(A) site; tail added; Pol II terminates downstream | Most protein-coding pre-mRNAs | Mature 3ʹ end and polymerase release point differ; Mediator links coactivator and termination logic |
| **RNA Pol III oligo(dT)** | Run of T residues in non-template strand | None required (intrinsic-like mechanism) | Short U-rich 3ʹ end; defined transcript boundary | tRNA and 5S rRNA 3ʹ ends | Short T-run sufficient; context of flanking sequence can influence efficiency |
| **Organellar** | Variable; often processing-site-coupled | Organelle-specific endonucleases; RNA stability factors | Mature RNA ends set by cleavage and trimming rather than by a canonical termination event | Mitochondrial rRNA and tRNA end formation | Termination and end formation are mechanistically separable; mature ends require downstream processing |
| **Viral** | Viral-specific poly(A) signals or intrinsic signals | Viral or host factors depending on genome type | Defined transcript ends for each temporal class; compact overlapping TUs | Herpesvirus poly(A) sites; retroviral LTR polyadenylation | Viral TUs can overlap; termination efficiency affects expression of neighboring viral genes |

RNA polymerase I and III termination use different mechanisms. RNA polymerase III often terminates at short oligo(dT) tracts in the non-template DNA that produce U-rich RNA, while RNA polymerase I termination is specialized for rDNA repeat transcription. Pol I-specific transcription-factor reviews and Pol III transcription machinery sources support the chapter-level point that promoter and terminator logic differs by polymerase system.

### Readthrough

Transcriptional readthrough occurs when polymerase continues beyond an expected terminator or annotation boundary. Readthrough production extends the primary transcription unit into downstream genes, regulatory elements, antisense regions, or repeats. The extended product may encounter alternative processing or surveillance, but those fates belong to [Chapter 31](chapter1030.md) and the processing chapters. A direct regulatory role for the product belongs to [Chapter 93](chapter1088.md). Apparent readthrough can also arise from incomplete annotation or technical artifacts that fail to distinguish nascent from processed RNA.

An example from imatinib-treated chronic myeloid leukemia cells places readthrough before altered downstream processing in a specific disease and drug context. The architectural conclusion is that the termination boundary changed. Whether the extended products are stabilized, removed, or causal for later phenotypes requires the separate analyses owned by [Chapter 31](chapter1030.md) and [Chapter 93](chapter1088.md).

Evidence for readthrough should include strand-specific signal across the expected boundary, transcript-end data showing altered termination or 3′ processing, nascent RNA evidence if the claim concerns ongoing transcription, and perturbation of termination or processing factors when possible. Long-read RNA sequencing can help connect upstream and downstream regions in the same molecule, but it must be interpreted with controls for template switching, internal priming, incomplete cDNA synthesis, and read alignment artifacts.

### Antisense Transcription and Boundary Interpretation

Antisense transcription is transcription from the DNA strand opposite a reference gene. It can arise from an independent promoter, bidirectional promoter activity, readthrough from another gene, enhancer transcription, repeat-derived promoters, viral transcription, or pervasive low-level initiation. The architectural task is to identify the strand-specific initiation site and distinguish independent initiation from readthrough or mapping ambiguity. Low abundance and rapid disposal can obscure the product, but surveillance mechanisms belong to [Chapter 31](chapter1030.md).

The evidence standard is therefore conservative. Opposite-strand RNA reads establish antisense production only if the assay is strand-specific and mapping is reliable. Start-site mapping, nascent transcription, promoter perturbation, and exclusion of upstream readthrough can localize antisense initiation. Whether the antisense RNA product regulates the sense locus is a different causal question handed to [Chapter 93](chapter1088.md).

> **Box 15.4. Readthrough Can Be Biology or Artifact**
>
> A readthrough claim is stronger when supported by:
> - Strand-specific data confirming opposite-strand signal is absent.
> - Transcription-end mapping showing altered termination or poly(A) site use.
> - Nascent RNA evidence that the downstream signal is actively produced, not a stable processed product.
> - Long-read molecule continuity connecting upstream and downstream sequence.
> - Perturbation of termination or 3ʹ-end processing factors.
> - Controls for internal priming and alignment ambiguity near repetitive sequence.
>
> A readthrough claim is weaker when it relies only on low-level downstream RNA-seq signal near repetitive sequence.

A readthrough claim is stronger when supported by strand-specific data, transcription-end mapping, nascent RNA evidence, long-read molecule continuity, perturbation of termination or processing factors, and controls for internal priming or alignment ambiguity. A readthrough claim is weaker when it relies only on low-level downstream RNA-seq signal near repetitive sequence.

## 15.5. Genome organization as an RNA-production architecture

### From Gene Lists to Production Architecture

A genome is not just a list of genes. It is an arrangement of initiation regions, regulatory elements, coding and noncoding intervals, processing signals, terminators, repeats, mobile elements, chromatin states, spatial contacts, replication domains, and nuclear or nucleoid compartments. That arrangement affects RNA output. Two genomes with similar gene repertoires can produce different RNA landscapes if promoters, enhancers, terminators, repeats, and chromatin context differ.

**Table 15.5. Genome Architecture Features Affecting RNA Output.** Major architectural features of genomes that shape RNA production, with examples and the perturbation evidence needed to establish causality.

| Feature | RNA-production effect | Example | Evidence needed | Related chapter |
| --- | --- | --- | --- | --- |
| **Promoter** | Determines initiation site, timing, and level of transcription | Bacterial operon promoter; CpG-island Pol II promoter | TSS mapping; polymerase occupancy; factor binding; promoter mutation | Ch. 19–23 |
| **Enhancer** | Increases promoter output; often cell-type-specific and distance-independent in orientation | Drosophila developmental enhancers; mammalian super-enhancers | Reporter assay; enhancer deletion; factor binding; chromatin contact | Ch. 75, 85 |
| **Insulator** | Limits enhancer-promoter communication; defines topological domain boundaries | CTCF/cohesin boundaries in mammals | Deletion; reporter flanking assay; Hi-C contact change | Ch. 75, 85 |
| **Terminator** | Ends transcription; prevents interference with downstream units | Bacterial intrinsic terminators; poly(A) signals | Transcript-end mapping; readthrough assay; mutation of terminator elements | Ch. 28 |
| **Repeat or mobile element** | Donates promoters, terminators, splice sites, or antisense promoters; creates RNA-processing hazards | Alu elements causing processing defects suppressed by DHX9; LTR retrotransposon promoters | TSS mapping; RNA-processing phenotype; factor depletion; reporter in native context | Ch. 15, 96, 109 |
| **Nucleosome positioning** | Controls DNA accessibility at promoters and enhancers; influences pause duration | Nucleosome-depleted region at active promoters | MNase-seq; ATAC-seq; chromatin remodeler perturbation | Ch. 19, 20 |
| **Topologically associating domain (TAD)** | Constrains and focuses enhancer-promoter contacts within a genomic neighborhood | CTCF-anchored TADs in mammalian genomes | Hi-C; TAD boundary deletion; contact rewiring assay | Ch. 75, 85 |
| **Chromatin state** | Affects polymerase access, pausing duration, and elongation rate | Heterochromatin silencing; active euchromatin at housekeeping genes | ChIP-seq; chromatin remodeler depletion; reporter integration at different loci | Ch. 19, 20 |
| **Nuclear compartment** | Concentrates or excludes transcription machinery; influences RNA processing environment | Nucleolus for rDNA transcription; nuclear lamina for repressed loci | Imaging; lamin-B DamID; nucleolar fractionation; factor depletion | Ch. 16, 17 |
| **Operon organization** | Couples transcription initiation across multiple functionally related genes | Bacterial lac operon; trypanosome polycistronic gene arrays | Co-transcription evidence; continuous RNA detection; upstream promoter mutation | Ch. 14, 21 |

In bacteria, genome architecture includes operon order, promoter placement, terminator strength, nucleoid-associated protein binding, DNA supercoiling, replication orientation, and proximity between genes with related functions. In archaea, compact organization and operons coexist with eukaryote-related transcription machinery. In eukaryotes, chromatin architecture, enhancers, promoters, insulators, topologically associating domains, lamina association, nucleolar organization, and nuclear bodies all affect transcription probability and RNA processing environment. In organelles, compact genomes and processing-dependent transcription create architecture in which gene order and processing sites are inseparable. In viruses and mobile elements, compact architecture often creates overlapping signals and multifunctional RNA segments.

### Repeats and Mobile Elements as Regulatory Architecture

Repeats and mobile elements can change RNA production in several ways. They can donate promoters, enhancers, splice sites, polyadenylation sites, terminators, insulators, antisense promoters, and binding sites for transcription factors. They can create overlapping transcription, double-stranded RNA potential, R-loops, cryptic exons, and RNA processing hazards. They can also provide raw material for evolutionary innovation. [Chapter 16](chapter1015.md) treats these topics in depth, but the architectural principle belongs here: repeat-derived sequence can become part of the transcription-unit grammar.

Human Alu elements provide a concrete boundary example. Alu-rich regions can create initiation signals, overlapping transcription, and paired repeat-derived sequence that changes the RNA products generated from a locus. DHX9 suppresses processing defects originating from Alu expansion in the human genome. The production-architecture conclusion belongs here; the helicase, quality-control, and disposal mechanisms hand off to [Chapter 31](chapter1030.md) and the relevant processing chapters.

### Spatial Contacts and Causality

Eukaryotic enhancer-promoter contacts and chromatin domains are often mapped with chromosome conformation methods, microscopy, chromatin immunoprecipitation, and nascent RNA assays. These data are powerful but easy to overinterpret. A contact between an enhancer and promoter can support a regulatory model, but contact alone does not prove causality. A chromatin mark can correlate with enhancer or promoter activity, but the mark may be a consequence, a permissive feature, or part of a causal pathway depending on the locus. A transcription factor peak can indicate binding, but binding may be nonfunctional.

Causal architecture claims require perturbation. Deleting or mutating a promoter, enhancer, insulator, terminator, repeat, or chromatin-binding site can test necessity. Moving an element can test positional constraints. Acute degradation of a factor can separate direct from indirect effects. Time-resolved nascent RNA can reveal the order of events. Reporter assays can test sufficiency, but reporters may miss native chromatin, distance, 3D context, and RNA processing. A strong architectural claim integrates several evidence classes rather than relying on one map.

> **Box 15.5. Architecture Claims Need Perturbation**
>
> - Contact maps, chromatin marks, factor binding, and RNA abundance are usually correlative.
> - These data become stronger when paired with targeted sequence perturbation (deletion, mutation, or insertion of the element), factor depletion or acute degradation, time-resolved nascent transcription to establish event order, transcript-end mapping, and rescue experiments.
> - Architecture can explain RNA output only when the tested feature is linked to a causal change in RNA production or processing, not merely to a correlated change in a chromatin or contact signature.

Contact maps, chromatin marks, factor binding, and RNA abundance are usually correlative. They become stronger when paired with targeted sequence perturbation, factor depletion, time-resolved nascent transcription, transcript-end mapping, and rescue experiments. Architecture can explain RNA output only when the tested feature is linked to a causal change in RNA production or processing.

## Experimental Foundations and Evidence

Transcription start site mapping identifies where RNA synthesis begins or where capped RNAs are detected. In bacteria and eukaryotes, start-site methods help separate true promoters from motif predictions. However, start-site data can be influenced by RNA processing, RNA stability, capping status, decapping, degradation, template switching, and library-specific biases. A start peak is strongest when supported by promoter motifs, polymerase occupancy, factor binding, perturbation, and nascent transcription.

Nascent RNA sequencing measures newly synthesized RNA or RNA associated with engaged polymerase. It is particularly useful for promoter-proximal pausing, enhancer transcription, immediate transcriptional responses, antisense transcription, and readthrough. Nascent methods differ in what they capture: polymerase-associated RNA, metabolically labeled RNA, run-on products, capped nascent RNA, or chromatin-associated RNA. These differences matter. A method optimized for stable polyadenylated RNA will under-detect many promoter-proximal RNAs and enhancer RNAs, whereas a nascent method may detect transient transcription whose mature product is rapidly destroyed.

RNA polymerase occupancy methods, including chromatin immunoprecipitation-based and related approaches, show where polymerase or initiation factors are located. They do not by themselves prove productive RNA synthesis. A paused polymerase peak near a promoter, a polymerase peak at an enhancer, or a broad gene-body signal requires integration with RNA output and factor perturbation. Precise polymerase mapping has been especially informative for promoter-directed initiation and pausing.

Reporter assays test whether a DNA fragment can drive or enhance transcription in an experimental construct. Reporters are valuable because they allow systematic mutation and quantitative readout. Their limitation is that reporter context differs from native chromatin, genomic distance, local topology, neighboring promoters, endogenous terminators, and RNA processing. A DNA fragment that acts as an enhancer in a reporter may not regulate the predicted endogenous gene; a fragment that fails in a reporter may require native chromatin or long-range context.

Perturbation is the main path to causal architecture. Deleting a promoter can test whether it is required for initiation. Mutating a terminator can test readthrough. Deleting an enhancer can test target-gene dependence, although enhancer deletions can also alter local chromatin or remove promoter-like activity. CRISPR interference can repress a regulatory element without cutting DNA, but it can spread chromatin effects. Acute degradation of Mediator or elongation factors can reveal direct transcriptional consequences before secondary responses dominate. Experiments that distinguish a DNA element, the act of transcription, and the RNA product are developed as RNA-function tests in [Chapter 93](chapter1088.md).

Transcript-end mapping, long-read transcriptomics, and direct RNA sequencing help define transcription units and mature RNA boundaries. Their value is highest when they connect starts, internal exons, readthrough regions, and ends on individual molecules. Their limitations include incomplete coverage, internal priming, RNA degradation, reverse-transcription artifacts, base-calling errors, alignment ambiguity in repeats, and difficulty distinguishing co-transcriptional intermediates from mature RNAs. [Chapter 128](chapter1117.md) treats long-read and direct RNA technologies, while [Chapter 129](chapter1154.md) treats nascent and time-resolved RNA measurement.

## Biological Contexts Across Systems

Bacteria use promoter, operon, terminator, and RNA-structure logic to coordinate growth, stress, metabolism, virulence, phage response, and environmental adaptation. Their compact genomes make transcription-unit architecture especially visible. A promoter can initiate a polycistron; a terminator can prevent interference with the next operon; attenuation can change the unit boundary; and opposite-strand initiation can create an overlapping antisense product whose later fate or function is addressed elsewhere.

Archaea combine bacterial-like compact organization with transcription machinery related to eukaryotic systems. This combination makes archaea important for understanding evolutionary transitions in RNA production. Archaeal operons, small RNAs, CRISPR arrays, and basal initiation factors show that gene organization and transcription machinery do not always evolve as a single package.

Eukaryotic nuclear genomes rely heavily on chromatin and long-range regulation. A protein-coding gene may be controlled by several enhancers, alternative promoters, promoter-proximal pausing, alternative polyadenylation, antisense transcription, and neighboring noncoding transcription. Developmental genes can be especially enhancer-rich. Housekeeping genes often have different promoter architectures from highly cell-type-specific genes. Many lncRNA and enhancer-transcribed loci require production models that record initiation, direction, overlap, and boundaries rather than assuming one isolated gene-like unit.

Organelles demonstrate processing-centered architecture. Mitochondrial and plastid genomes often produce transcripts whose boundaries are shaped by downstream processing rather than one promoter and one terminator per mature RNA. Their RNA outputs depend on nuclear-encoded factors, organelle-specific polymerases, RNA editing in some lineages, and RNA stability factors. This is why [Chapter 17](chapter1016.md) is adjacent to [Chapter 15](chapter1014.md) in the book architecture.

![Figure 15.5. Genome Architecture as an RNA-Production System](../assets/figures/chapter1014_figure5.png)

**Figure 15.5. Genome Architecture as an RNA-Production System.** A genome functions as an RNA-production system in which promoters, enhancers, insulators, terminators, chromatin domains, topological contacts, repeat-derived regulatory elements, and nuclear or nucleoid compartments determine which primary RNAs are made, where, and in which direction. Repeats and mobile elements can donate promoters, termination signals, and antisense initiation sites. Perturbation of an architectural component can alter the nascent RNA landscape, showing that production is a property of genomic arrangement rather than gene sequence alone.

Viruses, phage, viroids, and mobile elements show compact and opportunistic RNA-production strategies. Phage can encode antitermination systems to redirect host polymerase. Retroviruses use promoter and enhancer logic in long terminal repeats. RNA viruses may produce subgenomic RNAs or replication intermediates rather than ordinary DNA-derived transcription units. Mobile elements can insert new promoter or terminator signals into host genomes. These systems force annotation to record mechanism and life-cycle context rather than forcing every RNA into a standard cellular gene model.

## Technology, Computational, and Engineering Links

Promoter prediction uses sequence motifs, conservation, chromatin state, transcription factor binding, nucleosome positioning, DNA shape, CpG content, machine-learning models, and start-site evidence. The main output of a promoter model should be a probability or candidate annotation, not a functional conclusion. A promoter model trained in one organism or cell type may fail in another because the relevant factors, chromatin state, and polymerase system differ.

Enhancer prediction faces a similar problem. Candidate enhancers can be identified by chromatin accessibility, histone marks, transcription factor binding, coactivator occupancy, evolutionary conservation, enhancer RNA production, and reporter activity. None of these features alone proves endogenous target-gene regulation. Computational enhancer-promoter linking uses distance, chromatin contacts, co-activity across cell types, perturbation data, and sequence features. Its major failure mode is confusing correlation with causality.

Transcript-unit annotation integrates short-read RNA-seq, long-read RNA-seq, cap analysis, poly(A)-site mapping, nascent RNA, splice junctions, transcript-end data, and genome annotation. This integration is difficult when transcription is pervasive, overlapping, antisense, repeat-rich, or condition-specific. [Chapter 18](chapter1017.md) covers transcript models and annotation versioning; [Chapter 141](chapter1128.md) covers transcriptome assembly and isoform quantification.

Synthetic biology uses promoters, enhancers, terminators, insulators, ribosome-binding sites, untranslated regions, and polycistronic cassettes as engineering parts. The chapter's main caution applies directly: parts are context-dependent. A strong promoter in one organism or genomic position may be weak in another. A terminator that works in a plasmid may fail in a chromosome. An enhancer tested in a reporter may not behave predictably after integration. Synthetic RNA circuits and engineered expression systems are covered in [Chapter 147](chapter1146.md).

Clinical and disease interpretation also depends on transcription-unit architecture. Noncoding variants can disrupt promoters, enhancers, insulators, terminators, splice-linked transcription, or polyadenylation. Cancer cells can show enhancer rewiring, promoter switching, readthrough transcription, enhancer hijacking, and mobile-element activation. However, disease association is not mechanism. A variant near an enhancer or a readthrough RNA in a tumor requires causal evidence before it is interpreted as a driver.

## Recent Consensus

- Promoters are functional initiation regions interpreted by specific polymerase systems, not universal upstream motifs.
- Promoter output depends on initiation, start-site choice, promoter escape, pausing, re-initiation, chromatin context, and regulatory inputs.
- Operons and polycistronic transcription coordinate initiation across multiple products, but mature RNA and protein output can be decoupled by processing, translation, and decay.
- Enhancers regulate promoters through transcription factors, cofactors such as Mediator, chromatin accessibility, topology, promoter-proximal pausing, and dynamic contacts rather than through one universal physical model.
- Enhancer transcription is common in active regulatory regions, but enhancer DNA activity, enhancer transcription, and enhancer RNA function are distinct claims.
- Termination is an active regulated process; readthrough, antitermination, and alternative 3′ end formation can change transcription-unit boundaries.
- Antisense transcription is common enough that strand-specific initiation and boundary mapping are essential; whether the product has regulatory function is handed to [Chapter 93](chapter1088.md).
- Genome architecture shapes RNA output by arranging initiation regions, regulatory elements, processing signals, terminators, repeats, chromatin, and spatial contacts.

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

Open questions:

- How often do enhancer regions use distinct initiation sites, directions, and promoter partners across cell states, and how should those changing production units be annotated?
- Which enhancer-promoter communication models dominate at different time scales, loci, cell types, and developmental stages?
- How should annotations represent transient promoter-proximal RNAs, regulated readthrough, unstable antisense transcripts, and condition-specific transcription units?
- How often are repeat-derived regulatory elements adaptive, neutral, deleterious, or context-dependent?
- How should computational models distinguish direct promoter or enhancer causality from chromatin correlation?

Common misconceptions:

- "A promoter motif is a promoter." A motif is a candidate feature; promoter function requires initiation evidence in context.
- "All promoters are upstream of genes." Some promoters are internal, bidirectional, embedded in repeats, organellar, viral, or arranged in compact overlapping architectures.
- "An operon means equal protein expression." Processing, RNA structure, translation initiation, translation coupling, and decay can decouple products.
- "Enhancer transcription proves enhancer RNA function." The DNA element, the act of transcription, and the RNA molecule are separate mechanistic objects.
- "Antisense transcription automatically regulates the sense gene." Antisense regulation requires perturbation and strand-aware evidence.
- "Readthrough is always an artifact." Readthrough can be regulated or disease-associated, but artifact controls are essential.
- "Chromatin contact proves enhancer target identity." Contact supports a model but causality requires perturbation and transcriptional readout.

Deprecated or weakened claims:

- Several older simplifications should be treated as deprecated teaching shortcuts. "One gene, one promoter, one transcript" is useful for an introductory diagram but misleading for operons, organelles, eukaryotic alternative promoters, overlapping transcripts, enhancer transcription, antisense transcription, and viral genomes. "Enhancers loop to promoters" is too narrow if presented as a universal mechanism, because enhancer-promoter communication includes transient contacts and hub-like models. "Termination is where the mature RNA ends" is wrong for many RNA polymerase II genes because cleavage and polymerase release can be separated.