Eukaryotic RNA Polymerases I, II, and III
Why does a human cell need three separate enzymes to do the job that one bacterial RNA polymerase manages alone? The answer lies in the sheer diversity of RNA products a eukaryotic nucleus must generate. Each RNA product has wildly different abundance, processing needs, and regulatory logic. A single liver cell makes millions of copies of ribosomal RNA per generation. Yet it makes only a few hundred copies of most messenger RNAs. It must also shut down protein synthesis genes without touching the ribosome factory, or vice versa. Bacteria solve this with one core polymerase and swappable sigma factors. Eukaryotes instead built three structurally related but functionally dedicated polymerases. Each recognizes its own promoters, its own factors, and its own class of gene. This division of labor is one of the deep organizing principles of eukaryotic gene expression. It was discovered not by inspection but by a clever pharmacological trick. That trick was the differential sensitivity of the three enzymes to the mushroom toxin α-amanitin. That single reagent, more than any structural study, first proved something important. It showed that "RNA polymerase" in eukaryotes is not one enzyme but three. Understanding this division matters for an important reason. It explains why cells can independently tune ribosome biogenesis, mRNA output, and small structural RNA supply. These are three processes with very different growth-rate dependencies and disease relevance. RNA Pol I hyperactivity is a hallmark of cancer. Pol II is the target of most transcriptional regulation biology teaches. Pol III output tracks cell growth and is dysregulated in tumors. This article covers several key topics. These include the identity of the three polymerases and the RNA classes they make. It also covers the promoter architectures and transcription-factor logic that route each gene to the correct enzyme. Finally, it covers the classical amanitin experiments that established the division. It also covers the subunit/structural relationships connecting all three back to a common ancestral machine shared with archaea.
- Eukaryotes have three nuclear RNA polymerases — Pol I, Pol II, Pol III — each dedicated to a distinct class of RNA.
- Pol I makes the large pre-rRNA (28S, 18S, 5.8S) in the nucleolus; Pol II makes mRNA and most regulatory/lncRNAs; Pol III makes tRNA, 5S rRNA, and other small RNAs.
- The three enzymes were distinguished experimentally by their differential sensitivity to α-amanitin (Pol II most sensitive, Pol III intermediate, Pol I resistant).
- Each polymerase has a unique promoter grammar and a private set of general transcription factors that recruit it — Pol II alone uses TFIIH and CTD phosphorylation for capping/splicing/termination coupling.
- All three share five common core subunits and a two-lobed, DNA-clamping architecture inherited from a common ancestor also found in archaeal RNA polymerase.
- Because Pol I and Pol III output scales with growth rate and ribosome demand, both are frequent targets of oncogenic signaling and cancer therapy.
Eukaryotic cells split transcription across three specialized polymerases — Pol I for ribosomal RNA, Pol II for messenger RNA, Pol III for transfer and small RNAs — each defined by a unique promoter, factor set, and α-amanitin sensitivity.
The core idea
In bacteria, one RNA polymerase core enzyme, redirected by different sigma factors, transcribes every gene in the genome. Eukaryotes instead evolved three paralogous, multi-subunit polymerases — RNA Polymerase I (Pol I), RNA Polymerase II (Pol II), and RNA Polymerase III (Pol III) — that partition the transcriptome by RNA class rather than by regulatory signal. Each enzyme is a distinct 10–17-subunit complex, built around a conserved catalytic core but wrapped in enzyme-specific subunits that dictate which promoters it can recognize and which transcription factors it can be handed off to. The division is not arbitrary. It tracks a fundamental split in the logic of gene expression: genes whose product is the functional RNA itself (ribosomal RNA, transfer RNA) versus genes whose transcript is a template for further translation (messenger RNA) versus a residual class of small, abundant, structural or regulatory RNAs. Pol I is essentially a single-purpose ribosomal RNA factory, running from a handful of near-identical gene copies (in humans, roughly 200–400 copies of the rDNA repeat, though many are transcriptionally silent) at extraordinarily high rates in the nucleolus. Pol II is the generalist and regulatory workhorse, transcribing all protein-coding genes plus most long non-coding RNAs and microRNA precursors, and its promoters and enhancers form the primary substrate of combinatorial gene regulation. Pol III specializes in short, highly structured, high-copy-number products — tRNAs, 5S rRNA, U6 snRNA, and other small RNAs — often using promoter elements located inside the transcribed gene itself, a structural quirk unique among the three. The three-polymerase system rests on prerequisite machinery you have already met: the basic transcription cycle of initiation, elongation, and termination, and the contrast with the single-enzyme, sigma-factor logic of bacterial transcription. What eukaryotes add is enzyme identity itself as an axis of regulation — which polymerase is recruited becomes as important a decision as which transcription factors bind.
Mechanism
Shared architecture, private accessories. All three eukaryotic polymerases are large, two-lobed enzymes ("crab-claw" shaped) that clamp around double-stranded DNA, melt a transcription bubble of roughly 12–14 base pairs, and thread the DNA:RNA hybrid through an internal channel toward a catalytic center built from two Mg²⁺ ions coordinated by a conserved aspartate triad. This core structure, and five subunits that are literally shared or near-identical across all three (often labeled RPB5, RPB6, RPB8, RPB10, RPB12 in Pol II nomenclature, with orthologs in Pol I and Pol III), is homologous to the single RNA polymerase of archaea and to the bacterial core enzyme — direct structural evidence, confirmed by X-ray crystallography (Kornberg and colleagues solved the Pol II structure at atomic resolution in 2001), that all cellular RNA polymerases descend from one ancestral enzyme.
Pol I: the ribosomal RNA factory. Pol I is a 14-subunit enzyme found almost exclusively in the nucleolus, where it transcribes a single long precursor, the 47S (in humans) or 45S pre-rRNA, from tandem rDNA repeats. This precursor is processed into the mature 28S, 18S, and 5.8S rRNAs that form the catalytic core of ribosomes. Pol I initiation requires UBF (upstream binding factor), which bends the rDNA promoter, and SL1/TIF-IB, a complex containing the TATA-binding protein (TBP) plus Pol I-specific TAFs, which positions the enzyme precisely at the transcription start site. Because a cell may need tens of millions of ribosomes, Pol I devotes its promoter architecture (an upstream control element plus a core promoter) entirely to maximizing throughput rather than combinatorial regulation — up to roughly 60% of total cellular transcription by mass in a rapidly growing cell is Pol I output.
Pol II: the messenger and regulatory engine. Pol II is a 12-subunit enzyme distinguished by its C-terminal domain (CTD) on the largest subunit (RPB1), a repeated heptapeptide (Tyr-Ser-Pro-Thr-Ser-Pro-Ser) present in 52 tandem copies in humans. Differential phosphorylation of this CTD by kinases such as CDK7 (part of TFIIH) and CDK9 (part of P-TEFb) creates a "CTD code" that times the recruitment of the mRNA capping enzyme, splicing factors, and 3′-cleavage/polyadenylation machinery to specific stages of the transcription cycle. Pol II promoters are the most architecturally diverse of the three: core promoter elements (TATA box, initiator, downstream promoter element) combine with distal enhancers and silencers bound by sequence-specific activators and repressors, integrated through the Mediator coactivator complex. This diversity is the physical substrate for the combinatorial control that makes different cell types express different genes from an identical genome.
Pol III: small, structured, and internally promoted. Pol III is the largest of the three (17 subunits) and transcribes the shortest, most abundant products: the roughly 76-nucleotide tRNAs, the 120-nucleotide 5S rRNA, U6 snRNA, 7SL RNA (of the signal recognition particle), and several other small regulatory RNAs. Remarkably, most Pol III promoters — type 1 (5S rRNA) and type 2 (tRNA) — lie within the transcribed sequence itself, built from internal control regions called box A and box C (type 1) or box A and box B (type 2), recognized by the assembly factor TFIIIC, which then recruits TFIIIB (containing TBP) to position Pol III just upstream of the start site. A separate type 3 promoter (used by U6 snRNA and other genes) sits entirely upstream in the more conventional style. Pol III also uses a distinctive termination signal: a simple run of four or more thymidines (T) on the non-template strand (equivalently, a poly-U tract in the nascent RNA), sufficient on its own to destabilize the elongation complex — the simplest, most compact termination mechanism of any eukaryotic polymerase, and notably reminiscent in spirit (though not sequence) of intrinsic bacterial termination.
How the correct polymerase finds its gene. Promoter recognition is entirely factor-mediated, not intrinsic to the polymerase itself. None of the three polymerases binds DNA specifically on its own; each is delivered to its promoter by a private set of general transcription factors, which read out the promoter grammar (upstream elements for Pol I and most Pol II genes, internal elements for classic Pol III genes) and then dock the correct polymerase onto the DNA. This modularity — DNA sequence → assembly factors → correct polymerase — is what allows the same nuclear enzyme pool to be targeted with pinpoint specificity to tens of thousands of genes, and it foreshadows the detailed factor biology of preinitiation complex assembly.
The evidence
The α-amanitin experiments (Roeder and Rutter, 1969–1970). The existence of three distinct polymerases was established not by structural biology (which came decades later) but by a pharmacological differential-sensitivity assay. α-Amanitin is a bicyclic octapeptide toxin produced by the death cap mushroom (Amanita phalloides), which binds tightly in the funnel/bridge-helix region of the largest polymerase subunit and blocks the translocation step of elongation. Robert Roeder and William Rutter (1969, extended 1970) fractionated nuclear extracts from rat liver, sea urchin, and other eukaryotes by ion-exchange chromatography and assayed each fraction's ability to incorporate labeled nucleotides into RNA in vitro, then tested how much α-amanitin was needed to inhibit each activity. They found three chromatographically separable activities with dramatically different drug sensitivities:
| Enzyme | α-Amanitin sensitivity | Concentration for 50% inhibition |
|---|---|---|
| Pol I | Resistant | > 1 mg/mL (essentially insensitive) |
| Pol II | Extremely sensitive | ~0.01–0.02 µg/mL |
| Pol III | Intermediate | ~10–20 µg/mL |
This roughly thousand-fold range of sensitivities was the decisive proof: a single enzyme cannot show three quantitatively distinct inhibition curves, so eukaryotic cells must contain (at least) three separate RNA polymerases. Roeder and Rutter further showed that the resistant activity co-purified with, and localized to, the nucleolus (consistent with rRNA synthesis), while the highly sensitive activity correlated with nucleoplasmic, rapidly labeled RNA (consistent with heterogeneous nuclear RNA, the mRNA precursor) — linking each biochemical activity to a specific RNA product even before the gene-class assignments were confirmed independently.
In vivo confirmation. Subsequent experiments injected α-amanitin into cultured cells and into Xenopus oocyte nuclei, then measured which labeled RNA species disappeared. Low doses selectively abolished heterogeneous nuclear RNA/mRNA synthesis while leaving rRNA and 5S RNA/tRNA synthesis intact; higher doses additionally blocked 5S rRNA and tRNA synthesis; nucleolar (45S/47S pre-rRNA) synthesis persisted even at the highest doses tested. This dose-response mapping — reproduced across many species from yeast to mammals — confirmed that the three chromatographic activities correspond to the three functionally distinct classes of nuclear transcription, and it remains the standard operational assay biologists use even today to ask "which polymerase transcribes this gene": add a Pol II-selective dose of α-amanitin and see whether the RNA of interest disappears.
Confirming distinct genes and promoters (in vitro reconstitution, 1980s). A second, complementary line of evidence came from Robert Roeder's and Pieter Sollner-Webb's laboratories reconstituting accurate, promoter-specific transcription in cell-free extracts. By purifying rDNA templates, tRNA gene templates, and protein-coding gene templates separately, and supplying each with crude nuclear extract, researchers showed that correctly initiated, full-length transcripts of the expected class appeared only when the matching polymerase and its cognate factors (UBF/SL1 for Pol I; TFIID/TFIIB/TFIIE/TFIIF/TFIIH for Pol II; TFIIIA/B/C for Pol III) were present — directly demonstrating promoter-selective, factor-dependent recruitment rather than any intrinsic sequence preference of the polymerases themselves. This established the modular "assembly factor → correct polymerase" logic described above and set up the entire subsequent research program on general transcription factors.
Worked examples
Example 1 — Diagnosing a mystery RNA by drug sensitivity. A researcher isolates a novel 300-nucleotide non-coding RNA from HeLa cells and wants to know which polymerase transcribes it, without sequencing its promoter. She treats cells with three concentrations of α-amanitin and measures the RNA's synthesis by nuclear run-on assay (pulse-labeling nascent transcripts with radioactive UTP), alongside known markers: pre-rRNA (Pol I), a housekeeping mRNA (Pol II), and a tRNA (Pol III).
| α-Amanitin dose | Pre-rRNA | Housekeeping mRNA | tRNA | Mystery RNA |
|---|---|---|---|---|
| 0 µg/mL | 100% | 100% | 100% | 100% |
| 0.02 µg/mL | 100% | 8% | 97% | 95% |
| 20 µg/mL | 96% | 3% | 15% | 12% |
| 1000 µg/mL | 40% | 2% | 5% | 6% |
Reasoning: At 0.02 µg/mL, only Pol II is knocked down (mRNA drops to 8%), and the mystery RNA is unaffected (95%) — ruling out Pol II. At 20 µg/mL, Pol III activity collapses (tRNA falls to 15%), and the mystery RNA falls in parallel (12%) — matching the tRNA curve rather than the pre-rRNA curve, which stays nearly full (96%) until the very highest, near-toxic dose. The concordance between the mystery RNA's inhibition curve and the tRNA curve — not the pre-rRNA curve — identifies it as a Pol III transcript, consistent with it being, for example, a small structured RNA like a U6-related species or a novel SINE-derived transcript. This is exactly the logic Roeder and Rutter's original experiments established as a general assay.
Example 2 — Estimating relative transcriptional output. A growing HeLa cell has roughly 10 million ribosomes and needs to replace ~1% of them per hour to sustain growth, while making roughly 2×10⁵ new mRNA molecules per hour distributed over ~15,000 expressed genes, and roughly 2×10⁶ tRNA molecules per hour to keep pace with translation demand. Estimate the relative nucleotide-incorporation burden on each polymerase, given: a mature ribosome contains ~7,200 nucleotides of rRNA (28S ≈ 5,000 nt, 18S ≈ 1,900 nt, 5.8S ≈ 160 nt, plus spacers roughly doubling raw transcript length to ~13,000 nt for the 47S precursor); an average mature mRNA is ~2,000 nt; a tRNA is ~76 nt.
- Pol I load: 10⁷ ribosomes × 1%/hr = 10⁵ new ribosomes/hr, each requiring one ~13,000-nt pre-rRNA transcript → 1.3 × 10⁹ nucleotides/hr.
- Pol II load (mRNA only): 2×10⁵ transcripts/hr × 2,000 nt ≈ 4 × 10⁸ nucleotides/hr.
- Pol III load (tRNA only): 2×10⁶ transcripts/hr × 76 nt ≈ 1.5 × 10⁸ nucleotides/hr.
Interpretation: Even though Pol II transcribes vastly more distinct genes (~15,000 vs. essentially one rDNA repeat type for Pol I), Pol I's nucleotide throughput is more than three times that of Pol II and roughly nine times that of Pol III in this estimate — quantitatively explaining the classic observation that Pol I accounts for the majority of all nuclear transcription by mass in a proliferating cell, and why nucleolar size and rDNA transcription rate are used clinically as prognostic markers of tumor growth rate (large, prominent nucleoli in cancer histopathology reflect elevated Pol I output).
Common misconceptions & traps
| Misconception | Correction |
|---|---|
| "Eukaryotes have one RNA polymerase, like bacteria." | They have three distinct nuclear enzymes (Pol I, II, III), plus separate polymerases in mitochondria/chloroplasts. |
| "Pol II makes all RNA that isn't mRNA-untranslated." | Pol II also makes most lncRNAs, miRNA precursors, and many snRNAs — but not tRNA, 5S rRNA, or the large pre-rRNA. |
| "α-Amanitin sensitivity tells you the RNA's function." | It tells you the polymerase identity only; sensitivity is a biochemical fingerprint, not evidence about the RNA's biological role. |
| "Pol III promoters always sit upstream like Pol II promoters." | Classic Pol III promoters (type 1, type 2) are internal, embedded within the transcribed sequence itself. |
| "More Pol I activity always means faster growth is good/adaptive." | Elevated Pol I activity is a marker of proliferation and is exploited pathologically in many cancers — it is not inherently beneficial to the organism. |
| "The three polymerases are unrelated enzymes that evolved independently." | They are paralogs sharing five core subunits and a common ancestral architecture also found in the single archaeal RNA polymerase. |
The scientist behind it
Origins · 1 minRobert G. Roeder and how this was discovered
Robert Roeder trained at the University of Washington under William Rutter and, as a graduate student in the late 1960s, undertook the biochemical fractionation of nuclear extracts that would resolve eukaryotic transcription into three separable polymerase activities. Rather than assuming a single bacterial-like enzyme, Roeder systematically chromatographed nuclear extracts and characterized each peak's ionic strength requirements, subunit composition, and — decisively — its response to α-amanitin. He went on to purify and characterize the general transcription factors (TFIIA through TFIIH) required for accurate Pol II initiation and later the Pol III factors TFIIIA/B/C, essentially founding the biochemical, reconstitution-based study of eukaryotic transcription as a discipline. Roeder received the Albert Lasker Basic Medical Research Award in 2003 for this body of work.
Where this matters today
Cancer biology and therapy. Elevated Pol I transcription is a near-universal feature of cancer cells, driven by oncogenic MYC and RAS-pathway signaling that upregulates UBF and SL1 activity; drugs like CX-5461, which selectively inhibits Pol I initiation, are in clinical trials as a strategy to starve tumor cells of ribosome biogenesis capacity while sparing normal cells with lower Pol I demand. Enlarged, hyperactive nucleoli remain a textbook histopathological hallmark of malignancy, a direct visible correlate of increased Pol I output.
Toxicology and mushroom poisoning. α-Amanitin, the very reagent that revealed the three-polymerase system, is the lethal toxin of Amanita phalloides (death cap) poisoning, one of the leading causes of fatal mushroom ingestion worldwide; its mechanism — irreversible Pol II inhibition leading to catastrophic loss of hepatocyte mRNA synthesis — directly explains the delayed, severe liver failure characteristic of this poisoning, and understanding Pol II's vulnerability has informed antidote research (e.g., silibinin, which blocks hepatocyte uptake of the toxin).
Viral manipulation of Pol III. Several DNA tumor viruses and RNA viruses (including adenovirus VA RNA genes and Epstein–Barr virus EBER genes) hijack Pol III to produce abundant small RNAs that suppress host antiviral defenses (for example, by sequestering the interferon-induced kinase PKR), making Pol III regulation a target of interest in antiviral research.
Biotechnology and gene editing. Pol III promoters (particularly the U6 and H1 promoters) are the standard tool for expressing short guide RNAs in CRISPR–Cas9 systems and short hairpin RNAs for RNA interference, precisely because Pol III efficiently produces short, precisely 3′-end-defined transcripts without the capping and splicing machinery that Pol II transcripts require — a direct engineering payoff from understanding Pol III's distinct termination and processing logic.
Neurodegeneration and aging. Mutations in Pol III subunits or its assembly factors cause rare inherited leukodystrophies (e.g., 4H leukodystrophy, from POLR3A/POLR3B mutations), underscoring that tRNA and small-RNA supply, once thought to be a "housekeeping" background process, has tissue-specific developmental consequences, particularly in the developing nervous system.
Ask the tutor: How do the general transcription factors physically differ between the Pol I, Pol II, and Pol III preinitiation complexes, and why does only Pol II require a CTD-based coupling system?
Why do eukaryotes need three RNA polymerases instead of one, like bacteria?
Eukaryotic genomes support far more diverse transcriptional outputs than bacterial genomes, and each RNA class has different abundance, regulatory, and processing requirements. Ribosomal RNA must be made at enormous, relatively constant rates from few gene copies; tRNAs and small RNAs must be short, simple, and rapidly renewable; mRNAs require intricate combinatorial regulation and coupling to capping, splicing, and export. Rather than reusing one polymerase with swappable factors (the bacterial sigma-factor strategy), eukaryotes evolved dedicated enzymes whose intrinsic properties — promoter recognition partners, termination mechanisms, CTD-based coupling — are matched to each RNA class's needs. This also allows the cell to regulate ribosome biogenesis independently of protein-coding gene expression, a flexibility bacteria achieve differently.
How was α-amanitin used to distinguish the three polymerases, and why does it work?
α-Amanitin binds a conserved pocket near the bridge helix of the largest polymerase subunit and blocks the translocation step of nucleotide addition. Roeder and Rutter separated nuclear extracts by chromatography and tested each fraction's sensitivity, finding three activities requiring roughly 100,000-fold, ~1,000-fold, and negligible drug to achieve 50% inhibition — corresponding to Pol II, Pol III, and Pol I respectively. This worked because the binding pocket, though structurally conserved, differs subtly enough among the three paralogous largest subunits that their affinities for the toxin differ by orders of magnitude, providing a clean quantitative fingerprint for each enzyme.
Do Pol I, II, and III ever transcribe the same gene, or is the assignment absolute?
The assignment is essentially fixed per gene, because promoter recognition depends on which assembly factors (SL1 for Pol I, TFIID/TFIIB for Pol II, TFIIIB/C for Pol III) can bind that gene's specific promoter architecture — a given promoter sequence is generally compatible with only one polymerase's factor set. There are a few genuine exceptions and borderline cases (some genes near the boundary of Pol II/Pol III specificity, and a small number of genes such as certain snRNA genes that can be transcribed by either Pol II or Pol III depending on promoter type), but as a rule the polymerase-to-gene-class mapping is highly specific and stable.
What does the CTD of Pol II actually do mechanistically?
The C-terminal domain is an intrinsically disordered tail of tandem heptapeptide repeats (consensus Tyr-Ser-Pro-Thr-Ser-Pro-Ser) extending from the largest Pol II subunit. During the transcription cycle, different serine and tyrosine residues within these repeats are phosphorylated and dephosphorylated by specific kinases and phosphatases (notably CDK7 near initiation, phosphorylating Ser5, and CDK9 during elongation, phosphorylating Ser2). These phosphorylation marks act as a code, recruited specifically by the capping enzyme (which needs Ser5-P), splicing factors, and the cleavage/polyadenylation machinery (which favor Ser2-P), physically coupling transcription to co-transcriptional RNA processing in real time as Pol II moves along the gene.
Why is Pol I output considered a marker of cancer, and how is this exploited therapeutically?
Cell growth and proliferation require continuous ribosome production, and ribosome biogenesis is rate-limited by Pol I transcription of pre-rRNA. Oncogenic pathways (MYC amplification, RAS/MAPK activation, loss of tumor suppressors like RB and p53) converge on upregulating Pol I initiation factors such as UBF and SL1, producing the enlarged, prominent nucleoli long used by pathologists as a microscopic hallmark of malignancy. This dependency has been exploited therapeutically: small-molecule inhibitors of Pol I initiation (such as CX-5461) selectively stress cancer cells with high ribosome demand, and are in clinical development, particularly for lymphomas and certain solid tumors with MYC-driven biology.
How do Pol III promoters differ structurally between tRNA genes and U6 snRNA genes?
Classic Pol III genes like tRNA genes (type 2 promoters) and the 5S rRNA gene (type 1 promoter) carry their promoter elements internally, within the transcribed sequence — box A and box B for tRNA genes, box A and box C for the 5S gene — recognized by TFIIIC, which then recruits TFIIIB to set the correct start site upstream. In contrast, U6 snRNA and a few other genes use a type 3 promoter, which sits entirely upstream of the transcription start site in a more Pol II-like arrangement (including a TATA box), recruiting TFIIIB directly without requiring TFIIIC. This split shows that "Pol III promoter" is not one single architecture but at least three distinct grammars converging on the same enzyme.
Is the three-polymerase system universal across eukaryotes?
Yes, in essence — Pol I, Pol II, and Pol III (or clear orthologs) are found across the eukaryotic tree, from yeast to plants to animals, reflecting their presence in the last eukaryotic common ancestor. Plants additionally possess two more RNA polymerases, Pol IV and Pol V, specialized for producing small RNAs involved in RNA-directed DNA methylation and heterochromatin silencing — an evolutionary elaboration on the same underlying architecture, not a replacement of the core three-polymerase system.