Transcription: Overview and RNA Polymerase
A bacterium copies a single gene into hundreds of RNA transcripts within minutes. Each one is a disposable working copy, read by ribosomes and then degraded. Meanwhile, the master DNA blueprint stays locked in the archive, untouched. This asymmetry between a permanent original and disposable copies is the entire logic of transcription. Transcription is the process by which The Central Dogma: Chemical Logic of Information Flow moves information from DNA into RNA. The enzyme that performs this copying, RNA polymerase, is one of the most mechanistically instructive machines in biochemistry. It must find the right starting point on a genome of millions of base pairs. It reads one strand while ignoring the other. It synthesizes a chemically distinct polymer without ever needing a running start. And it must know precisely when to stop. Transcription matters because it is the first and most heavily regulated checkpoint in gene expression. Every cell in your body carries the same genome. Yet a neuron and a hepatocyte look and behave nothing alike. The difference is almost entirely which genes get transcribed, when, and how much. Transcription factors, hormones, chromatin marks, and small RNAs all converge on one molecular decision: does RNA polymerase bind this promoter and start moving? Understanding the physical chemistry of the polymerase itself is essential. This means grasping how it recognizes a template strand, why it needs no primer, and how nucleotides are added one at a time with high fidelity. This is the mechanistic foundation for everything built on top of it. It underlies cancer-driving mutations in transcription factors and antibiotics that jam bacterial RNA polymerase. The big picture is a story of asymmetric information transfer built from familiar chemistry. RNA polymerase performs a templated polymerization reaction chemically similar to DNA replication. This involves a nucleophilic attack by a 3′-OH on the α-phosphate of an incoming nucleotide, driven by pyrophosphate release. But it differs from DNA polymerase in three profound ways. It uses ribonucleotides instead of deoxyribonucleotides. It reads only one of the two DNA strands, the template strand, while the other strand (the coding/sense strand) is copied in name only. Most strikingly, it requires no primer at all. Each of these facts has a clean mechanistic explanation. This explanation is rooted in enzyme active-site chemistry and in the different fidelity requirements of a disposable RNA copy versus a permanent DNA genome. This article builds the physical picture of RNA polymerase as a molecular machine. It covers its multi-subunit architecture and the way it clamps around DNA, melting it into a transcription bubble. It explains the chemistry of the two-metal-ion mechanism for phosphodiester bond formation. It also resolves the primer paradox: why an enzyme that seems to need a free 3′-OH to extend, exactly like DNA polymerase, manages to start from scratch. The treatment here sets up later, more detailed accounts of promoter recognition, initiation, and termination. It also prepares the ground for the elaborate RNA processing that eukaryotic transcripts undergo after synthesis.
- RNA polymerase synthesizes RNA 5′→3′ by reading a DNA template strand 3′→5′, using ribonucleoside triphosphates (ATP, GTP, CTP, UTP) as substrates.
- The coding (sense) strand has the same sequence as the RNA transcript (with U for T); the template (antisense) strand is the one actually read by the enzyme.
- Bond formation uses a universal two-metal-ion mechanism: Mg²⁺ ions activate the 3′-OH nucleophile and stabilize the pyrophosphate leaving group.
- Unlike DNA polymerase, RNA polymerase requires no primer — it can initiate de novo because its active site stabilizes an initiating NTP without a covalent 3′-OH handle from a prior strand.
- Bacterial RNA polymerase is a single conserved multisubunit enzyme (core: α₂ββ′ω); eukaryotes use three specialized polymerases (Pol I, II, III) for different RNA classes.
- Fidelity requirements differ from replication: RNA is transient and made in many copies, so transcription tolerates a ~10⁴–10⁵-fold higher error rate than DNA replication.
- Key idea: RNA polymerase is a self-priming, strand-selective, two-metal-ion polymerase whose active-site chemistry — not any special "RNA-only" trick — explains why transcripts start without a primer.
Where this lives today
RNA polymerase is a first-line drug target and a diagnostic workhorse across medicine and biotechnology. The antibiotic rifampin (rifampicin), a cornerstone of tuberculosis treatment, binds directly in the DNA/RNA exit channel of bacterial RNA polymerase's β subunit, physically blocking extension of the nascent RNA past about 2–3 nucleotides — exploiting exactly the structural channel described here, and demonstrating why understanding polymerase architecture translates directly into rational antibiotic design. The mushroom toxin α-amanitin, historically crucial for distinguishing eukaryotic RNA polymerases I, II, and III (Pol II is exquisitely sensitive, Pol III is weakly sensitive, Pol I is resistant), remains a standard tool in molecular biology labs for dissecting which polymerase transcribes a given gene. Nucleoside analog drugs such as remdesivir target viral RNA-dependent RNA polymerases, which share the same two-metal-ion active-site chemistry described here for cellular RNA polymerase, showing how deeply conserved this catalytic mechanism is across all of biology, viral and cellular alike. In biotechnology, bacteriophage T7 RNA polymerase — small, single-subunit, extraordinarily processive, and requiring only a short 17-bp promoter — is the workhorse enzyme behind in vitro transcription reactions used to manufacture mRNA vaccines (including COVID-19 mRNA vaccines) and CRISPR guide RNAs at industrial scale, an application resting entirely on the primer-free initiation chemistry worked out in Derivation 1. Understanding transcription initiation, elongation rate, and termination is also the mechanistic backbone for interpreting genome-wide techniques such as ChIP-seq, PRO-seq, and single-molecule transcription assays that map RNA polymerase's real-time position and speed across entire genomes in living cells — techniques that, in turn, feed directly into the still-more-detailed pictures of Promoters, Initiation, and Termination of Transcription and of Eukaryotic RNA Processing: Capping, Splicing, Polyadenylation built on the foundation laid here.
Ask the tutor: How does σ factor actually recognize the −35 and −10 promoter elements, and what happens structurally during open complex formation?
The physical picture
RNA polymerase is best pictured as a crab-claw-shaped clamp that grips double-stranded DNA, prises the two strands apart over a short stretch, and slides along the genome reading one strand while extruding a growing RNA chain from an internal channel. The enzyme does not chew through the DNA covalently; it uses only noncovalent contacts and a swiveling clamp domain to hold the DNA in place under tension while the two strands are mechanically separated by roughly one helical turn's worth of base pairs — this open region is called the transcription bubble. Inside the bubble, 8–9 nucleotides of the nascent RNA remain base-paired to the template DNA strand as a transient RNA-DNA hybrid, which is peeled away and directed into an exit channel as new bases are added at the front (downstream) edge of the bubble. Bacterial RNA polymerase's core enzyme has the subunit composition α₂ββ′ω (roughly 400 kDa), and gains promoter-recognition ability only when joined by a dissociable σ factor to form the holoenzyme (a topic explored in more depth for initiation and promoters). The two large subunits β and β′ form the walls of the crab claw and together build the single active site at their interface, where three conserved aspartate residues in a signature "NADFDGD" motif chelate the catalytic magnesium ions. Eukaryotes have evolved three paralogous, more elaborate machines — RNA polymerase I (makes most ribosomal RNA), RNA polymerase II (makes messenger RNA and many regulatory RNAs), and RNA polymerase III (makes tRNA, 5S rRNA, and other small RNAs) — but all three retain the same two-large-subunit catalytic core and the same fundamental two-metal-ion chemistry as the bacterial enzyme, a conservation so deep it reflects a single evolutionary origin for templated RNA synthesis predating the divergence of the three domains of life. The chemistry of bond formation is identical in logic to DNA polymerase's, which is why it is useful to compare the two directly. An incoming ribonucleoside triphosphate (NTP) enters the active site and is checked for correct Watson–Crick pairing with the template base at the "insertion site." Two Mg²⁺ ions, held by the conserved aspartates, do the catalytic work: metal A lowers the pKa of the RNA 3′-OH and orients it for in-line nucleophilic attack on the α-phosphate of the NTP, while metal B stabilizes the developing negative charge on the β,γ-pyrophosphate leaving group and the pentacovalent transition state at phosphorus. The reaction is an SN2-like in-line displacement at phosphorus, releasing inorganic pyrophosphate (PPi), which is then hydrolyzed by pyrophosphatase to two phosphates — an essentially irreversible step that pulls the overall polymerization forward thermodynamically.
The core ideas
The template strand, read 3′→5′, dictates the RNA sequence read 5′→3′, by strict Watson–Crick–Franklin complementarity with U replacing T:
where: each template base pairs with the complementary incoming ribonucleotide (A↔U, T↔A, G↔C, C↔G), and the enzyme moves along the template in the 3′→5′ direction while synthesizing RNA 5′→3′.
where: this transcript is identical in sequence to the non-template (coding) DNA strand read 5′→3′, except U replaces T — a fact used constantly to read gene sequences directly off the coding strand without ever writing out the template. The catalytic step, common to RNA and DNA polymerases, is nucleotidyl transfer:
where: RNAₙ is the RNA chain of length n with a free 3′-OH, NTP is the incoming ribonucleoside triphosphate selected by base pairing with the template, and PPi is inorganic pyrophosphate released as the leaving group; two Mg²⁺ ions are obligatory cofactors. The overall reaction is driven forward by a large negative free energy change, principally because pyrophosphate is rapidly hydrolyzed by pyrophosphatase, removing product and making the net reaction essentially irreversible:
where: the first term (bond formation plus PPi release) is close to thermoneutral or mildly favorable on its own (~−1 to −4 kcal/mol depending on sequence context), while pyrophosphate hydrolysis contributes roughly −6.7 to −8 kcal/mol under cellular conditions, making the coupled, two-step process strongly exergonic overall. The number of RNA polymerase molecules actively transcribing a gene relates to promoter strength and elongation rate through a simple flux relationship useful for estimating transcript output:
is not the standard form; the cleaner and more commonly used relation is transcript output rate:
where: ρ is the linear density of actively elongating polymerases on the gene (molecules per unit length, limited by steric exclusion — a polymerase footprint is ~35 bp so ρ cannot exceed roughly 1 per 35 bp), and v is the elongation velocity (bacterial RNA polymerase: ~40–80 nucleotides per second in vivo; eukaryotic Pol II: ~25–50 nt/s).
Goal. Show mechanistically why DNA polymerase absolutely requires a primer with a free 3′-OH, while RNA polymerase can initiate synthesis de novo from two free NTPs, using only active-site chemistry — no appeal to "RNA is just different."
Step 1 — State the shared chemical requirement. Both enzymes catalyze the same nucleophilic substitution at phosphorus: a 3′-oxyanion attacks the α-phosphate of an incoming NTP (or dNTP), in-line, expelling pyrophosphate. This chemistry is identical in both enzymes and is why both use the two-metal-ion mechanism with conserved aspartates. The obligatory ingredient is a properly positioned 3′-OH nucleophile, activated by Mg²⁺-A, sitting immediately adjacent (in-line) to the α-phosphate of the incoming triphosphate positioned in the "insertion site" — a site defined by tight, sequence-independent contacts to the ribose-phosphate backbone plus positioning by the templating base.
Step 2 — Ask where that 3′-OH comes from at the very first bond. For the second nucleotide added and beyond, the 3′-OH belongs to the previous nucleotide already in the chain — no ambiguity there. The entire question of priming concerns only the first phosphodiester bond: something must occupy the "n−1" position (called the initiation site, or i-site) opposite the templating base, contributing a 3′-OH, before the "n" position (the i+1 site) can be filled by an incoming NTP.
Step 3 — Compare what each enzyme's initiation site can bind. DNA polymerase's initiation (n−1) site is built to bind only a primer terminus that is already base-paired to the template — the enzyme has essentially no separate "start from nothing" pocket; its clamp and fingers domains close around a pre-existing primer-template duplex and cannot stably position a lone free nucleotide there. RNA polymerase's active site, by contrast, has a bipartite structure with both an initiation subsite and an elongation subsite that can each, independently, bind a free NTP positioned by Watson–Crick pairing to the template — no primer terminus required. Structural and biochemical work (footprinting, and crystal structures of initiating complexes) shows RNA polymerase holds the first NTP (almost always a purine, ATP or GTP, at position +1) in the initiation subsite purely through template-directed base pairing plus core enzyme contacts, with its 3′-OH poised exactly where a primer's 3′-OH would sit in the analogous DNA polymerase complex.
Step 4 — Show the mechanism is otherwise unchanged. With NTP-1 sitting in the initiation site contributing its 3′-OH, and NTP-2 entering the adjacent elongation site paired to the next template base, the identical two-metal-ion in-line attack proceeds: Mg-A deprotonates/activates the 3′-OH of NTP-1, which attacks the α-phosphate of NTP-2, releasing PPi and forming the first phosphodiester bond, giving a dinucleotide with a free 3′-OH at its 3′ end. This dinucleotide's 3′-OH now occupies the elongation site's "primer-like" role for the next round, and the reaction becomes formally identical to normal elongation from that point on.
Step 5 — State why this asymmetry exists and its cost. De novo initiation is intrinsically inefficient: the very first phosphodiester bond forms far more slowly than subsequent ones, and the initiating dinucleotide is prone to falling off the template ("abortive initiation," in which RNA polymerase releases short 2–9 nucleotide RNAs repeatedly before achieving a stable elongation complex). DNA polymerase, needing far higher fidelity because its product is permanent and passed to daughter cells, evolved to outsource the error-prone act of first-bond formation entirely to a specialized low-fidelity primase (which synthesizes a short RNA primer) — a division of labor that RNA polymerase, whose product is transient and made in bulk, does not need.
Conclusion. No exotic RNA-specific chemistry is required. The primer-free initiation of transcription reduces to a single structural fact: RNA polymerase's active site contains an extra nucleotide-binding subsite (the initiation subsite) capable of positioning a free NTP's 3′-OH in-line for attack, a subsite that DNA polymerase's active site architecture lacks.
Go deeper · Derivation B · 2 minChoosing the template strand: how the enzyme knows which strand to read
Goal. Show that "template strand selection" is not a separate biochemical decision made by RNA polymerase gene-by-gene, but a direct geometric consequence of promoter orientation and enzyme directionality, fixed the moment the polymerase binds and orients on a promoter.
Step 1 — Recognize that both strands are chemically available. Double-stranded DNA is antiparallel; either strand could in principle serve as a template, since RNA polymerase's active-site chemistry (two-metal-ion nucleotidyl transfer) has no inherent strand preference — it will faithfully copy whichever single strand is threaded through its active-site channel in the 3′→5′ orientation.
Step 2 — Identify what actually fixes the choice: the promoter. The promoter is an asymmetric DNA sequence (in bacteria, the −35 and −10 elements recognized by σ factor; in eukaryotes, elements like the TATA box recognized by transcription factor complexes) that is inherently directional — it reads differently on the two strands and is only recognized in one specific orientation by the initiation machinery. Because the promoter is asymmetric, the polymerase (or its σ/initiation-factor partner) can bind it in only one productive orientation.
Step 3 — Follow the geometric consequence. Once σ factor (or the eukaryotic pre-initiation complex) is bound in its single correct orientation on the promoter, the DNA strand running 3′→5′ through the active-site cleft, in the downstream direction the enzyme will travel is automatically fixed as the template strand — there is no independent "reading" step in which the enzyme samples both strands and picks one. The clamp opens the DNA into a bubble at a fixed position relative to the bound σ factor, and the geometry of that opening presents one specific strand to the active-site channel.
Step 4 — Confirm with the asymmetry test. If you artificially invert a promoter sequence (flip which strand carries the −35/−10 consensus, e.g., by genetic engineering), transcription initiates on the opposite strand and produces the antisense-oriented transcript, and the gene is now read out in the reverse direction along the chromosome. This confirms that template choice is dictated entirely by promoter sequence and orientation, not by any strand-scanning search performed by the polymerase itself.
Step 5 — Note the biological consequence for gene density. Because template selection is fixed per-promoter and not per-molecule, neighboring genes on a chromosome can be transcribed from either strand independently; roughly half of human genes are transcribed from each strand, and convergent or divergent gene pairs (reading toward or away from each other) are common and functionally significant (e.g., in transcriptional interference and enhancer/promoter pairing).
Conclusion. Strand selection is a geometric readout of promoter asymmetry, locked in at the moment of polymerase (or transcription factor) binding — not a separate recognition event performed by the polymerase's catalytic core, which is strand-agnostic and simply copies whichever strand ends up threaded through it.
Worked examples
Example 1 — Foundational: writing the RNA transcript from a given template strand. Suppose the DNA template strand, read as given, is:
Write this explicitly with orientation labels to avoid the single most common transcription error: template strand, 3′→5′:
RNA polymerase reads this 3′→5′ (left to right as written) and synthesizes RNA 5′→3′, pairing A↔U, T↔A, C↔G, G↔C:
So the transcript is 5′-AUGGUCAAU-3′. As a check, the coding strand (complementary to the template, same polarity convention, written 5′→3′) must read 5′-AUGGUCAAT-3′ with T instead of U — indeed swapping U→T in the RNA gives exactly the coding strand, confirming self-consistency. Note this transcript begins with AUG, a start codon in the reading frame shown — a deliberately chosen non-round example to make the pairing arithmetic checkable base-by-base rather than falling into a symmetric or palindromic sequence that would hide an error.
Example 2 — Applied: estimating transcript yield from a moderately active bacterial gene. A bacterial operon is transcribed with an average of 1 new RNA polymerase initiating every 4.3 seconds (an asymmetric, measured-looking initiation interval, not a round number), and each polymerase elongates at 55 nucleotides per second once past the promoter-proximal pause. The gene is 2,850 nucleotides long. Estimate (a) the number of RNA polymerase molecules simultaneously transcribing the gene at steady state, and (b) the rate of complete-transcript production. Step 1: elongation time per transcript.
Step 2: number of polymerases simultaneously on the gene at steady state. Since a new polymerase initiates every 4.3 s and each occupies the gene for 51.8 s, by Little's Law (steady-state queueing: number in system = arrival rate × time in system):
Check this against the physical footprint limit: RNA polymerase occupies ~35 bp of DNA, so the maximum number that could physically fit on a 2850 nt gene is 2850/35 ≈ 81 — our estimate of 12 is comfortably below this steric ceiling, confirming the traffic density is physically plausible (not jammed). Step 3: transcript production rate. At steady state, the rate complete transcripts are finished equals the initiation rate (every initiated polymerase eventually finishes, assuming no premature termination):
This order of magnitude (tens of transcripts per minute for a moderately strong promoter) matches measured values for actively expressed bacterial operons such as the lac operon under inducing conditions.
Example 3 — Exam/competition level: distinguishing abortive initiation kinetics from productive elongation using a pulse-chase argument. An in vitro transcription reaction is assembled with RNA polymerase holoenzyme, a linear DNA template containing a single strong promoter, and all four NTPs, but ATP is radiolabeled (³²P at the α-phosphate, so label is incorporated into every internal phosphodiester bond, not just the 5′ end). The reaction is sampled at 2, 10, 30, and 120 seconds. At 2 s, the gel shows a ladder of short labeled RNAs 2–8 nucleotides long, at roughly constant total intensity across the four time points but with the fraction in the 2–8 nt range dropping from ~90% (2 s) to ~15% (120 s), while a distinct band corresponding to full-length run-off transcript (600 nt) grows from undetectable at 2 s to dominant at 120 s. Explain this pattern mechanistically and estimate the approximate fraction of initiation events that are abortive versus productive at late time, given that the 2–8 nt species plateau at 15% of total labeled RNA while full-length product is 85%. Mechanistic explanation. Immediately after promoter binding and DNA melting, RNA polymerase sits in a stressed, promoter-bound "initially transcribing complex" that repeatedly makes short RNAs (2–8 nt is the classic abortive range) and releases them without ever escaping the promoter, because σ factor's continued contacts with the −10/−35 elements create scrunching stress in the transcription bubble that is mechanically incompatible with stable, long-range translocation. This is abortive initiation: the polymerase does not move away from the promoter; it synthesizes a short RNA, releases it, and re-initiates from scratch (fully consistent with the primer-free-initiation mechanism in Derivation 1 being repeated many times at the same promoter). Only when the RNA reaches a critical length (roughly 9–12 nt) does σ factor's grip loosen enough for the enzyme to break its promoter contacts entirely and transition into a stable, highly processive elongation complex — an event called promoter escape or promoter clearance. Once escape occurs, further short abortive products are no longer generated from that polymerase, and it proceeds to make full-length RNA at the elongation rate. Early time points (2 s) are therefore dominated by the transient, repeatedly-cycling abortive phase (because many polymerase molecules are still promoter-bound and cycling), while late time points (120 s) are dominated by successfully escaped polymerases that have completed full-length transcripts; the abortive fraction never disappears entirely because some polymerases continue unsuccessfully cycling at the promoter even at late times, or because abortive products, once released, are chemically stable and simply persist as a fixed-size pool while new full-length product accumulates around them. Quantitative estimate. If at 120 s the abortive (2–8 nt) pool has plateaued at 15% of total labeled RNA (by radioactivity, meaning by total incorporated nucleotides weighted appropriately — but since these are short products, we approximate fraction of RNA molecules using total counts divided by rough average lengths), and full-length (600 nt) product is 85%, then per mole of RNA polymerase that has engaged the promoter, the ratio of molecules of abortive product to molecules of full-length product is much larger than the mass ratio suggests, because each abortive RNA is roughly (2 to 8, average ≈5) nucleotides versus 600 for full length:
So even though abortive RNA is only 15% of total labeled mass, there are roughly 21 abortive RNA molecules released for every one full-length transcript completed — a large number, correctly capturing the empirical fact (well documented for E. coli RNA polymerase at strong promoters such as T7A1 or lacUV5) that abortive initiation is a highly repetitive, wasteful, but ultimately self-limiting process that precedes productive elongation.
Limiting cases, scope & common traps
| Situation | What actually happens | Why |
|---|---|---|
| Template strand vs. coding strand confusion | RNA sequence matches the coding strand (with U for T), NOT the template strand directly | Template is read 3′→5′ and is complementary; coding strand is by definition the one with RNA-like sequence |
| "RNA polymerase reads both strands and picks the better match" | False — strand choice is fixed entirely by promoter orientation, not polymerase scanning | Promoter asymmetry, recognized by σ/TFs, fixes a single binding orientation (Derivation 2) |
| "RNA polymerase needs a primer just like DNA polymerase" | False — it initiates de novo using an initiation subsite that binds a free NTP's 3′-OH | Extra active-site subsite absent in DNA polymerase (Derivation 1) |
| Very short genes (< 35 bp) | Cannot support more than one polymerase at a time (footprint exceeds gene length) | Steric exclusion sets a hard ceiling on transcriptional throughput |
| High superhelical stress / negative supercoiling ahead of the bubble | Enhances promoter melting; positive supercoiling ahead of moving polymerase inhibits elongation ("twin supercoiled domain") | Torsional strain from strand separation must be resolved by topoisomerases |
| Extremely AT-rich vs. GC-rich templates | AT-rich promoters melt more easily (weaker base stacking/H-bonding) aiding open complex formation; GC-rich elongation regions can cause pausing | Base-pair stability directly affects strand separation energetics |
| Reverse transcriptase and RNA-dependent RNA polymerases | Use a strikingly similar two-metal-ion active site but read RNA templates or make DNA from RNA | Same catalytic chemistry is repurposed with different template/product specificity |
The chemist behind it
Roger D. Kornberg, the structural biochemist whose crystallographic work resolved the atomic architecture of eukaryotic RNA polymerase II in complete detail, earned the 2006 Nobel Prize in Chemistry for this achievement — a fitting continuation of a family tradition, since his father Arthur Kornberg had earlier won the 1959 Nobel Prize in Physiology or Medicine for discovering DNA polymerase. Roger Kornberg's laboratory, working through the 1990s and early 2000s, obtained the first high-resolution crystal structures of yeast RNA polymerase II — first as a ten-subunit core enzyme, then captured in the act of transcription with DNA template, nascent RNA, and incoming nucleotide all visible simultaneously in the active site. These structures directly visualized the two-metal-ion mechanism, the transcription bubble, the RNA-DNA hybrid, and the exit channels for RNA and displaced DNA strand — turning what had been inferred biochemically (from decades of work by researchers such as James Darnell, Robert Roeder, and Pierre Chambon on eukaryotic transcription, and earlier still by Jerard Hurwitz, Samuel Weiss, and Audrey Stevens, who independently discovered RNA polymerase activity in 1960) into a directly observed molecular movie. Kornberg's work exemplifies how structural biology transforms mechanistic hypotheses (built from kinetics, mutagenesis, and footprinting) into physically verified pictures — precisely the union of chemistry and structure this article has tried to convey for the polymerase active site.
Origins · 1 minRoger D. Kornberg and how this was discovered
Does RNA polymerase ever use the coding strand as a template by mistake?
No, not under normal conditions. Strand selection is locked in by the asymmetric promoter sequence and the single orientation in which σ factor (bacteria) or the pre-initiation complex (eukaryotes) can bind it, as shown in Derivation 2. Once that orientation is fixed, the DNA is opened into a bubble with a fixed geometry, and only one strand is physically threaded through the polymerase's active-site channel as template. There is no independent "strand-checking" step performed by the catalytic core itself, which is chemically strand-agnostic; the decision is made entirely upstream by promoter architecture and initiation-factor binding, not corrected or re-checked during elongation.
Why does RNA polymerase use ribonucleotides (with a 2′-OH) instead of deoxyribonucleotides, given the extra reactivity risk?
The presence of the 2′-OH is exactly what makes RNA transient and functionally versatile — it makes the phosphodiester backbone more susceptible to hydrolysis and to specific cleavage (both spontaneous, base-catalyzed transesterification via the 2′-OH attacking the adjacent phosphate, and enzymatic, via ribonucleases), which is desirable for a molecule meant to be turned over quickly rather than stored permanently like DNA. The 2′-OH also enables RNA to fold into complex tertiary structures (as in ribozymes, tRNA, and rRNA) that participate directly in catalysis, something DNA's 2′-deoxyribose backbone, lacking this hydroxyl, cannot readily do. Ribonucleotide use is therefore matched to RNA's biological role as an ephemeral, structurally versatile working copy, not an evolutionary oversight.
Is the "no primer needed" property unique to cellular RNA polymerases, or shared by viral polymerases too?
It is broadly shared. RNA-dependent RNA polymerases of RNA viruses (like poliovirus or SARS-CoV-2) and reverse transcriptases making the very first cDNA strand from an RNA template each face an analogous primer-provision problem, and different viral lineages have evolved distinct solutions: some RNA viral polymerases genuinely initiate de novo using a specialized initiation platform much like cellular RNA polymerase, while others (like poliovirus) use a small viral protein, VPg, covalently linked to a nucleotide, as a protein-primed initiation strategy. Reverse transcriptase, notably, does require a primer (a host tRNA, in the case of retroviruses like HIV) because its active site architecture, evolved from a DNA-polymerase-like fold, lacks the initiation subsite described in this article.
How is transcriptional fidelity actually measured, and why is it lower than DNA replication fidelity?
Fidelity is measured by quantifying misincorporation rates in reconstituted transcription reactions (comparing incorporation of correct vs. incorrect NTPs under matched conditions) or by deep sequencing of nascent RNA to detect mismatches relative to the genomic template, yielding error rates around 10⁻⁴ to 10⁻�³ per nucleotide for bacterial RNA polymerase (somewhat lower, ~10⁻⁵ to 10⁻⁶, for eukaryotic Pol II with proofreading assistance from TFIIS-stimulated backtracking and cleavage). This is several orders of magnitude worse than DNA replication fidelity (~10⁻⁹–10⁻¹⁰ after proofreading and mismatch repair) because an error in one of many transient RNA copies is diluted and eventually degraded, whereas a DNA replication error becomes a permanent, heritable mutation — so the evolutionary cost of imperfect transcription is far lower than the cost of imperfect replication.
Why do eukaryotes need three separate RNA polymerases instead of one, given that bacteria manage with a single enzyme?
The division into Pol I, II, and III likely reflects the need for highly specialized, differently regulated promoter recognition machinery for functionally distinct RNA classes with vastly different expression demands: Pol I dedicates itself entirely to making the enormous quantities of pre-ribosomal RNA needed to build ribosomes (in a specialized nucleolar compartment), Pol II handles the enormous complexity of protein-coding gene regulation (requiring elaborate enhancer/promoter integration via numerous transcription factors and Mediator), and Pol III specializes in short, highly transcribed structural RNAs like tRNA that need distinct internal promoter recognition strategies. Bacteria, lacking a nucleus and needing simpler, faster regulatory logic for a much smaller, more streamlined genome, manage with one core enzyme paired with swappable σ factors instead.
Does the transcription bubble stay a fixed size as RNA polymerase moves, or does it change?
It stays remarkably constant in size — roughly one turn of DNA, about 12–14 base pairs unwound at any instant, with about 8–9 of those base pairs forming a transient RNA-DNA hybrid inside the enzyme. As the polymerase translocates forward by one base pair, it simultaneously unwinds one new base pair at the leading (downstream) edge of the bubble and reanneals one base pair at the trailing (upstream) edge, so the bubble moves as a fixed-size "wave" of local strand separation traveling with the enzyme rather than growing indefinitely, which would otherwise create catastrophic topological strain far ahead of and behind the polymerase.
What happens to the coding strand while the template strand is being read?
The coding (non-template) strand is displaced into single-stranded form only within the transcription bubble itself; it is not degraded or chemically modified, and it re-anneals with the template strand immediately behind the trailing edge of the moving polymerase, restoring the normal DNA double helix once the enzyme has passed. This transient, localized single-strandedness is exactly analogous to (though independent of) the strand separation that occurs during DNA replication at the replication fork, and, like replication, it generates positive supercoiling ahead of and negative supercoiling behind the moving enzyme (the "twin supercoiled domain" model), which topoisomerases must relieve.