Transcription Elongation and Proofreading

Transcription elongation is the phase after promoter escape in which RNA polymerase (RNAP) synthesizes RNA processively — meaning it stays attached to the same DNA template and does not dissociate between successive nucleotide additions, unlike many DNA polymerases that must reassociate for each processing event via a sliding clamp. To do this, RNAP maintains a transcription bubble: a locally melted, single-stranded region of DNA roughly 12–14 base pairs wide that travels with the enzyme along the gene. Within the bubble, the nascent RNA transiently base-pairs with the template strand to form an RNA:DNA hybrid of about 8–9 base pairs, housed inside a groove of the enzyme called the main channel. As the enzyme advances, DNA is unwound ahead of the bubble and rewound behind it, while the RNA is peeled away from the template and extruded through an RNA exit channel, having earned no permanent double-stranded region with the DNA. Fidelity during elongation refers to the accuracy with which the correct ribonucleotide — Watson-Crick complementary to the template base being read — is selected and incorporated, and, when an error slips through, corrected. Two distinct fidelity mechanisms operate. The first is kinetic selection: the active site discriminates correct from incorrect NTPs based on the fit of the base pair in the insertion site, favoring rapid, high-affinity binding of the correct nucleotide and a conformational closure of the enzyme's "trigger loop" that is much slower or aborted for a mismatch. The second is intrinsic proofreading: after an incorrect nucleotide is incorporated, the enzyme senses the resulting distortion, backtracks (slides backward along the DNA without dissociating), and removes the terminal, mismatched nucleotide(s) either by reversing the polymerization reaction (pyrophosphorolysis) or by endonucleolytic cleavage of the extruded RNA 3′ end, often stimulated by accessory transcript-cleavage factors. This combination — kinetic proofreading at the point of insertion plus post-insertion excision — gives RNAP an overall transcriptional error rate on the order of 10⁻⁴ to 10⁻⁶ errors per nucleotide added, several orders of magnitude worse than DNA replication's fidelity (~10⁻⁹–10⁻¹⁰ with proofreading and mismatch repair) but adequate because RNA is not the permanent, heritable information store; errors are diluted out as transcripts turn over and are not passed to daughter cells.

The elongation complex: architecture of a moving machine. The core RNAP enzyme (in bacteria, subunit composition α₂ββ′ω; in eukaryotes, RNA Polymerase II with twelve subunits including Rpb1 and Rpb2) forms a crab-claw shaped structure with the DNA held in a positively charged main channel between two large lobes. A structural element called the bridge helix spans the cleft near the active site and, together with the trigger loop, forms the catalytic core around a two-metal-ion (Mg²⁺) mechanism analogous to that used by DNA polymerases and many other nucleotidyl transferases. The template DNA strand threads through the active site, the RNA:DNA hybrid occupies a cleft just downstream of the catalytic Mg²⁺ ions, and the nascent RNA is peeled off the template by a rudder or lid structural motif and directed into a separate RNA exit channel. A clamp domain closes over the DNA like a jaw, explaining why the polymerase, once initiated, essentially never falls off until it hits a terminator — it is topologically almost encircling the nucleic acid.

The nucleotide addition cycle. Elongation proceeds through a repeating four-step chemical and mechanical cycle for each nucleotide added: 1. NTP binding. An incoming nucleoside triphosphate diffuses into the secondary channel (in bacteria) or enters via a funnel-shaped pore and binds in the insertion site, base-pairing transiently with the templating DNA base. 2. Trigger loop closure and catalysis. If the base pair is correct (Watson-Crick geometry), the trigger loop folds closed over the NTP, correctly positioning it and the two catalytic Mg²⁺ ions for in-line nucleophilic attack: the 3′-OH of the growing RNA chain attacks the α-phosphate of the incoming NTP, forming a new phosphodiester bond and releasing pyrophosphate (PPi). 3. Pyrophosphate release. PPi is released (often hydrolyzed further to two inorganic phosphates, pulling the reaction forward thermodynamically) and the trigger loop reopens. 4. Translocation. The enzyme moves exactly one base pair forward along the DNA (Brownian-ratchet or power-stroke models are both debated, but the net effect is a single-nucleotide step), vacating the insertion site for the next NTP and repositioning the RNA 3′ end. This cycle repeats at typical E. coli elongation rates of 40–80 nucleotides per second in vivo (somewhat slower — around 10–20 nt/s — for RNA Pol II under some conditions, though rates vary with gene and species), pausing transiently at many template positions.

Kinetic (insertion-level) fidelity. The trigger loop acts as a conformational fidelity checkpoint: closure requires the correct Watson-Crick geometry of the nascent base pair; a mismatch destabilizes closure, dramatically slowing the catalytic step and giving the enzyme time to release the incorrect NTP before bond formation — an induced-fit discrimination step conceptually similar to that in DNA polymerases, though generally somewhat less stringent.

Intrinsic proofreading — the correction step. When a wrong nucleotide is incorporated anyway (roughly once per several thousand to ten-thousand additions before correction), the mismatched 3′ terminus distorts the active site and stalls forward translocation. The polymerase then often backtracks: it slides backward along the DNA-RNA complex by one or more base pairs without dissociating, extruding the 3′ end of the RNA (now containing the error) out through the secondary channel behind the active site, forming a "backtracked" RNA loop. Two mechanisms then remove the erroneous nucleotide(s):

• Pyrophosphorolysis: the reverse of the polymerization reaction — PPi re-enters the active site and attacks the terminal phosphodiester bond, releasing the terminal NTP and regenerating a 3′-OH.
• Endonucleolytic (hydrolytic) cleavage: intrinsic to the polymerase's own catalytic site but greatly stimulated by accessory factors — in bacteria, GreA and GreB; in eukaryotes, TFIIS — which insert an acidic hairpin loop into the secondary channel, supplying catalytic residues that activate a water molecule to hydrolyze the RNA backbone a few nucleotides upstream of the 3′ end, cleaving out the mismatched region and creating a new, correctly paired 3′-OH for resumed synthesis.

Both mechanisms restore a correctly matched 3′ end, after which elongation resumes from the corrected position. Backtracking and pause-associated cleavage are not purely accidental — they also occur at natural elongation pause sites (sequence-encoded hairpins or GC-rich hybrids) and are exploited regulatorily, for instance to allow ribosome loading in bacteria or splicing factor recruitment co-transcriptionally in eukaryotes, foreshadowing how pausing feeds into intrinsic termination.

Backtracking and cleavage in detail. When RNAP backtracks by, say, 2–3 nucleotides, the enzyme's main channel now contains RNA that has been displaced from the active site; that 3′ single-stranded RNA tail threads into the narrow secondary channel — the same channel through which NTPs normally diffuse in. GreA and GreB (or TFIIS in eukaryotes) work by inserting an acidic loop tip through this channel to reach the backtracked RNA, supplying (or properly positioning) a second catalytic metal ion or activating water for hydrolysis. GreA typically stimulates cleavage of short (2–3 nt) fragments useful for rescuing transient backtracks and misincorporations; GreB can act on much deeper backtracks (up to ~9 nt), rescuing arrested complexes that would otherwise be permanently stalled. Because these factors accelerate a reaction the enzyme can also do unassisted (albeit slowly), transcript cleavage is described as intrinsic but factor-stimulated proofreading.

Chamberlin's kinetic studies established processivity and rate. In the 1960s–70s, Michael Chamberlin and colleagues purified E. coli RNA polymerase and measured elongation directly with radiolabeled NTPs on defined DNA templates. They showed that once initiated, a single polymerase molecule synthesizes RNA continuously at a roughly constant rate without dissociating and rebinding — the operational definition of a processive enzyme — and estimated elongation rates in the range of tens of nucleotides per second, consistent with values later confirmed in vivo. These pulse-chase and rifampicin-block experiments (rifampicin blocks new initiation but not ongoing elongation) let the field distinguish elongating complexes from newly initiating ones and quantify elongation rate independent of promoter effects.

Miller spreads visualized transcription bubbles and multiple polymerases in action. Oscar Miller's chromatin/gene-spreading electron microscopy technique (the famous "Miller trees") directly visualized ribosomal RNA genes in E. coli being transcribed by multiple RNA polymerase molecules simultaneously, each with a growing RNA "feather" of increasing length trailing from the gene — providing direct visual proof that transcription is processive, polar (5′-to-3′), and that many polymerases can traverse a single gene in tandem at defined average spacing, consistent with kinetically measured elongation rates.

Structural biology revealed the bubble and hybrid directly. X-ray crystal structures of bacterial RNAP (Seth Darst's lab, Thermus aquaticus enzyme) and yeast RNA Polymerase II elongation complexes (Roger Kornberg's lab, awarded the 2006 Nobel Prize in Chemistry for this work) captured the enzyme bound to nucleic acid scaffolds mimicking the transcription bubble, showing directly: an ~8–9 bp RNA:DNA hybrid inside the main cleft, a downstream DNA duplex entering through a separate channel, a rudder/lid separating RNA from DNA, and a bridge helix and trigger loop positioned exactly where mechanistic models predicted. These structures converted decades of biochemical inference into an atomic-resolution picture and confirmed the secondary channel as the route for both NTP entry and backtracked-RNA extrusion — later corroborated by structures with GreB and TFIIS bound in that channel.

Single-molecule experiments proved the stepping, pausing, and backtracking mechanism. Steven Block's laboratory used optical tweezers to tether single RNAP molecules transcribing single DNA templates under calibrated tension, recording real-time position versus time traces of individual enzymes. These experiments directly observed: discrete single-base-pair steps consistent with the nucleotide addition cycle; frequent pauses of variable duration; and — crucially — backtracking events in which the polymerase's position moved backward along the DNA by several base pairs before resuming forward motion, exactly as predicted by the biochemical proofreading model. Because force could be applied to assist or resist backward movement, these experiments also showed that assisting forward force reduced pause duration, while resisting force lengthened pauses and backtracks — direct mechanical evidence that backtracking is a real, force-dependent physical state of the enzyme, not merely a kinetic inference from cleavage assays.

Biochemical reconstitution proved GreA/GreB and TFIIS function via cleavage. In vitro transcription assays with stalled or backtracked elongation complexes showed that adding purified GreB (or TFIIS) to arrested complexes containing a misincorporated 3′ nucleotide restored RNA synthesis, and that the mechanism involved endonucleolytic cleavage of the RNA — demonstrated by mapping the exact cleavage products by gel electrophoresis and sequencing, showing removal of precisely the 3′-terminal mismatched residues, after which the shortened, corrected transcript could be extended again by the same polymerase. Mutant polymerases lacking key catalytic residues in the active site (needed for the intrinsic, factor-independent cleavage activity) showed elevated error rates in reporter-gene fidelity assays, directly linking the cleavage mechanism to transcriptional accuracy in vivo.

Example 1 — quantifying transcriptional error load in a bacterium. Suppose E. coli RNA polymerase makes errors during elongation at an intrinsic (post-proofreading) rate of 1 error per 10⁵ nucleotides synthesized. A given operon's mRNA is 3,000 nucleotides long, and under fast growth this gene is transcribed at a rate of 20 initiation events per minute, sustained for 10 minutes. Number of transcripts made = 20 × 10 = 200 mRNA molecules. Expected errors per transcript = 3,000 × (1/10⁵) = 0.03 errors/transcript. Expected total erroneous transcripts ≈ 200 × 0.03 = 6 transcripts (assuming at most one error matters per transcript, a reasonable first approximation since 0.03 ≪ 1, so the Poisson probability of ≥1 error per transcript ≈ 0.03 also, giving ≈ 6 flawed transcripts out of 200). Contrast this with a single round of chromosomal replication of the same 3,000 bp region at DNA fidelity (~10⁻¹⁰ per bp after proofreading and mismatch repair): expected errors = 3,000 × 10⁻¹⁰ ≈ 3 × 10⁻⁷, i.e., a vanishingly small chance of even one replication error in that same stretch. This numeric contrast — errors roughly a million-fold more common in transcription than replication — illustrates precisely why RNA polymerase can tolerate a "sloppier" active site: transcriptional errors are diluted across many independent RNA copies and are never inherited, so a single bad transcript is a minor, non-heritable cost, whereas a single bad replication event is permanent.

Example 2 — reading a single-molecule trace to identify backtracking. In an optical-trap experiment, an RNAP–DNA tether shows the enzyme's position advancing steadily at ~35 bp per 5 seconds (a slope of 7 bp/s) for the first 20 seconds, then the trace suddenly reverses, moving backward by 8 base pairs over 3 seconds, pauses flat for 12 seconds, and then resumes forward motion at the same 7 bp/s rate. Interpreting this trace: the steady forward slope is normal elongation (consistent with in vivo-like rates). The backward excursion of 8 bp is a backtracking event — the enzyme has slid backward without dissociating, extruding roughly 8 nucleotides of its own nascent RNA through the secondary channel. The subsequent flat, paused region (12 seconds with zero net movement) corresponds to the time the enzyme spends either waiting for spontaneous cleavage/pyrophosphorolysis or (in a reconstituted system) waiting for a cleavage factor like GreB to act and restore a competent 3′-OH end. The resumption of forward motion at the pre-pause rate confirms that after correction, the elongation complex is fully catalytically restored — it has not been damaged or permanently arrested, and no new initiation event was needed. If the experiment were repeated with GreB added to the reaction chamber, the paused/backtracked region would shorten dramatically (cleavage becomes much faster), which is exactly the kind of dose-dependent rescue seen in reconstituted single-molecule assays testing Gre factors.

Understanding elongation and its intrinsic proofreading has direct biomedical and biotechnological payoffs. The antibiotic rifampicin targets the bacterial RNAP β subunit and blocks the transition from initiation to productive elongation, making it a frontline tuberculosis drug; resistance mutations map precisely to the RNAP structural regions defined by Darst-lab crystallography. The mushroom toxin α-amanitin inhibits eukaryotic RNA Polymerase II specifically by binding near the bridge helix and trigger loop, freezing translocation — a mechanistic insight from elongation structural biology now exploited in antibody-drug conjugates (amanitin-linked antibodies) for targeted cancer therapy. Backtracking and transcription-coupled repair are linked: persistently stalled, backtracked Pol II at DNA lesions is recognized by repair factors (transcription-coupled nucleotide excision repair), and mutations in this coupling underlie human diseases such as Cockayne syndrome. In synthetic biology and biotechnology, engineered RNA polymerase variants with altered trigger-loop dynamics are used to tune transcription speed and fidelity in cell-free expression systems and to design orthogonal transcription systems (e.g., T7 RNA polymerase variants) for industrial mRNA and protein production, including large-scale mRNA vaccine manufacturing, where elongation processivity and fidelity directly determine yield and product quality.

Ask the tutor: How does intrinsic transcription termination in bacteria exploit the same pausing and hybrid-destabilization machinery used during elongation and proofreading?