Promoters, Initiation, and Termination of Transcription

Sigma factor competition is exploited pharmacologically and biotechnologically. The antibiotic rifampin binds within the RNA exit channel of the β subunit and blocks RNA chain extension past a few nucleotides, a target chosen precisely because bacterial RNA polymerase's initiation machinery differs enough from the eukaryotic Pol II apparatus (with its unrelated general transcription factors) to allow selective inhibition; rifampin remains a first-line tuberculosis drug. Synthetic biologists routinely engineer intrinsic terminators as modular, ATP-free "off switches" in synthetic gene circuits and CRISPR guide RNA cassettes, exploiting exactly the hairpin/U-tract thermodynamics derived above to tune termination efficiency and thereby control relative expression levels of downstream genes in an operon-like construct. Orthogonal sigma factor engineering — pairing mutant sigma factors with correspondingly mutated -35/-10 elements — is used to build insulated synthetic promoters that respond only to an engineered RNA polymerase variant, minimizing crosstalk with host transcription in synthetic biology chassis organisms. Rho-dependent termination is exploited by phage and by bacteria themselves as a regulatory mechanism (transcriptional attenuation and polarity), and Rho itself is an actively pursued antibacterial target, since bicyclomycin, a natural product, specifically inhibits Rho's ATPase activity and is under continued study as a scaffold for novel antibiotics. Finally, the same closed-complex/open-complex/escape framework, generalized enormously, underlies how eukaryotic Pol II promoters are read by TFIID, TFIIB, and the rest of the general transcription factor set, a far more elaborate elaboration of the same basic logic taken up in Eukaryotic RNA Processing: Capping, Splicing, Polyadenylation and the combinatorial control schemes of Regulation of Gene Expression: Chemical Logic.

Ask the tutor: How does an activator protein like CAP/CRP actually increase the rate constant $k_a$ at a promoter it regulates — does it change $K_B$, $k_f$, or both?

Bacterial RNA polymerase core enzyme is a five-subunit assembly, α₂ββ'ω, shaped like a crab claw with a positively charged internal channel that grips DNA. On its own, core enzyme binds DNA with roughly equal affinity anywhere along the genome, and any transcription it initiates begins at random sites with no fixed start nucleotide. The problem is not catalytic — the two-metal-ion mechanism at the ββ' active site works identically whether the template is a real gene or synthetic junk DNA — the problem is entirely one of address-finding. Sigma factor solves the address-finding problem. Sigma70 (so named for its 70 kDa mass) is built from four structurally distinct regions, 1.1 through 4.2, connected by flexible linkers. In free solution sigma70 is autoinhibited: region 1.1 occupies the DNA-binding channel and region 4.2 is folded against region 2, sequestering the surfaces that must eventually read promoter DNA. When sigma70 binds core enzyme, it undergoes an extended conformational change: region 4.2 swings out to sit at the tip of the β-flap, positioned to scan roughly one helical turn of DNA 35 base pairs upstream of where transcription will start; region 2.4, in the main cleft, is positioned over the -10 element 10 base pairs upstream. This "molecular ruler" geometry — two reading heads separated by a fixed distance corresponding to the holoenzyme footprint — is what allows one holoenzyme to scan for a bipartite promoter sequence without needing to fully unwind DNA first. Recognition proceeds in two thermodynamically distinct phases. First, holoenzyme forms a closed complex (RPc): sigma reads the -35 hexamer TTGACA and the -10 hexamer TATAAT (consensus sequences derived from aligning hundreds of E. coli promoters) while the DNA duplex remains fully base-paired. This is a bimolecular binding event governed by an equilibrium association constant $K_B$. Second, the closed complex isomerizes into an open complex (RPo): the DNA strands separate over roughly 13 base pairs, from position -11 to +2 relative to the +1 transcription start site, forming the transcription bubble. Isomerization is governed by a forward rate constant $k_f$ and, once formed, RPo is extremely stable — the melted region is locked into the enzyme's active-site cleft and stabilized by sigma region 2.3, which flips out and stacks against a conserved adenine at the -11 position of the non-template strand.

Once RPo forms, transcription initiation begins with NTPs entering the active site and phosphodiester bond formation proceeding by the universal two-metal-ion mechanism (Mg²⁺-coordinated 3'-OH attacking the α-phosphate of the incoming NTP, releasing pyrophosphate). But the enzyme does not simply run off down the gene. Because sigma region 4.2 remains bound to the -35 element while the catalytic center tries to translocate downstream, the DNA between -35 and the active site must be pulled in and compressed — a process called scrunching. This creates growing torsional and mechanical strain. Short RNAs of 2 to ~9 nucleotides are made and released in a futile cycle called abortive initiation; only when accumulated strain is sufficient to break the sigma–(-35) interaction does the enzyme escape into processive elongation, at which point sigma dissociates (though not always immediately) and core polymerase transcribes for kilobases without needing promoter-specific contacts at all. At the far end of a gene, the elongating core enzyme must be stopped. Two chemically distinct devices exist. Intrinsic (Rho-independent) termination uses information built directly into the nascent RNA: a GC-rich inverted repeat that folds into a hairpin 7–9 base pairs long, immediately followed by a run of ~8 uridines being transcribed. The hairpin forms inside the RNA exit channel, mechanically wrenching the transcription bubble and destabilizing the weak rU:dA hybrid (the weakest of all Watson–Crick pairs) that holds RNA to template. Rho-dependent termination instead uses an accessory hexameric ATPase, Rho, which loads onto an upstream C-rich, unstructured "rut" (Rho utilization) site on the nascent RNA, translocates 5'→3' along the RNA using ATP hydrolysis, and catches up to a polymerase that has paused, unwinding the RNA:DNA hybrid via a mechanism related to that of a replicative helicase.

The overall promoter recognition equilibrium is usually decomposed as a two-step process:

where:

• $R$ is free holoenzyme, $P$ is promoter DNA
• $RP_c$ is the closed complex (duplex intact, sigma bound to -35/-10)
• $RP_o$ is the open complex (melted bubble, ready to initiate)
• $K_B = k_1/k_{-1}$ is the closed-complex association constant
• $k_f$ is the (essentially irreversible on physiological timescales) isomerization rate constant

The overall second-order rate constant for open-complex formation, $k_a$, measured experimentally by abortive-initiation assays, combines both steps:

Reading: because $k_f \gg k_{-1}$ for strong promoters, nearly every closed complex that forms proceeds forward to RPo rather than dissociating back to free DNA and enzyme; the reaction is "funneled" downhill. This is why promoter strength correlates strongly with the -35/-10 match to consensus (which sets $K_B$) and with -10 AT content and spacer length (which sets $k_f$). The free energy relationship for the two-step process:

where:

• $\Delta G^{\circ}_{B} = -RT\ln K_B$ is the binding free energy of the closed complex
• $\Delta G^{\ddagger}_{f}$ is the activation barrier for isomerization (melting + conformational change)
• $\Delta G^{\ddagger}_{-1}$ is the activation barrier for closed-complex dissociation

Reading: promoter strength is not simply "how tightly does polymerase bind DNA" — it is a kinetic competition between forward isomerization and backward dissociation from the closed complex, which is why some strong promoters have only moderate closed-complex affinity but very fast isomerization. For abortive initiation, the productive escape probability per initiation attempt, $P_{esc}$, competes against abortive cycling:

where:

• $k_{esc}$ is the rate constant for breaking sigma-promoter contacts and moving into elongation
• $k_{abort}$ is the rate constant for RNA release and cycle restart without escape

Reading: because $k_{abort}$ is typically comparable to or larger than $k_{esc}$ at many promoters, a single open complex often produces several abortive transcripts before one successful escape event — this is directly observable as a "ladder" of short RNAs in vitro. For intrinsic termination, the thermodynamic driving force can be expressed as a competition between hairpin folding free energy and hybrid stability:

where:

• $\Delta G_{\text{hairpin}}$ is the (negative, favorable) free energy of RNA hairpin folding
• $\Delta G_{\text{hybrid}}$ is the (also negative but weak) free energy of the rU:dA RNA:DNA hybrid in the transcription bubble

Reading: termination is favored when hairpin folding releases enough free energy to outcompete the weak rU:dA hybrid holding the 3' end of the RNA in the active site; because rU:dA is the weakest possible hybrid, only a modest hairpin is needed, which is why U-tract length (not just hairpin strength) is a strong predictor of termination efficiency.

Example 1 (foundational) — Consensus matching and relative promoter strength. A promoter has a -35 element of TTGCCA (two mismatches from consensus TTGACA, positions 4 and 5) and a -10 element of the perfect consensus TATAAT, with a 17 bp spacer (the optimal length). A second promoter has a perfect -35 (TTGACA) but a -10 of TATGAT (one mismatch, position 4) and a 16 bp spacer. Rank which is likely stronger, using the qualitative rule that -10 match and spacer length dominate $k_f$ (isomerization), while -35 match dominates $K_B$ (closed-complex affinity), and that both terms multiply in $k_a = K_B k_f$. Promoter 1 has a perfect -10 (favorable isomerization, since it is the -10 that must melt) but a weaker -35 (weaker closed-complex capture, hence smaller $K_B$). Promoter 2 has a perfect -35 (strong capture) but a mismatched -10 (slower melting) and a suboptimal spacer (16 bp, one short of the 17 bp optimum, which weakly stresses the geometry between the two sigma-reading heads). Because a mismatch in the -10 element directly impairs the base-flipping/melting step that is rate-limiting for many promoters (recall from the derivation that -10 AT content sets the melting free energy almost entirely), promoter 1's perfect -10 combined with only a two-base -35 mismatch is empirically the more common architecture of strong natural promoters (e.g., rrn promoters tolerate substantial -35 divergence but almost never diverge from -10 consensus at the critical -11 and -7 positions). The qualitative prediction: promoter 1 is likely stronger overall, primarily through a larger $k_f$, even though its $K_B$ is somewhat reduced — consistent with the general finding that $k_a$ correlates more tightly with -10 match than with -35 match across characterized E. coli promoters.

Example 2 (applied) — Abortive initiation cycling and escape probability from single-molecule data. A single-molecule fluorescence experiment on a strong promoter finds that, per open complex, the enzyme releases an average of 4.5 abortive transcripts (mean cycle number $\bar{n}$) before productive escape. Assuming each cycle has the same constant escape probability $p$ per attempt (a geometric distribution), estimate $p$ and the corresponding ratio $k_{abort}/k_{esc}$. For a geometric distribution with success probability $p$, the mean number of trials until first success is $1/p$; the mean number of failures before success is

Setting $\bar{n}_{\text{failures}} = 4.5$:

So the escape probability per cycle is about 18%. Since $P_{esc} = k_{esc}/(k_{esc}+k_{abort}) = p$,

Sensibly, this recovers $k_{abort}/k_{esc} \approx \bar n$ itself (as expected for a geometric process where mean failures before success equals the odds ratio against success), confirming internal consistency: abortive release is roughly 4.5-fold faster than promoter escape at this promoter, meaning strain must build across several scrunching/abortive cycles before the sigma–(-35) contact finally breaks.

Example 3 (exam/competition level) — Designing an intrinsic terminator and predicting termination efficiency from hybrid competition. A synthetic biologist designs an intrinsic terminator with a hairpin stem of 9 bp, all G:C pairs, with folding free energy $\Delta G_{\text{hairpin}} = -22$ kcal/mol (using standard nearest-neighbor RNA folding parameters for a GC-rich stem-loop with a 4-nt loop), followed immediately by a U-tract of 7 uridines. The RNA:DNA hybrid within the transcription bubble at the point of termination is dominated by the last 8 base pairs of the hybrid, of which 7 are rU:dA (each contributing roughly $-0.5$ kcal/mol of hybrid stability, among the weakest known Watson–Crick-type pairs in an RNA:DNA duplex) and 1 is a stronger rG:dC pair contributing about $-2.0$ kcal/mol.

(a) Estimate total hybrid stability $\Delta G_{\text{hybrid}}$ and compare with $\Delta G_{\text{hairpin}}$ to predict whether termination should be efficient.

Using $\Delta G_{\text{term}} = \Delta G_{\text{hairpin}} - \Delta G_{\text{hybrid}}$ as defined above (careful with sign convention: we want the net driving force for termination, i.e., how much more favorable it is for the RNA to fold into the hairpin and dissociate than to remain hybridized):

A strongly negative value indicates that hairpin folding vastly outcompetes retention of the weak rU-rich hybrid — this terminator is predicted to be highly efficient (in practice, well-characterized strong intrinsic terminators like the E. coli rrnB T1 terminator show termination efficiencies exceeding 90–95% in vitro, consistent with this large a driving force).

(b) Now suppose a mutation shortens the U-tract to 3 uridines (replacing 4 of the 7 U's with G/C, each contributing an additional $-2.0$ kcal/mol of hybrid stability instead of $-0.5$ kcal/mol). Recompute $\Delta G_{\text{hybrid}}$ and $\Delta G_{\text{term}}$, and predict the consequence.

Termination is still thermodynamically favorable, but the driving force has dropped by 6 kcal/mol (from $-16.5$ to $-10.5$ kcal/mol) — corresponding, via $\Delta\Delta G = -RT\ln(\text{efficiency ratio})$, to roughly a $e^{6/0.616}\approx e^{9.7}\approx 1.6\times10^4$-fold shift in the effective equilibrium between "terminate" and "read through" outcomes at $T=310$ K, predicting a substantial drop in termination efficiency — this is exactly the experimentally established rule that shortening the U-tract of natural terminators (from the typical 7–9 U's down to 3–4) causes measurable read-through, even when the hairpin itself is unchanged, because the weak rU:dA hybrid is mechanistically essential, not merely incidental, to intrinsic termination.

Much of the mechanistic dissection of sigma factor function and open-complex kinetics is associated with Michael Chamberlin, who in the 1970s at Berkeley pioneered quantitative kinetic assays for RNA polymerase–promoter interactions, establishing the closed-complex/open-complex two-step framework and the heparin-challenge assay still used to measure $k_a$ today. The Rho protein and Rho-dependent termination were characterized biochemically starting with Jeffrey Roberts, whose 1969 discovery of a factor required for correct termination on phage λ DNA templates first demonstrated that termination, like initiation, is an actively catalyzed and factor-dependent process rather than a passive falling-off. The structural basis for how sigma factor reprograms core enzyme specificity — the conformational extension and the placement of regions 2 and 4 at the two promoter elements — emerged from crystallographic work in the 2000s by groups including those of Seth Darst and Richard Ebright, whose structures of holoenzyme and open complexes gave atomic detail to decades of biochemical inference.