Bacterial Promoters and the -10/-35 Elements

A promoter is a DNA sequence, typically 40-60 bp long, located immediately upstream of a gene's transcription start site, that directs RNA polymerase to bind and initiate transcription at the correct position and on the correct strand. In bacteria, promoter recognition is not performed by the catalytic core enzyme of RNA polymerase (subunit composition α₂ββ'ω) alone. Core enzyme binds DNA nonspecifically and cannot distinguish a promoter from any other sequence. Specific recognition requires an additional subunit, sigma factor (σ), which associates with core enzyme to form the holoenzyme (α₂ββ'ωσ). Sigma factor is the actual DNA-sequence reader; it is released after initiation, leaving core enzyme to elongate the transcript. The dominant sigma factor under normal growth conditions in E. coli is σ⁷⁰ (so named because it has a molecular mass of about 70 kDa). σ⁷⁰-holoenzyme recognizes promoters through two hexameric sequence elements positioned by their distance upstream of the transcription start site, which is conventionally numbered +1. Nucleotides upstream of +1 are given negative numbers (there is no position "0"). The two elements are:

These are consensus sequences — the statistically most common base at each position, derived by aligning many known strong promoters. No single natural promoter matches the consensus at every position; most real promoters are imperfect matches, and this imperfection is precisely what allows genes to be expressed at different, tunable strengths rather than all firing at maximal, identical rates. The spacer between the two elements is not read for sequence but its length is critical, most efficient at 17 bp (range 16-19 bp), because it must position the two elements on the correct face and register of the DNA helix for simultaneous contact with the two sigma regions.

Step 1 — Holoenzyme assembly. Core RNA polymerase associates with a free σ⁷⁰ subunit. Sigma factor by itself has very low affinity for DNA in the relevant conformation; it is holoenzyme, with sigma properly positioned against the core enzyme's β and β' subunits, that acquires high-affinity, sequence-specific promoter binding activity.

Step 2 — Closed complex formation. Holoenzyme diffuses along DNA (partly by facilitated one-dimensional sliding) and samples sequences. When σ⁷⁰ region 4 (a helix-turn-helix domain) encounters a -35 sequence resembling TTGACA, and region 2 downstream encounters a -10 sequence resembling TATAAT at the correct spacing, the two contacts cooperate to form a stable closed complex (RPc), in which the DNA duplex is still fully base-paired.

Step 3 — Isomerization to the open complex. The A-T base pairs of the -10 element require only two hydrogen bonds each (versus three for G-C pairs) and are inherently easier to separate. Sigma region 2 does more than just read the -10 sequence in duplex form: it flips out and directly reads two bases (typically the A at position -11 and the T at position -7) in their single-stranded state, capturing them in aromatic pockets of the sigma protein. This nucleates local strand separation of roughly 13 bp, extending from about -11 to +2 or +3, producing the open complex (RPo), in which the template strand is exposed in the active site. This melting step is why promoter -10 regions are so consistently A-T rich: G-C pairs would resist melting and slow initiation.

Step 4 — de novo RNA synthesis and abortive initiation. With the template strand accessible, RNA polymerase begins synthesizing RNA from the +1 nucleotide using NTP substrates, without a primer. Early rounds frequently produce short abortive transcripts (2-9 nucleotides) that are released while polymerase remains promoter-bound, a repeated futile cycle before successful escape.

Step 5 — Promoter escape. Once the nascent RNA-DNA hybrid reaches roughly 9-10 nt, the enzyme breaks its contacts with the -35/-10 elements, sigma is typically released (though not obligatorily, and can remain associated further into elongation than once thought), and core polymerase moves off as a processive elongation complex. This transition — promoter clearance — is covered in depth in the article on Transcription Initiation and Promoter Escape. Two further, secondary sequence elements refine this picture and explain especially strong promoters:

• The extended -10 element: a TG dinucleotide immediately upstream of the -10 hexamer (around position -14/-15), contacted by σ⁷⁰ region 3.0, which can partially compensate for a weak or absent -35 element.
• The UP element: an A-T rich sequence further upstream (around -40 to -60), recognized not by sigma but by the two α-subunit C-terminal domains (αCTD) of core polymerase, which dock in the DNA minor groove and boost transcription up to 30-fold. UP elements are prominent in the extremely strong ribosomal RNA (rrn) promoters, which must be transcribed at enormous rates to supply ribosome biogenesis.

The overall logic — two spaced recognition boxes read by different protein domains, cooperating to bend and then locally melt DNA — is a mechanistic theme that reappears, with different sequences and different sigma factors, throughout bacterial gene regulation; it is elaborated further in Sigma Factors and Promoter Recognition.

The image above captures the static anatomy; the process is inherently dynamic, involving DNA bending of roughly 90° at the -10/-35 region and a wave of local melting. The next figure traces that dynamic sequence of events as a pathway.

The existence and consensus sequences of the -35 and -10 elements were established not by a single experiment but by a convergence of three lines of evidence: direct sequencing of many natural promoters, footprinting/protection assays that mapped exactly where polymerase touches DNA, and systematic mutagenesis that measured how deviations from consensus change transcription rate.

Sequence alignment of natural promoters. In the 1970s, as more bacterial and phage promoter sequences became available (from E. coli genes and from phages such as T7 and lambda), David Pribnow (1975) noted a short A-T rich region roughly 10 bp upstream of known transcription start sites shared strong sequence similarity across many unrelated promoters — this became known as the Pribnow box. Shortly after, comparison of a larger promoter collection by Mark Ptashne, Carl Pribnow's contemporaries, and especially the compilation by Harley and Reznikoff and later Hawley and McClure (1983, analyzing ~150-300 promoters), extended this to a second conserved region around -35, yielding the two-element consensus TTGACA...(16-19 bp)...TATAAT. The key evidential point: this consensus was not assumed, it emerged from aligning real sequenced promoters by their experimentally mapped +1 start sites (start sites had been mapped independently using primer extension and S1 nuclease mapping of mRNA 5' ends).

DNase I footprinting. To show that RNA polymerase physically contacts these exact regions, David Galas and Albert Schmitz introduced DNase I footprinting in 1978. The logic: end-label a DNA fragment containing a promoter, bind RNA polymerase, then briefly digest with DNase I, which cuts DNA roughly randomly except where a bound protein sterically blocks access. Run the partially digested DNA on a sequencing gel alongside a Maxam-Gilbert sequencing ladder of the same fragment. Wherever the protein is bound, a gap ("footprint") appears in the ladder of cleavage products because those phosphodiester bonds are protected from cutting. When this was applied to E. coli RNA polymerase bound to well-characterized promoters (e.g., the lac promoter, the lambda P_R and P_L promoters), the footprint extended from roughly -50 to +20, precisely spanning both the -35 and -10 consensus regions and the intervening spacer, with additional finer protection patterns (hypersensitive sites at specific phosphates due to DNA bending) exactly where the -10 element sits. This directly proved, at nucleotide resolution, that RNA polymerase makes contact across both elements simultaneously — confirming they are not just statistically associated with promoters but are the physical footprint of the bound enzyme.

Mutagenesis and quantitative promoter strength. The causal, quantitative link between sequence and function was established through systematic promoter mutagenesis, most influentially by William McClure and colleagues and by Michael Chamberlin's and Ptashne's laboratories using promoters such as lacUV5, the lambda phage P_R/P_RM promoters, and later synthetic promoter libraries. Single base-pair substitutions were introduced at each position of the -35 and -10 boxes, and the effect was measured directly by: 1. In vitro abortive initiation assays — measuring the rate of synthesis of short abortive RNA products, a proxy for the rate of open complex formation and initiation. 2. KMnO₄ footprinting — potassium permanganate specifically oxidizes unpaired thymines, so it reports directly on whether the -10 region has melted into an open complex; mutations that block melting eliminate the KMnO₄ reactivity at expected positions. 3. In vivo reporter fusions — placing mutant promoters upstream of a lacZ or later fluorescent reporter and measuring enzyme activity/fluorescence as a proxy for transcription rate in living cells. The consistent result: mutations that move a base toward the consensus (e.g., changing a non-consensus base at -35 position 3 to the consensus G) increase promoter strength, typically several-fold per position change, while mutations away from consensus decrease it. Changing the spacer length away from 17 bp (either shorter or longer, even while preserving both hexamer sequences perfectly) sharply reduces transcription, confirming that correct helical phasing — not just sequence identity — is required for simultaneous engagement by the two sigma domains. This body of work allowed construction of quantitative "promoter strength" models (weight matrices) that predict transcription rate from sequence with substantial accuracy, a foundational result for later synthetic biology promoter engineering. Together, these three evidence streams — comparative sequencing, footprinting, and mutagenesis — moved the -10/-35 model from a descriptive pattern to a mechanistically and causally validated one.

Example 1 — Scoring and ranking three promoter variants. Suppose three candidate promoter sequences are proposed for a synthetic operon, and you are asked to predict their relative strength using consensus matching. The consensus is -35: TTGACA, spacer 17 bp, -10: TATAAT.

Reasoning: P1 is a perfect consensus match at both boxes and ideal spacing — it should be the strongest, essentially a maximal-affinity σ⁷⁰ promoter (in fact, a perfect-consensus synthetic promoter of this kind is unusually strong compared to almost all natural E. coli promoters, most of which carry several mismatches — strong natural constitutive promoters typically deviate from perfect consensus at 2-4 positions, which keeps expression tunable and prevents pathological over-transcription). P2 has a single mismatch in the -10 box, at a position (position 5, normally A) known from mutagenesis studies to be moderately important for stable strand separation; expect a measurable but modest reduction in strength (roughly 2- to 5-fold weaker judged from typical single-position effect sizes in the literature) relative to P1. P3 combines a -35 mismatch with non-ideal spacing (19 bp rather than 17 bp) — even though its -10 box is perfect. Because both effects are independently deleterious, and spacing errors are especially punishing (they disrupt simultaneous two-domain engagement rather than a single contact), P3 should be predicted as the weakest of the three, likely an order of magnitude or more below P1, despite having only "one" formal -35 mismatch when spacing is not counted. Ranking of predicted strength: P1 > P2 > P3. This mirrors the logic used to design synthetic promoter libraries (e.g., the Anderson promoter collection used in synthetic biology), where systematic variation of -35/-10 sequence and spacing produces a graded series of constitutive promoter strengths spanning roughly 100-fold or more.

Example 2 — Diagnosing a transcription defect from footprinting data. A researcher isolates a mutant E. coli strain in which a normally strongly-expressed metabolic gene is transcribed at only ~5% of wild-type levels. Two candidate promoter mutations, X and Y, are found nearby; only one is causal. To determine which, the researcher performs DNase I footprinting and KMnO₄ footprinting on both variants with purified σ⁷⁰-holoenzyme in vitro.

• Mutant X: DNase I footprint is essentially identical to wild-type across -50 to +20 (holoenzyme still binds the full promoter region and bends DNA normally). KMnO₄ footprint, however, shows no reactivity at the expected melted thymines around -11/-7, meaning the DNA never opens.
• Mutant Y: DNase I footprint shows loss of protection specifically over the -35 region (roughly -40 to -30), while protection over the -10 region and downstream is retained; KMnO₄ reactivity is present but reduced, appearing after a longer incubation.

Interpretation: Mutant X's defect is at the level of isomerization — the polymerase can still find and bind the promoter (closed complex forms fine, per the normal DNase footprint) but something prevents the closed-to-open transition, consistent with a mutation weakening the A-T richness of the -10 element or disrupting the specific bases sigma region 2 must flip out and read. Mutant Y's defect is at the level of initial recognition — loss of DNase protection specifically over -35 indicates the mutation has damaged the -35 hexamer itself (probably a base substitution reducing similarity to TTGACA), weakening but not eliminating σ region 4 binding; because -35 contacts help nucleate the closed complex and stabilize the overall holoenzyme-DNA interaction, some open complex can still form (delayed KMnO₄ reactivity) but at a much lower rate, consistent with reduced overall transcription. This kind of reasoning — using two complementary footprinting methods that separately report on protein binding (DNase I) versus DNA strand opening (KMnO₄) — is the standard way to localize a promoter mutation's mechanistic effect, distinguishing a "recognition" defect from an "isomerization" defect.

The recognition of a conserved A-T rich region upstream of bacterial transcription start sites is credited to David Pribnow, working in the mid-1970s as sequence data from early bacteriophage and E. coli promoters began to accumulate following the development of rapid DNA sequencing methods. Pribnow compared the sequences just upstream of several known start sites — from phage T7 promoters and from E. coli genes such as the lactose operon — and noticed a recurring TATAAT-like motif positioned about 10 bp before the start site, a discovery that gave the element its informal name, the Pribnow box. The complementary -35 element and the two-element consensus model were subsequently developed through the broader comparative and biochemical work of investigators including Heinz Schaller, Michael Rosenberg, and later systematized quantitatively by David Hawley and William McClure, whose 1983 compilation and statistical analysis of over a hundred promoter sequences produced the widely cited consensus and spacing rules still taught today. The structural and mechanistic understanding of how sigma factor reads these elements — the helix-turn-helix recognition of -35 by region 4, and the base-flipping recognition of melted -10 bases by region 2 — came later still, from crystallography and biochemistry on sigma factor domains by groups including those of Seth Darst and Richard Ebright in the 1990s-2000s, revealing the atomic basis for a pattern first noticed simply by eye in a stack of sequenced base pairs.

Antibiotic development. Because bacterial transcription initiation depends on a bacteria-specific sigma-holoenzyme mechanism absent in the eukaryotic and archaeal RNA polymerase systems used by human cells, the initiation machinery is an attractive, selective antibacterial target; compounds that block sigma-core interaction or open complex formation are actively studied as antibiotic leads distinct from rifamycins (which target the catalytic β subunit rather than promoter recognition itself).

Synthetic biology and metabolic engineering. Promoter libraries built by systematically varying -35/-10 sequence and spacer length (such as the widely used Anderson promoter collection and related constitutive promoter panels) give bioengineers a toolkit of parts spanning roughly two orders of magnitude in expression strength, used to fine-tune metabolic pathway flux, balance enzyme stoichiometry in engineered production strains (for biofuels, pharmaceuticals, biosensors), and build synthetic genetic circuits with predictable behavior.

Diagnostics and strain engineering. Because promoter strength is quantitatively predictable from sequence, computational promoter-strength models trained on -35/-10/spacer/UP-element features are used to design expression cassettes for recombinant protein production (e.g., industrial enzyme manufacture, vaccine antigen production in bacterial hosts) without extensive trial-and-error screening.

Understanding stress and virulence gene regulation. Alternative sigma factors that recognize different consensus sequences than σ⁷⁰ (introduced in Sigma Factors and Promoter Recognition) allow bacteria to redirect the same core polymerase to entirely different gene sets during heat shock, nitrogen starvation, sporulation, or infection of a host — a regulatory strategy exploited by pathogens to switch on virulence genes, making sigma-factor-controlled promoters a target of interest in anti-virulence drug discovery.

Basic teaching and quantitative biology. The -10/-35 system remains the canonical, simplest example in all of molecular biology of how a protein "reads" DNA sequence to make a binary decision (bind strongly vs. weakly), making it a standard entry point for teaching position weight matrices, consensus sequences, and sequence-function relationships that recur throughout genomics.

Ask the tutor: How does a bound activator protein help RNA polymerase at a promoter with a weak -35 element?