Transcription in bacteria

A gene is useful only when it is read. Transcription, the copying of a DNA sequence into RNA, is the first and most heavily regulated step in gene expression. In bacteria it is also a remarkably simple process in outline: one kind of RNA polymerase, guided by a detachable subunit that recognizes start signals, copies genes into RNA that can be translated even before transcription is finished. This lesson follows a bacterial transcript from start to finish, introduces the DNA signals that mark the beginning and end of genes, and models how strongly promoters bind and how quickly messages decay. Lesson 6 contrasts this with the more elaborate eukaryotic process, and lesson 9 shows how bacteria switch genes on and off.

The RNA polymerase

Bacterial RNA polymerase is a large enzyme built from five kinds of subunit. The core enzyme, made of two α subunits, β, β′ and a small ω subunit, can synthesize RNA but cannot find the beginning of a gene by itself. Adding a sixth subunit, the sigma factor (σ), forms the holoenzyme, which recognizes promoters. Like DNA polymerase, RNA polymerase synthesizes 5′ to 3′, adding ribonucleoside triphosphates (ATP, GTP, CTP and UTP) to the 3′ end of the growing chain. Unlike DNA polymerase, it needs no primer: it can start a chain from scratch, and it uses only one strand of the DNA as a template.

Two strand names cause endless confusion, so it is worth fixing them now. The template strand is the strand that RNA polymerase reads; the RNA is complementary to it. The other strand, the coding strand (also called the sense or non-template strand), has the same sequence as the RNA, except that it contains T where the RNA has U. Gene sequences in databases are written as the coding strand, 5′ to 3′, so that they can be read directly as codons.

COMP = str.maketrans("ACGT", "TGCA")

def transcribe(coding_5to3):
    """RNA has the coding-strand sequence with U in place of T."""
    return coding_5to3.replace("T", "U")

def template_from_coding(coding_5to3):
    """Template strand written 3'->5' under the coding strand."""
    return coding_5to3.translate(COMP)

coding = "ATGCCTAGGTAA"
print("coding   5'", coding, "3'")
print("template 3'", template_from_coding(coding), "5'")
print("mRNA     5'", transcribe(coding), "3'")

coding   5' ATGCCTAGGTAA 3'
template 3' TACGGATCCATT 5'
mRNA     5' AUGCCUAGGUAA 3'

Promoters

A promoter is the DNA sequence where RNA polymerase binds and begins transcription. In E. coli the most common sigma factor, σ⁷⁰, recognizes two short sequences, located about 10 and 35 base pairs upstream of the transcription start site (called position +1). By convention, positions upstream are negative and downstream positive; there is no position 0. The −10 element, also called the Pribnow box, has the consensus sequence TATAAT. The −35 element has the consensus TTGACA. They are separated by a spacer of 16 to 18 bp, ideally 17.

A consensus sequence is the most common base at each position across many examples, and real promoters rarely match it perfectly. In general, the closer a promoter matches the consensus, the more tightly the holoenzyme binds and the more often transcription starts. Strong promoters, such as those for ribosomal RNA genes, start transcription every second or two; weak ones may start only once or twice per cell generation. Matching can be scored simply by counting agreements, or better by weighting each position by how strongly it is conserved, using a position weight matrix.

CONS35, CONS10 = "TTGACA", "TATAAT"

def score(seq, cons):
    return sum(a == b for a, b in zip(seq, cons))

def best_promoter(dna, spacers=(16, 17, 18)):
    best = None
    for i in range(len(dna) - 6):
        for sp in spacers:
            j = i + 6 + sp
            if j + 6 > len(dna):
                continue
            s35, s10 = score(dna[i:i+6], CONS35), score(dna[j:j+6], CONS10)
            total = s35 + s10 - abs(sp - 17)       # small penalty for non-ideal spacing
            if best is None or total > best[0]:
                best = (total, i, dna[i:i+6], s35, sp, dna[j:j+6], s10)
    return best

promoters = {
    "near-consensus": "GCATTGACAATTAATCATCGGCTCGTATAATGTGTGGAATTG",
    "lac-like (weak)": "GCTTTACACTTTATGCTTCCGGCTCGTATGTTGTGTGGAATTG",
    "poor match":     "GCAGGCCGAGCAAGTCGCATCCGCCTGCGCCAGCGGTAATTG",
}
for name, dna in promoters.items():
    total, i, m35, s35, sp, m10, s10 = best_promoter(dna)
    print(f"{name:16s} -35 {m35} ({s35}/6)  spacer {sp}  -10 {m10} ({s10}/6)  score {total}")

near-consensus   -35 TTGACA (6/6)  spacer 16  -10 TATAAT (6/6)  score 11
lac-like (weak)  -35 TTTACA (5/6)  spacer 18  -10 TATGTT (4/6)  score 8
poor match       -35 AAGTCG (2/6)  spacer 17  -10 GGTAAT (4/6)  score 6

The lac promoter in the second example is deliberately weak: its −35 and −10 elements match the consensus poorly, and it depends on an activator protein (CAP, lesson 9) to recruit RNA polymerase. This is a common design. A weak promoter that needs an activator gives the cell a way to turn a gene on strongly only when a signal is present.

Bacteria have several sigma factors, each recognizing a different class of promoters. In E. coli, σ⁷⁰ handles most genes during normal growth, σ³² recognizes promoters of heat-shock genes, σ⁵⁴ controls genes for nitrogen metabolism, and σ³⁸ takes over many genes as cells enter stationary phase. Switching sigma factors is a quick way to reprogram large sets of genes at once. Bacillus subtilis uses a cascade of sigma factors to drive the stages of spore formation.

Initiation, elongation and termination

Transcription proceeds in three stages. In initiation, the holoenzyme binds the promoter as a closed complex, then melts about 13 base pairs around the start site to form an open complex in which the template strand is exposed. It begins making RNA but often produces short abortive transcripts of 2 to 10 nucleotides, which are released, before it escapes from the promoter. Once the RNA is longer than about 10 nucleotides, the sigma factor is usually released and the core enzyme moves on.

During elongation, the polymerase moves along the template at about 40 to 80 nucleotides per second in E. coli, maintaining a transcription bubble of about 13 to 14 bp of unwound DNA in which about 8 to 9 bp of RNA are paired with the template as an RNA–DNA hybrid. Behind the bubble the DNA rewinds and the RNA exits. Like replication, transcription overwinds the DNA ahead and underwinds it behind, so topoisomerases are needed here too. RNA polymerase lacks a true proofreading exonuclease, but it can backtrack and, with help from GreA and GreB proteins, cut off a few nucleotides including a mistake. Its error rate is about 1 in 10⁴ to 10⁵, far higher than replication, which is acceptable because a mistake affects only one RNA copy among many, not the inherited genome.

Termination in bacteria happens in two main ways. In intrinsic (Rho-independent) termination, the newly made RNA contains a GC-rich inverted repeat followed by a run of about eight U residues. The inverted repeat folds into a stable hairpin as soon as it emerges from the polymerase, and the weak A–U pairs in the RNA–DNA hybrid behind it allow the transcript to slip off. In Rho-dependent termination, a ring-shaped helicase called Rho binds to a C-rich, unstructured stretch of RNA, moves along it using ATP, catches up with a paused polymerase and pulls the RNA away. About half of E. coli genes end at intrinsic terminators.

Coupled transcription and translation

Bacteria have no nucleus, so ribosomes can bind an mRNA as soon as its 5′ end emerges from RNA polymerase. Transcription and translation are therefore coupled: the leading ribosome often follows closely behind the polymerase, which helps prevent Rho from terminating transcription prematurely inside genes. Electron micrographs from the 1970s by Oscar Miller and colleagues show a DNA strand with RNA polymerases strung along it, each trailing an mRNA studded with ribosomes, like a Christmas tree. Lesson 9 shows how coupling allows a regulatory mechanism called attenuation.

Many bacterial mRNAs are polycistronic: a single transcript carries several protein-coding sequences, each with its own ribosome-binding site. Genes transcribed together from one promoter form an operon, and they usually encode proteins that work in the same pathway, such as the three lac genes (lesson 9). Operons ensure that the proteins of a pathway are made together and in fixed proportion.

mRNA lifetimes

Bacterial mRNAs are short-lived, with half-lives typically of 2 to 5 minutes at 37 °C, compared with hours for many eukaryotic mRNAs. Rapid decay means that when transcription of a gene stops, its protein soon stops being made, letting bacteria respond quickly to change. If mRNA decays with a constant half-life, the amount remaining after time t is given by exponential decay:

At steady state, when synthesis and decay balance, the number of mRNA molecules equals the synthesis rate divided by the decay constant. A gene transcribed once per minute, with a half-life of 3 minutes, has about 1 / (0.693 / 3) ≈ 4.3 mRNA copies per cell on average. This simple balance explains why cells can tune expression either by changing how often transcription starts or by changing how fast mRNA is degraded.

import math

def steady_state(rate_per_min, half_life_min):
    k = math.log(2) / half_life_min
    return rate_per_min / k

print("mRNA copies at steady state (synthesis rate / decay constant)")
for rate in (0.2, 1, 5):
    row = [f"{steady_state(rate, h):6.1f}" for h in (1, 3, 10)]
    print(f"  {rate:3} transcripts/min: half-life 1, 3, 10 min ->", "  ".join(row))

# Decay after transcription stops, half-life 3 min, starting at 13 copies
k = math.log(2) / 3
for t in range(0, 16, 3):
    print(f"t = {t:2d} min: {13 * math.exp(-k * t):5.1f} copies left")

mRNA copies at steady state (synthesis rate / decay constant)
  0.2 transcripts/min: half-life 1, 3, 10 min ->    0.3     0.9     2.9
    1 transcripts/min: half-life 1, 3, 10 min ->    1.4     4.3    14.4
    5 transcripts/min: half-life 1, 3, 10 min ->    7.2    21.6    72.1
t =  0 min:  13.0 copies left
t =  3 min:   6.5 copies left
t =  6 min:   3.2 copies left
t =  9 min:   1.6 copies left
t = 12 min:   0.8 copies left
t = 15 min:   0.4 copies left

Antibiotics that target transcription

Rifampicin (rifampin in the United States) binds the β subunit of bacterial RNA polymerase and blocks the path of the growing RNA after two or three nucleotides, so initiation fails. It does not affect human RNA polymerases, which differ in structure. Rifampicin is a mainstay of tuberculosis treatment, always in combination with other drugs because resistance, caused by mutations in the rpoB gene encoding the β subunit, arises readily. Rapid molecular tests that detect rpoB mutations directly from a patient sample are now used to identify drug-resistant tuberculosis within hours, an example of molecular biology moving from the bench to the clinic. Tuberculosis remains a serious problem in parts of northern Canada, where incidence among Inuit communities has been many times the national rate, and federal and Inuit organizations have committed to eliminating it.