Transcription and RNA processing in eukaryotes

Eukaryotic transcription takes place inside a nucleus, on DNA wrapped in chromatin, and produces RNAs that must be extensively modified before they can leave the nucleus and be translated. Where a bacterium makes a short-lived mRNA that ribosomes start reading immediately, a eukaryotic cell makes a precursor RNA, caps its 5′ end, cuts out internal segments and joins the rest, adds a tail of adenines to its 3′ end, and exports the finished message to the cytoplasm. Each step is a point of quality control and of regulation. This lesson describes the three eukaryotic RNA polymerases, the assembly of the transcription machinery at promoters, and the processing steps that turn a primary transcript into a mature mRNA.

Three RNA polymerases

Eukaryotes divide transcription among three nuclear RNA polymerases, each related to the bacterial enzyme but larger, with 12 to 17 subunits. Plants have two more, Pol IV and Pol V, which make small RNAs involved in silencing. Mitochondria and chloroplasts have their own polymerases, reflecting their bacterial ancestry.

The toxin α-amanitin, made by the death cap mushroom (Amanita phalloides), binds RNA polymerase II tightly and stops it moving along DNA. Cells of the liver, which absorb the toxin first, stop making mRNA and die over several days, which is why death-cap poisoning is so dangerous. Death caps have spread in British Columbia, where they are associated with imported trees, and in California, and they have caused fatal poisonings in both places. Differences in sensitivity to α-amanitin were how researchers first distinguished the three polymerases.

Promoters and general transcription factors

Unlike bacterial RNA polymerase with its sigma factor, eukaryotic RNA polymerase II cannot recognize promoters on its own. It relies on a set of general transcription factors (GTFs), named TFIIA, TFIIB, TFIID, TFIIE, TFIIF and TFIIH, which assemble with the polymerase at the core promoter to form the preinitiation complex. The core promoter is the stretch of about 40 bp around the start site. Many, but far from all, Pol II promoters contain a TATA box, with consensus TATAAA, about 25 to 30 bp upstream of the start site. Others use different elements, such as the initiator (Inr) at the start site or a downstream promoter element.

Assembly begins when TFIID, which includes the TATA-binding protein (TBP), binds the promoter. TBP binds in the minor groove and bends the DNA sharply. TFIIB then binds, and positions RNA polymerase II, which arrives with TFIIF. TFIIE and TFIIH join last. TFIIH has two key enzymatic activities: a helicase that opens the DNA at the start site, and a kinase that phosphorylates the carboxyl-terminal domain (CTD) of the largest polymerase subunit. The CTD is a long tail of repeats of seven amino acids, Tyr-Ser-Pro-Thr-Ser-Pro-Ser, repeated 52 times in humans. Phosphorylation of serine 5 in these repeats helps the polymerase escape the promoter; phosphorylation of serine 2 accumulates during elongation. The changing pattern of CTD phosphorylation acts as a platform that recruits capping, splicing and polyadenylation enzymes at the right moments, coupling RNA processing to transcription.

TFIIH is also part of the nucleotide excision repair system (BIOL-201 lesson 10), which is why some inherited defects in its subunits cause both repair deficiency and developmental problems. In living cells, the general factors and polymerase alone give only a low basal level of transcription. Strong expression requires activator proteins bound to enhancers, and a large co-activator complex called Mediator that relays their signals to the polymerase. Lesson 10 covers these regulatory layers.

Promoter-proximal pausing

In many animal genes, RNA polymerase II starts transcription and then pauses 20 to 60 nucleotides downstream of the start site, held by pausing factors. Release from the pause requires the kinase P-TEFb, which phosphorylates the polymerase CTD and the pausing factors. Pausing allows genes to be primed for rapid activation, as with the heat-shock genes of fruit flies, where paused polymerases can be released within minutes of a temperature rise. Pausing is now recognized as a major point of regulation across the genome, not a curiosity of a few genes.

Capping

Almost as soon as the 5′ end of a Pol II transcript emerges from the polymerase, about 20 to 30 nucleotides into synthesis, it receives a cap: a guanine nucleotide joined backwards, through an unusual 5′-to-5′ triphosphate bridge, and then methylated at position 7 of the guanine (m⁷G). The cap protects the RNA from 5′ to 3′ exonucleases, helps export it from the nucleus, and is recognized by the translation initiation factor eIF4E, which recruits the ribosome in the cytoplasm (lesson 7). Messenger RNAs that lack a cap are rapidly degraded.

Splicing

The most striking feature of eukaryotic genes, discovered in 1977 by Phillip Sharp and Richard Roberts working with adenovirus, is that their coding sequences are interrupted by non-coding segments. The segments that remain in the mature RNA are exons; the intervening segments that are removed are introns. The removal of introns and the joining of exons is splicing. In human genes, exons average about 150 nucleotides while introns average several thousand, so most of a typical gene is intron. The gene for dystrophin, which is defective in Duchenne muscular dystrophy (BIOL-201), spans about 2.2 million bp, but its mature mRNA is only about 14,000 nucleotides; transcribing it takes around 16 hours.

Introns are recognized by short sequences at their ends and inside them. Nearly all introns begin with GU at the 5′ splice site and end with AG at the 3′ splice site, the GU–AG rule. Upstream of the 3′ splice site lies a polypyrimidine tract (a run of C and U), and upstream of that a branch point containing a conserved adenine. Splicing occurs in two chemical steps. First, the 2′-OH of the branch-point adenine attacks the phosphate at the 5′ splice site, cutting the RNA and forming a looped structure called a lariat. Second, the newly freed 3′-OH of the upstream exon attacks the 3′ splice site, joining the two exons and releasing the intron as a lariat, which is then degraded.

These reactions are carried out by the spliceosome, a huge, dynamic complex of five small nuclear RNAs (U1, U2, U4, U5 and U6) and well over 100 proteins. U1 snRNA pairs with the 5′ splice site; U2 pairs with the branch point. Structural studies by cryo-electron microscopy have shown that the catalytic centre of the spliceosome is formed by RNA, the U6 snRNA, coordinating metal ions, making the spliceosome a ribozyme, like the ribosome. Some introns in fungi, plants and organelles can even splice themselves without any protein, which supports the idea that the spliceosome evolved from such self-splicing introns.

def splice(pre, introns):
    """Remove introns given as (start, end) positions, 0-based, end exclusive."""
    out, pos = [], 0
    for s, e in sorted(introns):
        assert pre[s:s+2] == "GU" and pre[e-2:e] == "AG", "GU-AG rule violated"
        out.append(pre[pos:s])
        pos = e
    out.append(pre[pos:])
    return "".join(out)

pre = "AUGGCUAAGGUAAGUCUUACUAACCCUUUCAGGCAUUAUGA"
mature = splice(pre, [(9, 32)])
print("pre-mRNA length:", len(pre))
print("mature mRNA:    ", mature, f"({len(mature)} nt)")
print("codons:", " ".join(mature[i:i+3] for i in range(0, len(mature), 3)))

pre-mRNA length: 41
mature mRNA:     AUGGCUAAGGCAUUAUGA (18 nt)
codons: AUG GCU AAG GCA UUA UGA

Alternative splicing

A single gene can be spliced in more than one way. Exons can be skipped or included, alternative 5′ or 3′ splice sites can be used, introns can be retained, and mutually exclusive exons can be chosen. More than 90 per cent of human multi-exon genes produce more than one splice form, which helps explain how about 20,000 genes produce a much larger variety of proteins. Splicing choices are controlled by RNA-binding proteins that bind exonic and intronic enhancers and silencers. Many differ between tissues: neurons and muscle cells, for example, have distinctive splicing programs.

The number of possible products grows quickly. A gene with several independently chosen alternative exons can, in principle, produce many isoforms. The most famous example is the Dscam gene of the fruit fly, which has four clusters of mutually exclusive exons containing 12, 48, 33 and 2 alternatives. Choosing one exon from each cluster gives 12 × 48 × 33 × 2 = 38,016 possible proteins, more than twice the number of genes in the fly genome. Different neurons express different combinations, which helps them distinguish their own branches from those of other neurons.

from math import prod
from itertools import product

# Dscam: four clusters of mutually exclusive alternative exons
clusters = {"exon 4": 12, "exon 6": 48, "exon 9": 33, "exon 17": 2}
print("Dscam possible isoforms:", f"{prod(clusters.values()):,}")

# A smaller gene: 3 cassette exons, each independently included or skipped
cassettes = ["E2", "E4", "E6"]
forms = []
for choice in product([True, False], repeat=len(cassettes)):
    exons = ["E1"] + [c for c, keep in zip(cassettes, choice) if keep] + ["E7"]
    forms.append("-".join(exons))
print(f"\n{len(forms)} isoforms from {len(cassettes)} cassette exons:")
for f in forms:
    print("  ", f)

Dscam possible isoforms: 38,016

8 isoforms from 3 cassette exons:
   E1-E2-E4-E6-E7
   E1-E2-E4-E7
   E1-E2-E6-E7
   E1-E2-E7
   E1-E4-E6-E7
   E1-E4-E7
   E1-E6-E7
   E1-E7

Splicing goes wrong in disease. An estimated 10 to 15 per cent or more of disease-causing point mutations affect splicing, by destroying a splice site, creating a new one or disrupting an enhancer. Spinal muscular atrophy (SMA), once the most common genetic cause of infant death, results from loss of the SMN1 gene. A nearly identical backup gene, SMN2, differs by a single base that causes exon 7 to be skipped in most transcripts, giving an unstable protein. Two treatments work by correcting this splicing. Nusinersen is an antisense oligonucleotide, injected into the spinal fluid, that blocks a splicing silencer in SMN2 intron 7. Risdiplam is a small molecule taken by mouth that stabilizes U1 snRNP binding at the exon 7 splice site. Both raise the amount of full-length SMN protein. Newborn screening for SMA, now part of provincial and state programs, matters because treatment works best before symptoms appear.

Polyadenylation and termination

The 3′ end of a Pol II transcript is not set by where the polymerase stops. Instead, a signal sequence in the RNA, AAUAAA in most cases, located 10 to 30 nucleotides upstream of the cleavage site, is recognized by the cleavage and polyadenylation specificity factor (CPSF). Along with a GU-rich element downstream, it directs cleavage of the RNA. Poly(A) polymerase then adds a tail of about 200 adenines in mammals, without any template. The tail protects the mRNA, helps export and promotes translation; its gradual shortening in the cytoplasm marks the mRNA for decay. After cleavage, the polymerase continues for a while, but the uncapped RNA it is still making is degraded by an exonuclease that chases it down and helps knock the polymerase off the template, the so-called torpedo model of termination.

Export, surveillance and decay

Only properly processed mRNAs are exported through nuclear pore complexes. Splicing deposits a protein complex, the exon junction complex, about 20 to 24 nucleotides upstream of each exon–exon junction. During the first round of translation, ribosomes displace these complexes. If a ribosome reaches a stop codon more than about 50 nucleotides upstream of the last exon junction, a junction complex remains downstream, signalling that the stop codon is premature. The mRNA is then destroyed by nonsense-mediated decay. This surveillance protects cells from truncated proteins that might be harmful, and it explains why many nonsense mutations reduce mRNA levels rather than producing short proteins.

Eukaryotic mRNAs are generally more stable than bacterial ones. Half-lives in mammalian cells range from minutes for mRNAs encoding signalling proteins and transcription factors, which often carry AU-rich destabilizing elements in their 3′ untranslated regions, to more than a day for mRNAs of housekeeping proteins. Decay usually begins with shortening of the poly(A) tail, followed by removal of the cap and exonuclease digestion.

RNA editing

In some RNAs, the sequence is changed after transcription. In humans, the ADAR enzymes convert adenosine to inosine, which ribosomes read as guanosine; editing of a glutamate receptor mRNA in the brain changes one amino acid that controls calcium flow through the channel and is essential for normal function. Another enzyme, APOBEC1, converts a cytidine to uridine in apolipoprotein B mRNA in the intestine, creating a stop codon and producing a shorter protein than the liver makes from the same gene. Editing is a reminder that the sequence of a protein cannot always be read directly from the genome.